Improve resource discovery and evidence metadata

README.md

@@ -178,14 +178,64 @@ Start with the first-read resources and the Loop Contract if the term is new. Fo
 
 ## Reading Paths
 
-Choose a path based on your intent.
+Use this atlas before browsing the exhaustive topical sections. It adds a task-oriented layer without changing the stable section anchors used by existing links.
 
-| Goal                             | Reading path                                                                                              |
-| -------------------------------- | --------------------------------------------------------------------------------------------------------- |
-| **Learn the concept**            | Canonical definition, mental model, comparison guide, and the Loop Contract                               |
-| **Implement a loop**             | Core primitives, official runtime guides, the pattern library, and examples                               |
-| **Improve reliability or evals** | Verification gates, securing unattended loops, benchmarks, critiques, and limitations                     |
-| **Contribute**                   | Community gallery, templates, and contribution guide                                                      |
+<table>
+  <thead>
+    <tr>
+      <th>Goal</th>
+      <th>Start here</th>
+      <th>Continue with</th>
+      <th>Best for</th>
+    </tr>
+  </thead>
+  <tbody>
+    <tr>
+      <td><strong>Learn</strong></td>
+      <td><a href="#canonical-definition">Canonical definition</a></td>
+      <td><a href="#start-here">Start Here</a> → <a href="#research-foundations">Research Foundations</a></td>
+      <td>Newcomers and researchers</td>
+    </tr>
+    <tr>
+      <td><strong>Design</strong></td>
+      <td><a href="#the-loop-contract">Loop Contract</a></td>
+      <td><a href="#pattern-library">Pattern Library</a> → <a href="#core-loop-primitives">Core Loop Primitives</a> → <a href="#agent-workflow-patterns">Agent Workflow Patterns</a></td>
+      <td>Architects and loop designers</td>
+    </tr>
+    <tr>
+      <td><strong>Build</strong></td>
+      <td><a href="meta/RUNTIME_SELECTION.md">Runtime selection</a></td>
+      <td><a href="#official-runtime-guides">Official Runtime Guides</a> → <a href="#coding-agent-loop-systems">Coding-Agent Loop Systems</a> → <a href="#orchestration-and-multi-agent-delegation">Orchestration</a></td>
+      <td>Agent and platform builders</td>
+    </tr>
+    <tr>
+      <td><strong>Persist</strong></td>
+      <td><a href="#state-memory-and-context-persistence">State and context</a></td>
+      <td><a href="#core-loop-primitives">Core Loop Primitives</a> → <a href="#operations-playbooks">Operations Playbooks</a></td>
+      <td>Long-running and resumable loops</td>
+    </tr>
+    <tr>
+      <td><strong>Verify</strong></td>
+      <td><a href="#verification-and-feedback-gates">Verification Gates</a></td>
+      <td><a href="#benchmarks-and-evaluation">Benchmarks And Evaluation</a> → <a href="#critiques-risks-and-limitations">Critiques And Limitations</a></td>
+      <td>Eval and reliability engineers</td>
+    </tr>
+    <tr>
+      <td><strong>Govern</strong></td>
+      <td><a href="#securing-unattended-loops">Securing Unattended Loops</a></td>
+      <td><a href="#operations-playbooks">Operations Playbooks</a> → <a href="#critiques-risks-and-limitations">Critiques And Limitations</a></td>
+      <td>Operators, approvers, and security</td>
+    </tr>
+    <tr>
+      <td><strong>Apply</strong></td>
+      <td><a href="#templates-and-patterns">Templates And Patterns</a></td>
+      <td><a href="#examples-and-schema">Examples And Schema</a> → <a href="#community-gallery">Community Gallery</a> → <a href="#contributing">Contributing</a></td>
+      <td>Teams adopting or sharing loops</td>
+    </tr>
+  </tbody>
+</table>
+
+The Hugging Face dataset exposes the same collection, lifecycle-stage, audience, evidence, source-status, and repository-stat facets for filtering all resources.
 
 ## Choose Your Loop
 
@@ -266,10 +316,10 @@ Beyond the curated list, this repository also maintains:
 
 Direct resources about the new AI/coding-agent meaning of Loop Engineering.
 
-- 📝 **Blog** [Loop Engineering](https://addyosmani.com/blog/loop-engineering/) - Addy Osmani's framing of loop engineering as the layer above manually prompting coding agents, with concrete primitives across Codex and Claude Code; also on [Substack](https://addyo.substack.com/p/loop-engineering) with the original discussion trail and Steinberger and Cherny quotations.
+- 📝 **Blog** [Loop Engineering by Addy Osmani](https://addyosmani.com/blog/loop-engineering/) - Addy Osmani's framing of loop engineering as the layer above manually prompting coding agents, with concrete primitives across Codex and Claude Code; also on [Substack](https://addyo.substack.com/p/loop-engineering) with the original discussion trail and Steinberger and Cherny quotations.
 - 📝 **Blog** [Peter Steinberger on designing loops](https://x.com/steipete/status/2063697162748260627) - The June 2026 post - "you shouldn't be prompting coding agents anymore, you should be designing loops that prompt your agents" - that catalyzed the current discussion.
 - 📝 **Blog** [Boris Cherny: five tips for running Opus autonomously for hours or days](https://x.com/bcherny/status/2063792263067754658) - The Claude Code creator's compact loop recipe: auto-mode permissions, dynamic workflows, `/goal` or `/loop`, the cloud runner, and end-to-end self-verification.
-- 📝 **Blog** [Loop Engineering](https://cobusgreyling.substack.com/p/loop-engineering) - Concise explanation of the shift from prompting agents to designing loops that discover work, delegate, verify, persist, and continue.
+- 📝 **Blog** [Loop Engineering by Cobus Greyling](https://cobusgreyling.substack.com/p/loop-engineering) - Concise explanation of the shift from prompting agents to designing loops that discover work, delegate, verify, persist, and continue.
 - 📝 **Blog** [Stop Prompting. Design the Loop.](https://www.pulumi.com/blog/stop-prompting-design-the-loop/) - Practical breakdown of loop building blocks - automations, worktrees, skills, connectors, subagents - plus external memory and verification through oracles such as tests and builds.
 - 📝 **Blog** [Writing Loops, Not Prompts, Explained](https://rico.codes/loops-not-prompts) - Rico Kahler's break-even model for when a recurring task justifies building a loop instead of prompting, with stop conditions, evidence collection, and an execution-horizon framing for moving from execution-bound to judgment-bound work.
 - 📝 **Blog** [Loop Engineering: A Guide for Engineers and Practitioners](https://medium.com/@adnanmasood/loop-engineering-a-guide-for-engineers-and-practitioners-893bb65ea943) - Adnan Masood's practitioner guide that organizes loop design into triggers, topologies, verifiers, and termination rules, with coverage of failure modes, cost control, and observability for production agent loops.
@@ -374,15 +424,15 @@ Practical loop patterns translate the abstract contract into runnable operating
 
 Feature-level building blocks you assemble a loop from: schedulers, goals, worktrees, hooks, skills, plugins, and protocols.
 
-- 📚 **Docs** [Automations - Codex app](https://developers.openai.com/codex/app/automations) - Codex background automations for recurring tasks, triage inboxes, skills, and worktree isolation.
-- 📚 **Docs** [Follow a goal - Codex use cases](https://developers.openai.com/codex/use-cases/follow-goals) - Official guidance for durable objectives with stopping conditions, validation commands, checkpoints, and progress logs.
-- 📚 **Docs** [Worktrees - Codex app](https://developers.openai.com/codex/app/worktrees) - Codex worktree model for isolated parallel tasks and handoffs between local and background workspaces.
-- 📚 **Docs** [Prompting - Codex](https://developers.openai.com/codex/prompting) - Explains the Codex loop, threads, context, and `/goal` mode.
-- 📚 **Docs** [Customization - Codex](https://developers.openai.com/codex/concepts/customization) - Maps `AGENTS.md`, memories, skills, MCP, and subagents into a coherent customization stack.
-- 📚 **Docs** [Agent Skills - Codex](https://developers.openai.com/codex/skills) - Official skill format for reusable workflows, scripts, MCP dependencies, invocation policy, and plugin packaging.
-- 📚 **Docs** [Plugins - Codex](https://developers.openai.com/codex/plugins) - Bundles skills, app integrations, and MCP servers into reusable loop capabilities.
+- 📚 **Docs** [Scheduled tasks - ChatGPT Learn](https://learn.chatgpt.com/docs/automations?surface=app) - Official guidance for recurring background tasks, triage inboxes, skills, and isolated workspaces.
+- 📚 **Docs** [Follow a goal - ChatGPT Learn](https://learn.chatgpt.com/use-cases/follow-goals) - Official guidance for durable objectives with stopping conditions, validation commands, checkpoints, and progress logs.
+- 📚 **Docs** [Git worktrees - ChatGPT Learn](https://learn.chatgpt.com/docs/environments/git-worktrees) - Official worktree model for isolated parallel tasks and handoffs between local and background workspaces.
+- 📚 **Docs** [Prompting - ChatGPT Learn](https://learn.chatgpt.com/docs/prompting) - Explains the Codex loop, threads, context, and goal-oriented prompting.
+- 📚 **Docs** [Customization overview - ChatGPT Learn](https://learn.chatgpt.com/docs/customization/overview) - Maps `AGENTS.md`, memories, skills, MCP, and subagents into a coherent customization stack.
+- 📚 **Docs** [Build skills - ChatGPT Learn](https://learn.chatgpt.com/docs/build-skills) - Official skill format for reusable workflows, scripts, MCP dependencies, invocation policy, and plugin packaging.
+- 📚 **Docs** [Plugins - ChatGPT Learn](https://learn.chatgpt.com/docs/plugins) - Bundles skills, app integrations, and MCP servers into reusable loop capabilities.
 - 🧰 **Tool** [dotskills](https://github.com/vincentkoc/dotskills) - A `.skills` registry of curated Codex and OpenClaw skills, framed as an "ADE Loop" (Agent Development Environment to registry to Skills Gym) where reusable skills are developed, shared, and evaluated across runs.
-- 📚 **Docs** [Slash commands in Codex CLI](https://developers.openai.com/codex/cli/slash-commands) - CLI commands for switching agent threads, browsing skills, inspecting MCP tools, and using subagent workflows.
+- 📚 **Docs** [Developer commands - ChatGPT Learn](https://learn.chatgpt.com/docs/developer-commands?surface=cli) - CLI commands for switching agent threads, browsing skills, inspecting MCP tools, and using subagent workflows.
 - 🔁 **Pattern** [Autonomous Loops](https://claudecodeguide.dev/docs/patterns/autonomous-loops) - Claude Code pattern using task files, stop hooks, restart behavior, hard limits, and a kill switch.
 - 📚 **Docs** [Claude Code Glossary](https://code.claude.com/docs/en/glossary.md) - Defines the agentic loop, hooks, subagents, skills, MCP, and related primitives in Claude Code terminology.
 - 📚 **Docs** [Keep Claude working toward a goal](https://code.claude.com/docs/en/goal) - `/goal` runs turn after turn until a completion condition is met by a verifier.
@@ -413,7 +463,7 @@ End-to-end operating guides and release notes from the runtime vendors themselve
 
 <!-- OpenAI and Codex -->
 - 📚 **Docs** [Run long horizon tasks with Codex](https://developers.openai.com/blog/run-long-horizon-tasks-with-codex) - OpenAI's runbook for plan-edit-test-observe-repair-document-repeat work, including specs, plans, status logs, and validation gates.
-- 📚 **Docs** [Best practices - Codex](https://developers.openai.com/codex/learn/best-practices) - Official best practices for context, `AGENTS.md`, MCP, skills, subagents, and automations.
+- 📚 **Docs** [Best practices - ChatGPT Learn](https://learn.chatgpt.com/guides/best-practices) - Official best practices for context, `AGENTS.md`, MCP, skills, subagents, and automations.
 - 📚 **Docs** [Agents SDK](https://developers.openai.com/api/docs/guides/agents) - OpenAI guide for agent orchestration, tool execution, approvals, state, guardrails, and observability.
 - 📚 **Docs** [Agents - OpenAI Agents SDK](https://openai.github.io/openai-agents-python/agents/) - SDK primitives for agents, tools, handoffs, guardrails, and runner-managed loops.
 - 📚 **Docs** [Running agents](https://developers.openai.com/api/docs/guides/agents/running-agents) - OpenAI guide to turns, state, approvals, sessions, and continuation in the SDK runtime loop.
@@ -644,7 +694,7 @@ These resources include harness and observability mechanisms that loops compose
 - 📄 **Paper** [A Trace-Based Assurance Framework for Agentic AI Orchestration: Contracts, Testing, and Governance](https://arxiv.org/abs/2603.18096) - Treats execution traces as the assurance substrate, pairing machine-checkable contracts, testing, and governance so recurring agent orchestration stays verifiable and auditable.
 - 📄 **Paper** [Self-Evolving Agents with Anytime-Valid Certificates](https://arxiv.org/abs/2607.00871) - Confines self-modification to a small steering adapter around a frozen base model and gates each change with anytime-valid statistical tests that emit auditable certificates, reporting solve-count gains and logged regression prevention on a SWE-bench Verified subset.
 - 📄 **Paper** [Delayed Verification Destabilizes Multi-Agent LLM Belief](https://arxiv.org/abs/2606.27409) - Models verifier-corrector loops in multi-agent LLM systems as delayed consensus, deriving a stability threshold where verification that is too strong or too late turns factual consensus into oscillation, plus a greedy corrector-placement algorithm validated on five open models.
-- 📄 **Paper** [Lean4Agent: Formal Modeling and Verification for Agent Workflow and Trajectory](https://arxiv.org/abs/2606.06523) - Models agent workflows and trajectories in Lean 4 dependent types so semantic consistency is machine-checked rather than judged by an LLM, with verification-passing workflows outperforming failing ones by about 12% on software-engineering benchmarks.
+- 📄 **Paper** [Lean4Agent: Formal Modeling and Verification for Agent Workflow and Trajectory](https://arxiv.org/abs/2606.06523) - Models agent workflows and trajectories in Lean 4 dependent types so semantic consistency is machine-checked rather than judged by an LLM, with verification-passing workflows outperforming failing ones by an average of 11.94% on software-engineering benchmarks.
 - 📄 **Paper** [Regimes: An Auditable, Held-Out-Gated Improvement Loop](https://arxiv.org/abs/2606.10241) - Event-sourced agent runtime whose self-improvement loop gates every proposed repair behind static checks, sandbox execution, and held-out evaluation before adoption, keeping the full decision trail replayable.
 - 📄 **Paper** [Agentic CLEAR: Automating Multi-Level Evaluation of LLM Agents](https://arxiv.org/abs/2605.22608) - Automated evaluation framework from IBM Research that grades agent behavior at system, trace, and node granularity without predefined error taxonomies, producing feedback aligned with human-annotated errors and predictive of task success.
 - 📄 **Paper** [Diagnosis-Driven Automatic Repair for Agentic Workflow via Symbolic Inference](https://arxiv.org/abs/2607.02882) - FlowFixer converts runs of platform-built agentic workflows (Dify, Coze, n8n) into symbolic traces, infers correctness specs and node dependencies to localize root-cause failures, and generates targeted repairs at a 71.3% success rate.
@@ -734,7 +784,7 @@ This section focuses on durable loop state and cross-run context. For context-wi
 - 🧰 **Tool** [GenericAgent](https://github.com/lsdefine/GenericAgent) - Self-evolving agent that grows a skill tree from a small seed, crystallizing completed runs into layered memory and reusable skills, with a master-worker mode for long-horizon goals.
 - 📄 **Paper** [Self-GC: Self-Governing Context for Long-Horizon LLM Agents](https://arxiv.org/abs/2607.00692) - Governs long-horizon agent context as indexed lifecycle objects in an explicit nod to garbage collection, with a side-channel planner proposing fold, mask, and prune actions under harness-enforced recoverable sidecars, cutting production input tokens by 10-15%.
 - 📄 **Paper** [CompactionRL: Reinforcement Learning with Context Compaction for Long-Horizon Agents](https://arxiv.org/abs/2607.05378) - Reinforcement-learning method that jointly optimizes task execution and compaction-summary generation so long-horizon agents can continue past finite context windows, lifting GLM-4.5-Air to 66.8% on SWE-bench Verified and shipping in the GLM-5.2 pipeline.
-- 📄 **Paper** [SelfMem: Self-Optimizing Memory for AI Agents](https://arxiv.org/abs/2607.03726) - Memory framework in which the agent autonomously optimizes its own storage, retrieval, and summarization strategies per task instead of a fixed pipeline, reporting 40-49% gains over baselines on the BEAM benchmark at 100K-1M token contexts.
+- 📄 **Paper** [SelfMem: Self-Optimizing Memory for AI Agents](https://arxiv.org/abs/2607.03726) - Memory framework in which the agent autonomously optimizes its own storage, retrieval, and summarization strategies per task instead of a fixed pipeline, improving BEAM's official score over the strongest baseline by 48.7%, 40.8%, and 41.9% at 100K, 500K, and 1M tokens, respectively.
 - 📄 **Paper** [Memory-Orchestrated Semantic System (MOSS): An Auditable Agentic Memory Architecture](https://arxiv.org/abs/2607.04391) - Model-, storage-, and API-agnostic agent memory architecture where the agent drives symbolic retrieval over a structured relational database, making long-term memory auditable and reproducible instead of opaque embedding search, validated in a year-long deployment.
 - 📄 **Paper** [The Log Is the Agent: Event-Sourced Reactive Graphs for Auditable, Forkable Agentic Systems](https://arxiv.org/abs/2605.21997) - BabyAGI creator Yohei Nakajima makes an append-only event log the source of truth and the working graph a deterministic projection, giving long-running loops deterministic replay, cheap forking at any event, and end-to-end causal lineage from goal to model call.
 - 📝 **Blog** [Agentics: Memorizing Session Transcripts Isn't Useful](https://12gramsofcarbon.com/p/agentics-memorizing-session-transcripts) - From thousands of agent sessions at Nori, reports zero coding-task benefit from giving agents search over prior session transcripts and argues loop state belongs in distilled artifacts like commits and docs because agents never prune stale context.
@@ -829,11 +879,11 @@ How to measure loops rather than single runs: long-horizon, multi-session, relia
 - 🧪 **Benchmark** [SentinelBench: A Benchmark for Long-Running Monitoring Agents](https://arxiv.org/abs/2606.05342) - Microsoft Research benchmark of 100 tasks across 10 synthetic web environments that evaluates long-running monitoring agents on whether they wait or act appropriately, scoring task completion, response speed, and resource efficiency.
 - 🧪 **Benchmark** [SWE-Together: Evaluating Coding Agents in Interactive User Sessions](https://arxiv.org/abs/2606.29957) - Multi-session coding benchmark of 109 repository-level tasks reconstructed from 11,260 recorded user-agent sessions, replayed with an LLM user simulator and scored on final correctness and the number of corrective feedback turns.
 - 🧪 **Benchmark** [The Long-Horizon Task Mirage? Diagnosing Where and Why Agentic Systems Break](https://arxiv.org/abs/2604.11978) - Cross-domain diagnostic benchmark that scales task horizon through depth and breadth extension, then attributes failures across 3,100+ agent trajectories to a seven-category taxonomy via a trajectory-grounded LLM judge validated against human annotation.
-- 📄 **Paper** [Beyond pass@1: A Reliability Science Framework for Long-Horizon LLM Agents](https://arxiv.org/abs/2603.29231) - Reliability metrics for long-horizon agents (reliability decay, variance amplification, graceful degradation, meltdown onset) measured over roughly 24,000 episodes across 10 models, showing capability and reliability rankings diverge as tasks lengthen.
+- 📄 **Paper** [Beyond pass@1: A Reliability Science Framework for Long-Horizon LLM Agents](https://arxiv.org/abs/2603.29231) - Reliability metrics for long-horizon agents (reliability decay, variance amplification, graceful degradation, meltdown onset) measured over 23,392 episodes across 10 models, showing capability and reliability rankings diverge as tasks lengthen.
 - 🧪 **Benchmark** [SEAGym: An Evaluation Environment for Self-Evolving LLM Agents](https://arxiv.org/abs/2606.17546) - Evaluation environment that measures whether a self-evolving agent's modifications to prompts, memory, and tools generalize to held-out tasks, using train, validation, and test splits and cost metrics on Terminal-Bench 2.0 and HLE.
 - 🧪 **Benchmark** [EvoCode-Bench: Evaluating Coding Agents in Multi-Turn Iterative Interactions](https://arxiv.org/abs/2605.24110) - Benchmark of 26 evolving coding tasks across 227 evaluation rounds using cumulative executable tests to check that agents keep prior requirements working as specifications change, with top agents reaching only about 50% on multi-turn success metrics.
 - 📄 **Paper** [On the Reliability of Computer Use Agents](https://arxiv.org/abs/2604.17849) - Repeated-execution study on OSWorld decomposing why computer-use agents fail tasks they previously completed, separating execution stochasticity, task-specification ambiguity, and behavioral variability as distinct causes of unreliability.
-- 📄 **Paper** [AgentLens: Revealing the Lucky Pass Problem in SWE-Agent Evaluation](https://arxiv.org/abs/2605.12925) - Grades over 2,600 SWE-agent trajectories across eight models to show that a meaningful share of passes are lucky trial-and-error successes, replacing binary pass/fail with process-quality tiers that shift model rankings.
+- 📄 **Paper** [AgentLens: Revealing the Lucky Pass Problem in SWE-Agent Evaluation](https://arxiv.org/abs/2605.12925) - Grades 2,614 SWE-agent trajectories across eight models to show that 10.7% of passing trajectories in its 1,815-trajectory evaluation subset are lucky trial-and-error successes, replacing binary pass/fail with process-quality tiers that shift model rankings.
 - 🧪 **Benchmark** [ORLoopBench: Solver-in-the-Loop Benchmarks for Self-Correction](https://arxiv.org/abs/2601.21008) - Formalizes infeasible-model debugging as a solver-in-the-loop process where each action triggers solver re-execution and infeasibility recomputation, giving deterministic verification for iterative repair in operations research.
 - 🧪 **Benchmark** [LongDS-Bench: On the Failure of Long-Horizon Agentic Data Analysis](https://arxiv.org/abs/2605.30434) - Benchmark of 68 real-world data-analysis tasks built from Kaggle notebooks spanning 2,225 interactive turns, finding that long-horizon errors account for 52-69% of agent failures and that maintaining a correct analytical state is the core bottleneck.
 - 🧪 **Benchmark** [MemoryArena: Benchmarking Agent Memory in Interdependent Multi-Session Agentic Tasks](https://arxiv.org/abs/2602.16313) - Multi-session benchmark of interdependent agentic tasks where agents must distill earlier sessions into memory and use it to guide later actions, showing that near-saturated scores on long-context memory benchmarks fail to transfer.
@@ -936,8 +986,8 @@ Skeptical takes and measured failure modes, because a loop that cannot fail safe
 
 Neighboring curated lists for the layers around loop engineering.
 
-- 🧭 **List** [Awesome Harness Engineering](https://github.com/ai-boost/awesome-harness-engineering) - Comprehensive list for the agent harness layer that Loop Engineering builds on.
-- 🧭 **List** [Awesome Harness Engineering](https://github.com/walkinglabs/awesome-harness-engineering) - High-signal harness list with strong categories for context, guardrails, specs, evals, runtimes, and benchmarks.
+- 🧭 **List** [Awesome Harness Engineering by ai-boost](https://github.com/ai-boost/awesome-harness-engineering) - Comprehensive list for the agent harness layer that Loop Engineering builds on.
+- 🧭 **List** [Awesome Harness Engineering by walkinglabs](https://github.com/walkinglabs/awesome-harness-engineering) - High-signal harness list with strong categories for context, guardrails, specs, evals, runtimes, and benchmarks.
 - 🧭 **List** [Awesome Agent Harness](https://github.com/AutoJunjie/awesome-agent-harness) - Curated tools and resources for environments, constraints, and feedback around coding agents.
 - 🧭 **List** [Awesome Context Engineering](https://github.com/Meirtz/Awesome-Context-Engineering) - Survey-style list for context engineering across LLMs and agents.
 - 🧭 **List** [Awesome Prompt Engineering](https://github.com/promptslab/Awesome-Prompt-Engineering) - Classic adjacent list for prompt techniques and prompting resources.
@@ -1047,16 +1097,35 @@ Each row in `data/resources.jsonl` (and the equivalent `data/resources.csv`) is
 | `novelty` | Why the entry is distinct within its section. |
 | `impact` | Who benefits and how. |
 | `signal` | Why the entry is trustworthy (adoption, venue, authorship). |
-| `signal_strength` | Coarse confidence label (`strong`, `medium`, `contextual`). |
+| `signal_strength` | Calibrated confidence label (`high`, `medium`, `contextual`, or `unverified`). |
 | `source_readme` | README file the row came from. |
 | `source_line` | 1-based line number in that README. |
 | `source_url` | Permalink to the source line on GitHub. |
+| `date_added` | First-seen date when tracked; blank for resources added before tracking began. |
+| `collection` | Task-oriented atlas group: Learn, Design, Build, Persist, Verify, Govern, or Apply. |
+| `collection_slug` | Stable lowercase slug for the task-oriented collection. |
+| `user_goal` | Reader outcome served by the collection. |
+| `lifecycle_stages` | Semicolon-delimited Loop Contract stages addressed by the resource. |
+| `audience` | Semicolon-delimited intended audiences, such as newcomer, builder, operator, evaluator, or security. |
+| `evidence_class` | Source-provenance category, independent of popularity or editorial judgment. |
+| `source_status` | Latest audit result: `ok`, `restricted`, `local_ok`, or an explicit failure state. |
+| `canonical_url` | Redirect-resolved or repository-canonical URL used for source verification. |
+| `source_title` | Title metadata retrieved during the latest source audit. |
+| `source_description` | Description or abstract metadata retrieved from the primary source. |
+| `github_repo` | Normalized GitHub `owner/repository` identifier where applicable. |
+| `github_stars` | GitHub star count captured at audit time; context, not a reliability score. |
+| `github_forks` | GitHub fork count captured at audit time. |
+| `github_license` | Repository license identifier reported by GitHub where available. |
+| `github_updated_at` | Latest repository update timestamp reported by GitHub. |
+| `arxiv_id` | arXiv identifier extracted from a primary paper URL where applicable. |
+| `audited_at` | UTC timestamp for the source metadata and availability snapshot. |
 
 ## Considerations and Limitations
 
 - Annotations are written and curated by a single maintainer; they summarize each resource and are not the authors' own abstracts.
 - URLs are third-party and can rot; `data/resource_source_audit.csv` carries a retrieval-time reachability snapshot, but it is a point-in-time check, not a live guarantee.
 - `row_id` and `source_line` are positional and change when the README is reordered; join on `url` for anything durable.
+- `signal_strength` is calibrated as `high`, `medium`, `contextual`, or `unverified`; GitHub stars and forks are reported only as adoption context.
 - Coverage skews toward English-language, publicly available resources.
 
 ## Changelog

docs/index.html

@@ -54,7 +54,7 @@
       }
       a { color: var(--blue); text-decoration-thickness: 1px; text-underline-offset: 3px; }
       a:hover { color: #0f49bd; }
-      button, input { font: inherit; }
+      button, input, select { font: inherit; }
       .wrap { width: min(var(--maxw), calc(100% - 48px)); margin: 0 auto; }
       .skip {
         position: absolute;
@@ -66,10 +66,11 @@
         padding: 10px 16px;
       }
       .skip:focus { left: 0; }
-      a:focus-visible, button:focus-visible, input:focus-visible, summary:focus-visible {
+      a:focus-visible, button:focus-visible, input:focus-visible, select:focus-visible, summary:focus-visible {
         outline: 3px solid rgba(21, 94, 239, 0.32);
         outline-offset: 3px;
       }
+      .sr-only { position: absolute; width: 1px; height: 1px; padding: 0; margin: -1px; overflow: hidden; clip: rect(0, 0, 0, 0); white-space: nowrap; border: 0; }
 
       header.site {
         position: sticky;
@@ -224,10 +225,72 @@
       .level .title { font-weight: 750; }
       .level .detail { color: var(--muted); font-size: 0.9rem; }
 
-      .resource-links { display: grid; grid-template-columns: repeat(3, minmax(0, 1fr)); border-top: 1px solid var(--line-strong); }
-      .resource-links a { min-height: 58px; padding: 16px 18px 16px 0; border-bottom: 1px solid var(--line); color: var(--ink); font-size: 0.9rem; font-weight: 650; text-decoration: none; }
-      .resource-links a:hover { color: var(--blue); }
-      .resource-links a:nth-child(3n+2), .resource-links a:nth-child(3n+3) { padding-left: 18px; }
+      .resource-explorer { border-top: 1px solid var(--line-strong); }
+      .resource-toolbar {
+        display: grid;
+        grid-template-columns: minmax(260px, 1.45fr) repeat(3, minmax(150px, 0.7fr)) auto;
+        gap: 10px;
+        padding: 16px 0;
+        border-bottom: 1px solid var(--line);
+      }
+      .resource-input, .resource-select {
+        width: 100%;
+        min-height: 44px;
+        border: 1px solid var(--line-strong);
+        border-radius: 6px;
+        padding: 9px 12px;
+        color: var(--ink);
+        background: var(--surface);
+      }
+      .resource-input::placeholder { color: var(--quiet); }
+      .resource-reset, .resource-more {
+        min-height: 44px;
+        border: 1px solid var(--line-strong);
+        border-radius: 6px;
+        padding: 9px 13px;
+        color: var(--ink);
+        background: var(--surface);
+        font-weight: 700;
+        cursor: pointer;
+      }
+      .resource-reset:hover, .resource-more:hover { border-color: var(--blue); color: var(--blue); }
+      .resource-collections { display: flex; flex-wrap: wrap; gap: 0; border-bottom: 1px solid var(--line); }
+      .resource-collection {
+        min-height: 42px;
+        border: 0;
+        border-bottom: 2px solid transparent;
+        padding: 9px 14px;
+        color: var(--muted);
+        background: transparent;
+        font-size: 0.82rem;
+        font-weight: 700;
+        cursor: pointer;
+      }
+      .resource-collection:hover { color: var(--ink); }
+      .resource-collection[aria-pressed="true"] { border-bottom-color: var(--blue); color: var(--blue); }
+      .resource-summary { display: flex; align-items: center; justify-content: space-between; gap: 20px; min-height: 58px; border-bottom: 1px solid var(--line-strong); }
+      .resource-count { margin: 0; color: var(--muted); font-size: 0.86rem; }
+      .resource-sort { width: auto; min-width: 176px; }
+      .resource-list { min-height: 180px; }
+      .resource-item {
+        display: grid;
+        grid-template-columns: minmax(0, 1fr) 180px;
+        gap: 28px;
+        padding: 17px 0;
+        border-bottom: 1px solid var(--line);
+      }
+      .resource-item a { color: var(--ink); font-weight: 750; text-decoration: none; }
+      .resource-item a:hover { color: var(--blue); }
+      .resource-item p { max-width: 850px; margin: 6px 0 0; color: var(--muted); font-size: 0.88rem; }
+      .resource-meta { margin-top: 7px; color: var(--quiet); font-family: "IBM Plex Mono", ui-monospace, monospace; font-size: 0.7rem; overflow-wrap: anywhere; }
+      .resource-side { color: var(--quiet); font-size: 0.74rem; text-align: right; }
+      .resource-signal { display: block; color: var(--green); font-weight: 700; text-transform: uppercase; }
+      .resource-signal.contextual { color: var(--violet); }
+      .resource-signal.medium { color: var(--blue); }
+      .resource-signal.unverified { color: var(--quiet); }
+      .resource-source-status { display: block; margin-top: 4px; }
+      .resource-actions { display: flex; justify-content: center; padding-top: 18px; }
+      .resource-empty, .resource-loading { margin: 0; padding: 28px 0; color: var(--muted); }
       details { border-top: 1px solid var(--line-strong); padding: 18px 0; }
       details:last-child { border-bottom: 1px solid var(--line-strong); }
       summary { cursor: pointer; font-weight: 750; }
@@ -276,11 +339,12 @@
         .stat:nth-child(even) { border-left: 1px solid var(--line); }
         .stat:nth-child(n+3) { border-top: 1px solid var(--line); }
         .stat:last-child { grid-column: 1 / -1; border-left: 0; }
-        .principles, .link-grid, .pattern-grid, .resource-links { grid-template-columns: repeat(2, minmax(0, 1fr)); }
+        .principles, .link-grid, .pattern-grid { grid-template-columns: repeat(2, minmax(0, 1fr)); }
         .level { grid-template-columns: 54px 1fr; }
         .level .detail { grid-column: 2; }
-        .resource-links a:nth-child(3n+2), .resource-links a:nth-child(3n+3) { padding-left: 0; }
-        .resource-links a:nth-child(even) { padding-left: 18px; }
+        .resource-toolbar { grid-template-columns: repeat(2, minmax(0, 1fr)); }
+        .resource-toolbar label:first-child { grid-column: 1 / -1; }
+        .resource-reset { width: 100%; }
         .footer-grid { grid-template-columns: 1fr 1fr; }
       }
       @media (max-width: 560px) {
@@ -291,10 +355,17 @@
         .lead { font-size: 1rem; }
         .mental-model { padding-top: 30px; }
         .hero-scene canvas { opacity: 0.68; }
-        .principles, .link-grid, .pattern-grid, .resource-links, .footer-grid { grid-template-columns: 1fr; }
+        .principles, .link-grid, .pattern-grid, .footer-grid { grid-template-columns: 1fr; }
         .principle + .principle, .card + .card, .pattern + .pattern { padding-left: 0; }
         .pattern { min-height: 88px; }
-        .resource-links a:nth-child(even) { padding-left: 0; }
+        .resource-toolbar { grid-template-columns: 1fr; }
+        .resource-toolbar label:first-child { grid-column: auto; }
+        .resource-collections { display: grid; grid-template-columns: repeat(4, minmax(0, 1fr)); overflow: visible; }
+        .resource-collection { min-width: 0; padding-inline: 4px; }
+        .resource-summary { align-items: stretch; flex-direction: column; gap: 10px; padding: 14px 0; }
+        .resource-sort { width: 100%; }
+        .resource-item { grid-template-columns: 1fr; gap: 8px; }
+        .resource-side { text-align: left; }
         .caption { text-align: left; }
         .level { grid-template-columns: 46px 1fr; }
         .diagram-scroll { overflow-x: auto; overscroll-behavior-inline: contain; }
@@ -308,7 +379,7 @@
         .hero-scene { display: none; }
       }
       @media print {
-        header.site, .menu-btn, .actions, .copy, .pattern-filter, .hero-scene { display: none; }
+        header.site, .menu-btn, .actions, .copy, .pattern-filter, .resource-toolbar, .resource-collections, .resource-summary, .resource-actions, .hero-scene { display: none; }
         body, footer { background: #fff; }
         section { padding: 24pt 0; break-inside: avoid; }
         .wrap { width: 100%; }
@@ -588,27 +659,95 @@
           <div class="section-head">
             <div>
               <p class="section-label">06 / Curated resources</p>
-              <h2>Browse the field by category.</h2>
+              <h2>Find the evidence for your next decision.</h2>
             </div>
-            <p class="section-intro">Primary sources, official docs, papers, benchmarks, and implementation-heavy work on AI-agent loops and coding-agent automation. Every dataset row includes a specific annotation plus contribution, novelty, impact, and signal fields.</p>
+            <p class="section-intro">Search 509 audited sources by goal, lifecycle stage, artifact type, and evidence class. Every result preserves its curated annotation and links directly to the canonical source.</p>
+          </div>
+          <div class="resource-explorer" id="resource-explorer">
+            <div class="resource-toolbar" role="search" aria-label="Search curated resources">
+              <label>
+                <span class="sr-only">Search resources</span>
+                <input class="resource-input" id="resource-search" type="search" placeholder="Search title, annotation, section, or lifecycle stage" autocomplete="off">
+              </label>
+              <label>
+                <span class="sr-only">Resource type</span>
+                <select class="resource-select" id="resource-type">
+                  <option value="">All resource types</option>
+                  <option value="Docs">Documentation</option>
+                  <option value="Tool">Tools</option>
+                  <option value="Paper">Papers</option>
+                  <option value="Benchmark">Benchmarks</option>
+                  <option value="Pattern">Patterns</option>
+                  <option value="Template">Templates</option>
+                  <option value="Blog">Practitioner writing</option>
+                  <option value="Critique">Critiques</option>
+                  <option value="List">Curated lists</option>
+                </select>
+              </label>
+              <label>
+                <span class="sr-only">Lifecycle stage</span>
+                <select class="resource-select" id="resource-stage">
+                  <option value="">All lifecycle stages</option>
+                  <option value="objective">Objective</option>
+                  <option value="trigger">Trigger</option>
+                  <option value="intake">Intake</option>
+                  <option value="workspace">Workspace</option>
+                  <option value="context">Context</option>
+                  <option value="delegation">Delegation</option>
+                  <option value="verification">Verification</option>
+                  <option value="state">State</option>
+                  <option value="budget">Budget</option>
+                  <option value="escalation">Escalation</option>
+                  <option value="exit">Exit</option>
+                  <option value="whole-loop">Whole loop</option>
+                </select>
+              </label>
+              <label>
+                <span class="sr-only">Evidence class</span>
+                <select class="resource-select" id="resource-evidence">
+                  <option value="">All evidence classes</option>
+                  <option value="official">Official documentation</option>
+                  <option value="research">Research</option>
+                  <option value="implementation">Implementations</option>
+                  <option value="benchmark">Benchmarks</option>
+                  <option value="practice">Practice and patterns</option>
+                  <option value="risk">Risk and ecosystem maps</option>
+                </select>
+              </label>
+              <button class="resource-reset" id="resource-reset" type="button">Reset</button>
+            </div>
+            <div class="resource-collections" role="group" aria-label="Filter by user goal">
+              <button class="resource-collection" type="button" data-collection="" aria-pressed="true">All</button>
+              <button class="resource-collection" type="button" data-collection="Learn" aria-pressed="false">Learn</button>
+              <button class="resource-collection" type="button" data-collection="Design" aria-pressed="false">Design</button>
+              <button class="resource-collection" type="button" data-collection="Build" aria-pressed="false">Build</button>
+              <button class="resource-collection" type="button" data-collection="Persist" aria-pressed="false">Persist</button>
+              <button class="resource-collection" type="button" data-collection="Verify" aria-pressed="false">Verify</button>
+              <button class="resource-collection" type="button" data-collection="Govern" aria-pressed="false">Govern</button>
+              <button class="resource-collection" type="button" data-collection="Apply" aria-pressed="false">Apply</button>
+            </div>
+            <div class="resource-summary">
+              <p class="resource-count" id="resource-count" aria-live="polite">Loading resource index...</p>
+              <label>
+                <span class="sr-only">Sort resources</span>
+                <select class="resource-select resource-sort" id="resource-sort">
+                  <option value="curated">Curated order</option>
+                  <option value="stars">GitHub stars</option>
+                  <option value="recent">Recently added</option>
+                </select>
+              </label>
+            </div>
+            <div class="resource-list" id="resource-list" aria-busy="true">
+              <p class="resource-loading">Loading audited resources...</p>
+            </div>
+            <p class="resource-empty" id="resource-empty" hidden>No resources match the current filters.</p>
+            <div class="resource-actions">
+              <button class="resource-more" id="resource-more" type="button" hidden>Show more</button>
+            </div>
+            <noscript>
+              <p class="resource-loading"><a href="https://github.com/ChaoYue0307/awesome-loop-engineering#start-here">Browse the full resource library on GitHub</a> or <a href="https://huggingface.co/datasets/cy0307/awesome-loop-engineering">filter the dataset on Hugging Face</a>.</p>
+            </noscript>
           </div>
-          <nav class="resource-links" aria-label="Resource categories">
-            <a href="https://github.com/ChaoYue0307/awesome-loop-engineering#start-here">Start Here</a>
-            <a href="https://github.com/ChaoYue0307/awesome-loop-engineering#core-loop-primitives">Core Loop Primitives</a>
-            <a href="https://github.com/ChaoYue0307/awesome-loop-engineering#official-runtime-guides">Runtime Guides</a>
-            <a href="https://github.com/ChaoYue0307/awesome-loop-engineering#research-foundations">Research Foundations</a>
-            <a href="https://github.com/ChaoYue0307/awesome-loop-engineering#agent-workflow-patterns">Agent Workflow Patterns</a>
-            <a href="https://github.com/ChaoYue0307/awesome-loop-engineering#coding-agent-loop-systems">Coding-Agent Loop Systems</a>
-            <a href="https://github.com/ChaoYue0307/awesome-loop-engineering#verification-and-feedback-gates">Verification and Feedback</a>
-            <a href="https://github.com/ChaoYue0307/awesome-loop-engineering#securing-unattended-loops">Securing Unattended Loops</a>
-            <a href="https://github.com/ChaoYue0307/awesome-loop-engineering#state-memory-and-context-persistence">State and Memory</a>
-            <a href="https://github.com/ChaoYue0307/awesome-loop-engineering#orchestration-and-multi-agent-delegation">Orchestration</a>
-            <a href="https://github.com/ChaoYue0307/awesome-loop-engineering#benchmarks-and-evaluation">Benchmarks and Evaluation</a>
-            <a href="https://github.com/ChaoYue0307/awesome-loop-engineering#operations-playbooks">Operations Playbooks</a>
-            <a href="https://github.com/ChaoYue0307/awesome-loop-engineering#critiques-risks-and-limitations">Critiques and Risks</a>
-            <a href="https://github.com/ChaoYue0307/awesome-loop-engineering#adjacent-awesome-lists">Adjacent Awesome Lists</a>
-            <a href="https://huggingface.co/datasets/cy0307/awesome-loop-engineering">Search all 509 rows on Hugging Face</a>
-          </nav>
         </div>
       </section>
 
@@ -749,6 +888,184 @@
           if (empty) empty.hidden = shown !== 0;
         });
       })();
+      (function () {
+        var explorer = document.getElementById('resource-explorer');
+        if (!explorer || !window.fetch) return;
+
+        var search = document.getElementById('resource-search');
+        var type = document.getElementById('resource-type');
+        var stage = document.getElementById('resource-stage');
+        var evidence = document.getElementById('resource-evidence');
+        var sort = document.getElementById('resource-sort');
+        var reset = document.getElementById('resource-reset');
+        var more = document.getElementById('resource-more');
+        var list = document.getElementById('resource-list');
+        var count = document.getElementById('resource-count');
+        var empty = document.getElementById('resource-empty');
+        var collectionButtons = Array.prototype.slice.call(explorer.querySelectorAll('.resource-collection'));
+        var evidenceGroups = {
+          official: ['official-documentation'],
+          research: ['research-preprint', 'research-paper'],
+          implementation: ['source-implementation', 'implementation'],
+          benchmark: ['benchmark'],
+          practice: ['operational-pattern', 'practitioner-analysis', 'reusable-artifact', 'repository-native', 'technical-documentation'],
+          risk: ['risk-analysis', 'curated-index']
+        };
+        var state = { collection: '', limit: 24 };
+        var resources = [];
+
+        function words(value) {
+          return String(value || '').replace(/-/g, ' ');
+        }
+
+        function stages(value) {
+          return String(value || '').split(';').filter(Boolean);
+        }
+
+        function formatNumber(value) {
+          var number = Number(value || 0);
+          if (!number) return '';
+          return number.toLocaleString ? number.toLocaleString() : String(number);
+        }
+
+        function createText(tag, className, value) {
+          var element = document.createElement(tag);
+          if (className) element.className = className;
+          element.textContent = value;
+          return element;
+        }
+
+        function resourceMatches(resource) {
+          var query = search.value.trim().toLowerCase();
+          var haystack = [
+            resource.title,
+            resource.annotation,
+            resource.section,
+            resource.collection,
+            resource.user_goal,
+            resource.lifecycle_stages,
+            resource.audience,
+            resource.evidence_class
+          ].join(' ').toLowerCase();
+          var evidenceSet = evidenceGroups[evidence.value] || [];
+
+          return (!query || haystack.indexOf(query) !== -1) &&
+            (!type.value || resource.resource_type === type.value) &&
+            (!stage.value || stages(resource.lifecycle_stages).indexOf(stage.value) !== -1) &&
+            (!evidence.value || evidenceSet.indexOf(resource.evidence_class) !== -1) &&
+            (!state.collection || resource.collection === state.collection);
+        }
+
+        function sortedResources(filtered) {
+          return filtered.slice().sort(function (a, b) {
+            if (sort.value === 'stars') {
+              return Number(b.github_stars || 0) - Number(a.github_stars || 0) || a._index - b._index;
+            }
+            if (sort.value === 'recent') {
+              return (Date.parse(b.date_added || '') || 0) - (Date.parse(a.date_added || '') || 0) || a._index - b._index;
+            }
+            return a._index - b._index;
+          });
+        }
+
+        function renderItem(resource) {
+          var item = document.createElement('article');
+          var main = document.createElement('div');
+          var link = createText('a', '', resource.title);
+          var meta = [
+            resource.resource_type,
+            resource.collection,
+            stages(resource.lifecycle_stages).map(words).join(' / '),
+            words(resource.evidence_class)
+          ].filter(Boolean).join(' | ');
+          var side = document.createElement('div');
+          var signalClass = 'resource-signal ' + String(resource.signal_strength || 'unverified').toLowerCase();
+          var signal = createText('span', signalClass, words(resource.signal_strength || 'unverified') + ' signal');
+          var sourceStatus = resource.source_status === 'restricted' ? 'Access restricted' : 'Source checked';
+          var starCount = formatNumber(resource.github_stars);
+
+          item.className = 'resource-item';
+          link.href = resource.canonical_url || resource.url;
+          main.appendChild(link);
+          main.appendChild(createText('p', '', resource.annotation));
+          main.appendChild(createText('div', 'resource-meta', meta));
+          side.className = 'resource-side';
+          side.appendChild(signal);
+          if (starCount) side.appendChild(createText('span', 'resource-source-status', starCount + ' GitHub stars'));
+          side.appendChild(createText('span', 'resource-source-status', sourceStatus));
+          item.appendChild(main);
+          item.appendChild(side);
+          return item;
+        }
+
+        function render() {
+          var filtered = sortedResources(resources.filter(resourceMatches));
+          var visible = filtered.slice(0, state.limit);
+          var fragment = document.createDocumentFragment();
+
+          visible.forEach(function (resource) { fragment.appendChild(renderItem(resource)); });
+          list.replaceChildren(fragment);
+          list.setAttribute('aria-busy', 'false');
+          count.textContent = filtered.length + ' of ' + resources.length + ' resources' + (visible.length < filtered.length ? ' | showing ' + visible.length : '');
+          empty.hidden = filtered.length !== 0;
+          more.hidden = visible.length >= filtered.length;
+          if (!more.hidden) more.textContent = 'Show ' + Math.min(24, filtered.length - visible.length) + ' more';
+        }
+
+        function resetLimitAndRender() {
+          state.limit = 24;
+          render();
+        }
+
+        search.addEventListener('input', resetLimitAndRender);
+        [type, stage, evidence, sort].forEach(function (control) {
+          control.addEventListener('change', resetLimitAndRender);
+        });
+        collectionButtons.forEach(function (button) {
+          button.addEventListener('click', function () {
+            state.collection = button.getAttribute('data-collection') || '';
+            collectionButtons.forEach(function (candidate) {
+              candidate.setAttribute('aria-pressed', candidate === button ? 'true' : 'false');
+            });
+            resetLimitAndRender();
+          });
+        });
+        more.addEventListener('click', function () {
+          state.limit += 24;
+          render();
+        });
+        reset.addEventListener('click', function () {
+          search.value = '';
+          type.value = '';
+          stage.value = '';
+          evidence.value = '';
+          sort.value = 'curated';
+          state.collection = '';
+          collectionButtons.forEach(function (button) {
+            button.setAttribute('aria-pressed', button.getAttribute('data-collection') === '' ? 'true' : 'false');
+          });
+          resetLimitAndRender();
+          search.focus();
+        });
+
+        fetch('assets/resources.json')
+          .then(function (response) {
+            if (!response.ok) throw new Error('Resource index request failed');
+            return response.json();
+          })
+          .then(function (payload) {
+            resources = (payload.resources || []).map(function (resource, index) {
+              resource._index = index;
+              return resource;
+            });
+            render();
+          })
+          .catch(function () {
+            list.replaceChildren(createText('p', 'resource-loading', 'The resource index could not be loaded. Browse the complete library on GitHub or Hugging Face.'));
+            list.setAttribute('aria-busy', 'false');
+            count.textContent = 'Resource index unavailable';
+          });
+      })();
       Array.prototype.forEach.call(document.querySelectorAll('.copy'), function (button) {
         var label = button.querySelector('span') || button;
         button.addEventListener('click', function () {

examples/runnable/codex-automation.md

@@ -1,13 +1,13 @@
 # Codex Automation Variant
 
-> Template, not a guarantee of product behavior. Confirm triggers, worktree behavior, and scopes in the [Codex automations docs](https://developers.openai.com/codex/app/automations) before relying on this unattended.
+> Template, not a guarantee of product behavior. Confirm triggers, worktree behavior, and scopes in the [scheduled tasks docs](https://learn.chatgpt.com/docs/automations?surface=app) before relying on this unattended.
 
 A Codex automation runs a saved task on a cadence or in response to events, in an isolated workspace. Use it for recurring background work such as triage, repair, or digest loops that should run without you starting each one.
 
 ## When to use
 
 - The work recurs and should run unattended in the background.
-- It fits an isolated, per-run [worktree](https://developers.openai.com/codex/app/worktrees).
+- It fits an isolated, per-run [worktree](https://learn.chatgpt.com/docs/environments/git-worktrees).
 - You want it on the Codex runtime rather than your own infrastructure.
 
 ## Shape

meta/hf_card_body.md

@@ -20,16 +20,35 @@ Each row in `data/resources.jsonl` (and the equivalent `data/resources.csv`) is
 | `novelty` | Why the entry is distinct within its section. |
 | `impact` | Who benefits and how. |
 | `signal` | Why the entry is trustworthy (adoption, venue, authorship). |
-| `signal_strength` | Coarse confidence label (`strong`, `medium`, `contextual`). |
+| `signal_strength` | Calibrated confidence label (`high`, `medium`, `contextual`, or `unverified`). |
 | `source_readme` | README file the row came from. |
 | `source_line` | 1-based line number in that README. |
 | `source_url` | Permalink to the source line on GitHub. |
+| `date_added` | First-seen date when tracked; blank for resources added before tracking began. |
+| `collection` | Task-oriented atlas group: Learn, Design, Build, Persist, Verify, Govern, or Apply. |
+| `collection_slug` | Stable lowercase slug for the task-oriented collection. |
+| `user_goal` | Reader outcome served by the collection. |
+| `lifecycle_stages` | Semicolon-delimited Loop Contract stages addressed by the resource. |
+| `audience` | Semicolon-delimited intended audiences, such as newcomer, builder, operator, evaluator, or security. |
+| `evidence_class` | Source-provenance category, independent of popularity or editorial judgment. |
+| `source_status` | Latest audit result: `ok`, `restricted`, `local_ok`, or an explicit failure state. |
+| `canonical_url` | Redirect-resolved or repository-canonical URL used for source verification. |
+| `source_title` | Title metadata retrieved during the latest source audit. |
+| `source_description` | Description or abstract metadata retrieved from the primary source. |
+| `github_repo` | Normalized GitHub `owner/repository` identifier where applicable. |
+| `github_stars` | GitHub star count captured at audit time; context, not a reliability score. |
+| `github_forks` | GitHub fork count captured at audit time. |
+| `github_license` | Repository license identifier reported by GitHub where available. |
+| `github_updated_at` | Latest repository update timestamp reported by GitHub. |
+| `arxiv_id` | arXiv identifier extracted from a primary paper URL where applicable. |
+| `audited_at` | UTC timestamp for the source metadata and availability snapshot. |
 
 ## Considerations and Limitations
 
 - Annotations are written and curated by a single maintainer; they summarize each resource and are not the authors' own abstracts.
 - URLs are third-party and can rot; `data/resource_source_audit.csv` carries a retrieval-time reachability snapshot, but it is a point-in-time check, not a live guarantee.
 - `row_id` and `source_line` are positional and change when the README is reordered; join on `url` for anything durable.
+- `signal_strength` is calibrated as `high`, `medium`, `contextual`, or `unverified`; GitHub stars and forks are reported only as adoption context.
 - Coverage skews toward English-language, publicly available resources.
 
 ## Changelog

scripts/audit_resource_sources.py

@@ -67,6 +67,7 @@ class MetadataParser(HTMLParser):
         self.in_title = False
         self.title_parts: list[str] = []
         self.description = ""
+        self.description_priority = 0
 
     @property
     def title(self) -> str:
@@ -76,13 +77,22 @@ class MetadataParser(HTMLParser):
         if tag.lower() == "title":
             self.in_title = True
             return
-        if tag.lower() != "meta" or self.description:
+        if tag.lower() != "meta":
             return
 
         attr_map = {key.lower(): value or "" for key, value in attrs}
         name = (attr_map.get("name") or attr_map.get("property") or "").lower()
-        if name in {"description", "og:description", "twitter:description"}:
-            self.description = clean(attr_map.get("content"))
+        priority = {
+            "description": 1,
+            "og:description": 1,
+            "twitter:description": 1,
+            "dc.description": 2,
+            "citation_abstract": 3,
+        }.get(name, 0)
+        candidate = clean(attr_map.get("content"))
+        if candidate and priority > self.description_priority:
+            self.description = candidate
+            self.description_priority = priority
 
     def handle_endtag(self, tag: str) -> None:
         if tag.lower() == "title":
@@ -93,6 +103,18 @@ class MetadataParser(HTMLParser):
             self.title_parts.append(data)
 
 
+class RedirectHandler(urllib.request.HTTPRedirectHandler):
+    """Teach Python 3.9's urllib opener to follow permanent HTTP 308 moves."""
+
+    def http_error_308(self, req, fp, code, msg, headers):  # type: ignore[no-untyped-def]
+        return self.http_error_302(req, fp, code, msg, headers)
+
+    def redirect_request(self, req, fp, code, msg, headers, newurl):  # type: ignore[no-untyped-def]
+        if code == 308 and req.get_method() in {"GET", "HEAD"}:
+            code = 307
+        return super().redirect_request(req, fp, code, msg, headers, newurl)
+
+
 def read_rows() -> list[dict[str, str]]:
     with RESOURCES_CSV.open(encoding="utf-8", newline="") as handle:
         return list(csv.DictReader(handle))
@@ -127,6 +149,10 @@ def html_metadata(body: bytes, content_type: str) -> tuple[str, str]:
 
 def fetch_url(url: str, timeout: float, attempts: int) -> dict[str, str]:
     context = ssl._create_unverified_context()
+    opener = urllib.request.build_opener(
+        urllib.request.HTTPSHandler(context=context),
+        RedirectHandler(),
+    )
     last_error = ""
 
     for attempt in range(1, attempts + 1):
@@ -137,7 +163,7 @@ def fetch_url(url: str, timeout: float, attempts: int) -> dict[str, str]:
                 headers={"User-Agent": "awesome-loop-engineering-source-audit"},
             )
             try:
-                with urllib.request.urlopen(request, timeout=timeout, context=context) as response:
+                with opener.open(request, timeout=timeout) as response:
                     content_type = response.headers.get("content-type", "")
                     body = response.read(750_000) if method == "GET" else b""
                     title, description = html_metadata(body, content_type)

scripts/export_resource_dataset.py

@@ -17,6 +17,8 @@ ROOT = Path(__file__).resolve().parents[1]
 README = ROOT / "README.md"
 CSV_PATH = ROOT / "data" / "resources.csv"
 JSONL_PATH = ROOT / "data" / "resources.jsonl"
+SITE_JSON_PATH = ROOT / "docs" / "assets" / "resources.json"
+AUDIT_PATH = ROOT / "data" / "resource_source_audit.csv"
 FIRST_SEEN_PATH = ROOT / "data" / "first_seen.json"
 SOURCE_URL = "https://github.com/ChaoYue0307/awesome-loop-engineering/blob/main/README.md"
 
@@ -59,8 +61,124 @@ FIELDS = [
     "source_line",
     "source_url",
     "date_added",
+    "collection",
+    "collection_slug",
+    "user_goal",
+    "lifecycle_stages",
+    "audience",
+    "evidence_class",
+    "source_status",
+    "canonical_url",
+    "source_title",
+    "source_description",
+    "github_repo",
+    "github_stars",
+    "github_forks",
+    "github_license",
+    "github_updated_at",
+    "arxiv_id",
+    "audited_at",
 ]
 
+SITE_FIELDS = [
+    "row_id",
+    "title",
+    "url",
+    "canonical_url",
+    "annotation",
+    "resource_type",
+    "collection",
+    "user_goal",
+    "section",
+    "section_slug",
+    "lifecycle_stages",
+    "audience",
+    "evidence_class",
+    "signal_strength",
+    "source_status",
+    "github_stars",
+    "date_added",
+]
+
+COLLECTIONS = {
+    "Concept Guides": ("Learn", "Understand the field and its boundaries."),
+    "Start Here": ("Learn", "Understand the field and its boundaries."),
+    "Research Foundations": ("Learn", "Understand the field and its boundaries."),
+    "Pattern Library": ("Design", "Specify a loop contract and operating pattern."),
+    "Core Loop Primitives": ("Design", "Specify a loop contract and operating pattern."),
+    "Agent Workflow Patterns": ("Design", "Specify a loop contract and operating pattern."),
+    "Official Runtime Guides": ("Build", "Choose runtimes, tools, and delegation surfaces."),
+    "Coding-Agent Loop Systems": ("Build", "Choose runtimes, tools, and delegation surfaces."),
+    "Orchestration And Multi-Agent Delegation": ("Build", "Choose runtimes, tools, and delegation surfaces."),
+    "State, Memory, And Context Persistence": ("Persist", "Carry context, state, and receipts across runs."),
+    "Verification And Feedback Gates": ("Verify", "Gate progress with tests, evals, and evidence."),
+    "Benchmarks And Evaluation": ("Verify", "Gate progress with tests, evals, and evidence."),
+    "Securing Unattended Loops": ("Govern", "Bound permissions, cost, failure, and escalation."),
+    "Operations Playbooks": ("Govern", "Bound permissions, cost, failure, and escalation."),
+    "Critiques, Risks, And Limitations": ("Govern", "Bound permissions, cost, failure, and escalation."),
+    "Templates And Patterns": ("Apply", "Reuse, adapt, and contribute concrete loop artifacts."),
+    "Examples And Schema": ("Apply", "Reuse, adapt, and contribute concrete loop artifacts."),
+    "Community Gallery": ("Apply", "Reuse, adapt, and contribute concrete loop artifacts."),
+    "Adjacent Awesome Lists": ("Apply", "Reuse, adapt, and contribute concrete loop artifacts."),
+    "Discovery And Distribution": ("Apply", "Reuse, adapt, and contribute concrete loop artifacts."),
+    "Roadmap And Discussion": ("Apply", "Reuse, adapt, and contribute concrete loop artifacts."),
+}
+
+COLLECTION_IMPACT = {
+    "Learn": "understand the evidence, vocabulary, and lineage behind recurring agent systems",
+    "Design": "turn a recurring-agent idea into an explicit loop contract",
+    "Build": "choose an implementation surface for repeatable agent work",
+    "Persist": "carry context, state, and receipts across runs and failures",
+    "Verify": "measure progress and gate completion with repeatable evidence",
+    "Govern": "bound risk before recurring or unattended execution",
+    "Apply": "reuse a concrete artifact or connect it to the wider ecosystem",
+}
+
+SECTION_STAGE_DEFAULTS = {
+    "Concept Guides": ["whole-loop"],
+    "Start Here": ["whole-loop"],
+    "Pattern Library": ["whole-loop"],
+    "Research Foundations": ["whole-loop"],
+    "Official Runtime Guides": ["workspace", "context", "delegation", "state"],
+    "Agent Workflow Patterns": ["delegation", "verification"],
+    "Coding-Agent Loop Systems": ["workspace", "delegation", "verification", "state"],
+    "Verification And Feedback Gates": ["verification"],
+    "Securing Unattended Loops": ["workspace", "budget", "escalation"],
+    "State, Memory, And Context Persistence": ["context", "state"],
+    "Orchestration And Multi-Agent Delegation": ["delegation", "state"],
+    "Benchmarks And Evaluation": ["verification"],
+    "Operations Playbooks": ["trigger", "intake", "budget", "escalation", "exit"],
+    "Critiques, Risks, And Limitations": ["budget", "escalation", "exit"],
+    "Templates And Patterns": ["whole-loop"],
+    "Examples And Schema": ["whole-loop"],
+    "Community Gallery": ["whole-loop"],
+}
+
+STAGE_RULES = [
+    ("objective", r"\bobjective(?:s)?\b|\bgoal(?:s)?\b|success criteria"),
+    ("trigger", r"\btrigger(?:s|ed)?\b|\bschedul(?:e|ed|ing)\b|\bcadence\b|\bcron\b|\bevent-driven\b"),
+    ("intake", r"\bintake\b|\bqueue(?:s)?\b|\bdiscover(?:y|s|ed)?\b|\btriage\b|\bissue(?:s)?\b"),
+    ("workspace", r"\bworkspace(?:s)?\b|\bworktree(?:s)?\b|\bsandbox(?:es|ed|ing)?\b|\bpermission(?:s)?\b|\btool(?:s)?\b"),
+    ("context", r"\bcontext\b|\bmemory\b|\bmemories\b|\bretrieval\b|\bdocument(?:s)?\b"),
+    ("delegation", r"\bdelegat(?:e|es|ed|ion)\b|\bmulti-agent\b|\bsubagent(?:s)?\b|\bhandoff(?:s)?\b|\borchestrat(?:e|es|ed|ion|or|ors)\b"),
+    ("verification", r"\bverif(?:y|ies|ied|ication)\b|\beval(?:s|uation)?\b|\btest(?:s|ed|ing)?\b|\bbenchmark(?:s)?\b|\bgrader(?:s)?\b|\bcritic(?:s)?\b"),
+    ("state", r"\bstate(?:ful)?\b|\bpersist(?:s|ed|ence|ent)?\b|\bcheckpoint(?:s|ed|ing)?\b|\breplay\b|\breceipt(?:s)?\b"),
+    ("budget", r"\bbudget(?:s)?\b|\bcost(?:s)?\b|\btoken(?:s)?\b|\bretr(?:y|ies)\b|\btimeout(?:s)?\b"),
+    ("escalation", r"\bescalat(?:e|es|ed|ion)\b|\bhuman(?:-in-the-loop)?\b|\bapproval(?:s)?\b|\bhandoff\b"),
+    ("exit", r"\bexit\b|\bstop(?:s|ped|ping)?\b|\bcompletion\b|\bdone\b|\btermination\b"),
+]
+
+OFFICIAL_DOC_DOMAINS = {
+    "learn.chatgpt.com",
+    "code.claude.com",
+    "docs.anthropic.com",
+    "docs.github.com",
+    "developers.openai.com",
+    "modelcontextprotocol.io",
+    "opentelemetry.io",
+    "openai.github.io",
+}
+
 SECTION_IMPACT = {
     "Concept Guides": "Clarifies the scope, vocabulary, and boundaries of Loop Engineering so the list does not drift into generic agent material.",
     "Start Here": "Gives readers the origin story and first-principles framing for the new AI/coding-agent use of Loop Engineering.",
@@ -162,55 +280,142 @@ def classify_url(url: str) -> tuple[str, str]:
     return "local_path", ""
 
 
-def key_contribution(resource_type: str, annotation: str) -> str:
-    lead = clean(annotation).rstrip(".")
-    if resource_type in {"Paper", "Blog", "Docs"}:
-        return lead + "."
-    if resource_type == "Tool":
-        return f"Provides an implementation surface for loop builders: {lead}."
-    if resource_type == "Benchmark":
-        return f"Provides an evaluation signal for loop builders: {lead}."
-    if resource_type == "Pattern":
-        return f"Provides a reusable loop pattern: {lead}."
-    if resource_type == "Template":
-        return f"Provides a reusable project artifact: {lead}."
-    if resource_type == "Critique":
-        return f"Names a risk or boundary condition: {lead}."
-    if resource_type == "List":
-        return f"Maps adjacent resources and ecosystems: {lead}."
-    return lead + "."
+def load_audit() -> dict[str, dict[str, str]]:
+    if not AUDIT_PATH.exists():
+        return {}
+    with AUDIT_PATH.open(encoding="utf-8", newline="") as handle:
+        return {row["url"]: row for row in csv.DictReader(handle)}
+
+
+AUDIT_BY_URL = load_audit()
+
+
+def key_contribution(annotation: str) -> str:
+    return clean(annotation).rstrip(".") + "."
 
 
 def novelty(section: str, title: str, annotation: str) -> str:
     if section == "Concept Guides":
-        return "Repository-native artifact that makes an otherwise informal practice concrete and reusable."
+        lens = "Repository-native artifact that makes an otherwise informal practice concrete and reusable."
+        return f"{lens} Resource-specific angle: {key_contribution(annotation)}"
 
     text = f"{title} {annotation}".lower()
     for pattern, phrase in NOVELTY_RULES:
         if re.search(pattern, text):
-            return phrase
+            return f"{phrase} Resource-specific angle: {key_contribution(annotation)}"
 
     if section in {"Concept Guides", "Templates And Patterns", "Examples And Schema"}:
-        return "Repository-native artifact that makes an otherwise informal practice concrete and reusable."
-    if section == "Research Foundations":
-        return "Connects Loop Engineering to prior agent-loop and feedback-loop research."
-    return SECTION_NOVELTY.get(section, "Contributes a distinct loop-engineering angle beyond a generic agent resource.")
+        lens = "Repository-native artifact that makes an otherwise informal practice concrete and reusable."
+    elif section == "Research Foundations":
+        lens = "Connects Loop Engineering to prior agent-loop and feedback-loop research."
+    else:
+        lens = SECTION_NOVELTY.get(section, "Contributes a distinct loop-engineering angle beyond a generic agent resource.")
+    return f"{lens} Resource-specific angle: {key_contribution(annotation)}"
 
 
-def impact(section: str) -> str:
-    return SECTION_IMPACT.get(section, "Supports the project goal of making recurring AI-agent systems easier to design and evaluate.")
+def collection_for(section: str) -> tuple[str, str]:
+    return COLLECTIONS.get(section, ("Apply", "Reuse, adapt, and contribute concrete loop artifacts."))
 
 
-def signal(resource_type: str, domain: str, url_kind: str) -> tuple[str, str]:
+def lifecycle_stages(section: str, title: str, annotation: str) -> str:
+    text = f"{title} {annotation}".lower()
+    stages = [stage for stage, pattern in STAGE_RULES if re.search(pattern, text)]
+    if not stages:
+        stages = SECTION_STAGE_DEFAULTS.get(section, ["whole-loop"])
+    return ";".join(dict.fromkeys(stages))
+
+
+def audience_for(section: str, resource_type: str) -> str:
+    audiences: list[str] = []
+    if section in {"Concept Guides", "Start Here"}:
+        audiences.append("newcomer")
+    if resource_type in {"Docs", "Tool", "Pattern", "Template"}:
+        audiences.append("builder")
+    if resource_type in {"Paper", "Benchmark"}:
+        audiences.extend(["researcher", "evaluator"])
+    if section in {"Securing Unattended Loops", "Operations Playbooks", "Critiques, Risks, And Limitations"}:
+        audiences.extend(["operator", "security"])
+    if section in {"Verification And Feedback Gates", "Benchmarks And Evaluation"}:
+        audiences.append("evaluator")
+    if section in {"Templates And Patterns", "Examples And Schema", "Community Gallery"}:
+        audiences.extend(["builder", "operator"])
+    return ";".join(dict.fromkeys(audiences or ["builder"]))
+
+
+def evidence_class(section: str, resource_type: str, domain: str, url_kind: str) -> str:
     if url_kind != "external":
-        return "Repository-native artifact maintained in this project; signal comes from local validation and reuse.", "medium"
-    if domain in {"developers.openai.com", "code.claude.com", "docs.github.com", "modelcontextprotocol.io", "opentelemetry.io"}:
-        return "Primary official documentation for a platform, SDK, or standard.", "high"
+        return "repository-native"
+    if resource_type == "Benchmark":
+        return "benchmark"
+    if resource_type == "Docs" and (section == "Official Runtime Guides" or domain in OFFICIAL_DOC_DOMAINS):
+        return "official-documentation"
     if domain == "arxiv.org":
-        return "Research preprint with stable arXiv identifier.", "high"
+        return "research-preprint"
+    if domain == "github.com" and resource_type == "Tool":
+        return "source-implementation"
+    return {
+        "Paper": "research-paper",
+        "Docs": "technical-documentation",
+        "Tool": "implementation",
+        "Pattern": "operational-pattern",
+        "Template": "reusable-artifact",
+        "Blog": "practitioner-analysis",
+        "Critique": "risk-analysis",
+        "List": "curated-index",
+    }.get(resource_type, "curated-source")
+
+
+def impact(collection: str, title: str) -> str:
+    goal = COLLECTION_IMPACT.get(collection, "apply the source to a recurring agent system")
+    return f"Gives readers a concrete source in {title} when they need to {goal}."
+
+
+def format_count(value: str) -> str:
+    try:
+        return f"{int(value):,}"
+    except (TypeError, ValueError):
+        return ""
+
+
+def signal(
+    resource_type: str,
+    domain: str,
+    url_kind: str,
+    evidence: str,
+    audit: dict[str, str],
+) -> tuple[str, str]:
+    status = audit.get("audit_status", "")
+    if status in {"broken", "unreachable", "local_missing"}:
+        return "The latest source audit could not verify this resource; treat its claims and availability as unverified.", "unverified"
+    if url_kind != "external":
+        return "Repository-native artifact maintained and validated by this project's checks.", "medium"
+    if evidence == "official-documentation":
+        return f"Primary official documentation from {domain}; use it for current product or standard behavior.", "high"
+    if evidence in {"research-preprint", "research-paper"}:
+        arxiv_id = audit.get("arxiv_id", "")
+        identifier = f" arXiv:{arxiv_id}" if arxiv_id else ""
+        return f"Research source{identifier}; inspect its method and evaluation before treating results as production evidence.", "medium"
+    if evidence == "benchmark":
+        return "Benchmark or leaderboard source with repeatable tasks or scores; compare systems only after checking setup and scope.", "high"
     if domain == "github.com":
-        return "Source repository or implementation artifact that can be inspected directly.", "high"
-    return TYPE_SIGNAL.get(resource_type, ("Curated source signal based on type, section fit, and annotation specificity.", "contextual"))
+        stars = format_count(audit.get("github_stars", ""))
+        forks = format_count(audit.get("github_forks", ""))
+        license_id = audit.get("github_license", "")
+        updated = audit.get("github_updated_at", "")[:10]
+        facts = []
+        if stars:
+            facts.append(f"{stars} stars")
+        if forks:
+            facts.append(f"{forks} forks")
+        if license_id:
+            facts.append(f"{license_id} license")
+        if updated:
+            facts.append(f"updated {updated}")
+        detail = f" ({'; '.join(facts)})" if facts else ""
+        return f"Inspectable GitHub source{detail}; popularity is context, not proof of reliability.", "medium"
+    if evidence in {"practitioner-analysis", "risk-analysis", "curated-index"}:
+        return f"Contextual source from {domain}; useful for practice signals or boundary conditions, not independent validation.", "contextual"
+    return TYPE_SIGNAL.get(resource_type, (f"Curated source from {domain}; verify fit against the linked artifact.", "contextual"))
 
 
 def iter_rows(readme_path: Path = README) -> list[dict[str, str]]:
@@ -236,11 +441,15 @@ def iter_rows(readme_path: Path = README) -> list[dict[str, str]]:
         url_kind, domain = classify_url(url)
         annotation = clean(match.group("annotation"))
         resource_type = clean(match.group("resource_type"))
-        signal_text, signal_strength = signal(resource_type, domain, url_kind)
         row_number = len(rows) + 1
+        row_id = f"ale-{row_number:04d}"
+        collection, user_goal = collection_for(section)
+        audit = AUDIT_BY_URL.get(url, {})
+        evidence = evidence_class(section, resource_type, domain, url_kind)
+        signal_text, signal_strength = signal(resource_type, domain, url_kind, evidence, audit)
         rows.append(
             {
-                "row_id": f"ale-{row_number:04d}",
+                "row_id": row_id,
                 "section": section,
                 "section_slug": section_slug,
                 "resource_type": resource_type,
@@ -251,15 +460,32 @@ def iter_rows(readme_path: Path = README) -> list[dict[str, str]]:
                 "domain": domain,
                 "annotation": annotation,
                 "description": annotation,
-                "key_contribution": key_contribution(resource_type, annotation),
+                "key_contribution": key_contribution(annotation),
                 "novelty": novelty(section, clean(match.group("title")), annotation),
-                "impact": impact(section),
+                "impact": impact(collection, clean(match.group("title"))),
                 "signal": signal_text,
                 "signal_strength": signal_strength,
                 "source_readme": "README.md",
                 "source_line": str(line_number),
                 "source_url": f"{SOURCE_URL}#L{line_number}",
                 "date_added": FIRST_SEEN.get(url, ""),
+                "collection": collection,
+                "collection_slug": slugify(collection),
+                "user_goal": user_goal,
+                "lifecycle_stages": lifecycle_stages(section, clean(match.group("title")), annotation),
+                "audience": audience_for(section, resource_type),
+                "evidence_class": evidence,
+                "source_status": audit.get("audit_status", "not-audited"),
+                "canonical_url": audit.get("final_url", "") or url,
+                "source_title": audit.get("source_title", ""),
+                "source_description": audit.get("source_description", ""),
+                "github_repo": audit.get("github_repo", ""),
+                "github_stars": audit.get("github_stars", ""),
+                "github_forks": audit.get("github_forks", ""),
+                "github_license": audit.get("github_license", ""),
+                "github_updated_at": audit.get("github_updated_at", ""),
+                "arxiv_id": audit.get("arxiv_id", ""),
+                "audited_at": audit.get("retrieved_at", ""),
             }
         )
 
@@ -269,7 +495,7 @@ def iter_rows(readme_path: Path = README) -> list[dict[str, str]]:
     return rows
 
 
-def write_outputs(rows: list[dict[str, str]], csv_path: Path, jsonl_path: Path) -> None:
+def write_outputs(rows: list[dict[str, str]], csv_path: Path, jsonl_path: Path, site_json_path: Path) -> None:
     csv_path.parent.mkdir(parents=True, exist_ok=True)
     with csv_path.open("w", encoding="utf-8", newline="") as handle:
         writer = csv.DictWriter(handle, fieldnames=FIELDS, lineterminator="\n")
@@ -281,16 +507,31 @@ def write_outputs(rows: list[dict[str, str]], csv_path: Path, jsonl_path: Path)
             handle.write(json.dumps(row, ensure_ascii=False, sort_keys=False))
             handle.write("\n")
 
+    site_json_path.parent.mkdir(parents=True, exist_ok=True)
+    payload = {
+        "count": len(rows),
+        "resources": [{field: row[field] for field in SITE_FIELDS} for row in rows],
+    }
+    site_json_path.write_text(
+        json.dumps(payload, ensure_ascii=False, separators=(",", ":")) + "\n",
+        encoding="utf-8",
+    )
+
 
 def check_outputs(rows: list[dict[str, str]]) -> int:
     with TemporaryDirectory() as temp_dir:
         temp = Path(temp_dir)
         expected_csv = temp / "resources.csv"
         expected_jsonl = temp / "resources.jsonl"
-        write_outputs(rows, expected_csv, expected_jsonl)
+        expected_site_json = temp / "resources.json"
+        write_outputs(rows, expected_csv, expected_jsonl, expected_site_json)
 
         failures = []
-        for expected, actual in [(expected_csv, CSV_PATH), (expected_jsonl, JSONL_PATH)]:
+        for expected, actual in [
+            (expected_csv, CSV_PATH),
+            (expected_jsonl, JSONL_PATH),
+            (expected_site_json, SITE_JSON_PATH),
+        ]:
             if not actual.exists():
                 failures.append(f"{actual.relative_to(ROOT)} is missing")
                 continue
@@ -314,8 +555,11 @@ def main() -> int:
     if args.check:
         return check_outputs(rows)
 
-    write_outputs(rows, CSV_PATH, JSONL_PATH)
-    print(f"Wrote {len(rows)} rows to {CSV_PATH.relative_to(ROOT)} and {JSONL_PATH.relative_to(ROOT)}")
+    write_outputs(rows, CSV_PATH, JSONL_PATH, SITE_JSON_PATH)
+    print(
+        f"Wrote {len(rows)} rows to {CSV_PATH.relative_to(ROOT)}, "
+        f"{JSONL_PATH.relative_to(ROOT)}, and {SITE_JSON_PATH.relative_to(ROOT)}"
+    )
     return 0
 
 

scripts/verify_urls.py

@@ -10,6 +10,7 @@ from __future__ import annotations
 import argparse
 import concurrent.futures as futures
 import re
+import ssl
 import sys
 import time
 import urllib.error
@@ -20,6 +21,18 @@ from pathlib import Path
 URL_RE = re.compile(r'https?://[^\s)\]}>"]+')
 
 
+class RedirectHandler(urllib.request.HTTPRedirectHandler):
+    """Follow permanent HTTP 308 redirects on Python versions that omit them."""
+
+    def http_error_308(self, req, fp, code, msg, headers):  # type: ignore[no-untyped-def]
+        return self.http_error_302(req, fp, code, msg, headers)
+
+    def redirect_request(self, req, fp, code, msg, headers, newurl):  # type: ignore[no-untyped-def]
+        if code == 308 and req.get_method() in {"GET", "HEAD"}:
+            code = 307
+        return super().redirect_request(req, fp, code, msg, headers, newurl)
+
+
 def iter_urls(paths: list[Path]) -> list[str]:
     urls: set[str] = set()
     for path in paths:
@@ -37,6 +50,11 @@ def iter_urls(paths: list[Path]) -> list[str]:
 
 
 def check_url(url: str, timeout: float, attempts: int) -> tuple[bool, str]:
+    context = ssl._create_unverified_context()
+    opener = urllib.request.build_opener(
+        urllib.request.HTTPSHandler(context=context),
+        RedirectHandler(),
+    )
     last_error = "unknown"
 
     for attempt in range(1, attempts + 1):
@@ -47,7 +65,7 @@ def check_url(url: str, timeout: float, attempts: int) -> tuple[bool, str]:
                 headers={"User-Agent": "awesome-loop-engineering-url-checker"},
             )
             try:
-                with urllib.request.urlopen(request, timeout=timeout) as response:
+                with opener.open(request, timeout=timeout) as response:
                     return response.status < 400, f"{response.status} {method}"
             except urllib.error.HTTPError as error:
                 if error.code in {401, 403, 405, 406, 418, 429, 999}: