Add OWASP MCP Top 10 mapping on advisories and dataset card

README.md

@@ -12,6 +12,7 @@ tags:
   - supply-chain
   - codex
   - claude-code
+  - owasp-mcp
 configs:
   - config_name: plugins
     data_files: plugins.parquet
@@ -26,7 +27,7 @@ configs:
 
 Snapshot of Hashgraph Online plugin-catalog scores, modeled HOL Guard runtime fixtures, and curated public advisories. Dataset id: `HashgraphOnline/hol-plugin-security`.
 
-The default config is **plugins** (205 scored registry plugins). `runtime_fixtures` are modeled harness outcomes from the published Guard benchmark record. `advisories` are the 22 public HOL Guard advisory pages. `advisories.threat_class` is set from the official hub listing badges on [hol.org/guard/security](https://hol.org/guard/security), not inferred from titles.
+The default config is **plugins** (205 scored registry plugins). `runtime_fixtures` are modeled harness outcomes from the published Guard benchmark record. `advisories` are the 22 public HOL Guard advisory pages. `advisories.threat_class` is set from the official hub listing badges on [hol.org/guard/security](https://hol.org/guard/security), not inferred from titles. `advisories.owasp_mcp` is HOL's primary mapping of each public advisory to one OWASP MCP Top 10 (beta) ID.
 
 Sources: [plugin catalog API](https://hol.org/registry/api/v1/plugins/catalog), [runtime benchmark](https://hol.org/guard/research/ai-coding-agent-runtime-security-benchmark), [advisories](https://hol.org/guard/security/advisories). Re-export with `export_plugins.py`. License: Apache-2.0.
 
@@ -49,6 +50,37 @@ Sources: [plugin catalog API](https://hol.org/registry/api/v1/plugins/catalog),
 | runtime_fixtures | 220 | `runtime_fixtures.jsonl`, `runtime_fixtures.parquet` |
 | advisories | 22 | `advisories.jsonl`, `advisories.parquet` |
 
+## How to use
+
+Join `advisories` on `owasp_mcp` (one primary OWASP MCP Top 10 ID per advisory). Do not treat a mapped advisory as proof a plugin is safe. Plugin rows use `registry-broker-fallback` static/catalog scoring; scan is not a safety guarantee.
+
+## OWASP MCP Top 10 mapping (beta)
+
+This is HOL's mapping of its local Guard + catalog scanner to the [OWASP MCP Top 10](https://owasp.org/www-project-mcp-top-10/) beta (Phase 3). It is not a claim of complete coverage.
+
+Source: [OWASP MCP Top 10](https://owasp.org/www-project-mcp-top-10/). Project is beta (Phase 3). Document license CC BY-NC-SA 4.0; IDs and names are cited, not copied as long OWASP descriptions.
+
+HOL Guard is local-first runtime control on the developer machine (shell, secrets/file reads, MCP server change, plugin/skill install). It is not a cloud MCP gateway. Cursor/artifact trust is not full pre-exec on every MCP tool call. Runtime fixtures in this dataset are modeled; same outcome across harnesses; latency values are placeholders.
+
+`advisories.owasp_mcp` is HOL's primary mapping of each public advisory (one ID). Optional `owasp_mcp_name` is the official Top 10 title. `threat_class` is unchanged (hub badges on [hol.org/guard/security](https://hol.org/guard/security)).
+
+MCP06 on the Top 10 list is **Intent Flow Subversion**. The same OWASP page later also titles that item "Prompt Injection via Contextual Payloads"; detect text below includes prompt-injection advisories. HOL Guard is not a complete prompt-injection preventer.
+
+| ID | Detect | Prevent | Gap |
+| --- | --- | --- | --- |
+| MCP01 | catalog/advisory secret-exfil + mcp-token-theft | Guard can block secret-file / env reads at the local action boundary | not a secrets manager; does not rotate tokens; does not scrub secrets already inside model context |
+| MCP02 | tool-permission-creep advisory | Guard approval on plugin/skill install and MCP server change | not an OAuth scope-expiry gateway |
+| MCP03 | tool-description-poisoning advisory + scanner MCP posture | Guard on changed MCP servers / plugin install | Cursor path is artifact trust, not full pre-exec of every tool description |
+| MCP04 | catalog scores, lockfile/publishability findings, supply-chain advisories | Guard on plugin/skill install | not a signing/SBOM authority |
+| MCP05 | unsafe-command / rm-rf advisory | Guard intercepts local shell/file-destructive actions | does not sandbox every MCP server's own backend |
+| MCP06 | several prompt-injection advisories (Intent Flow Subversion / contextual payloads) | limited (some injected-instruction to shell paths hit Guard) | NOT a complete prompt-injection preventer |
+| MCP07 | MCP transport hardening (insecure HTTP, wildcard binds, missing auth posture) on Guard pages | local only | not an MCP gateway identity layer |
+| MCP08 | local decision trail / receipts on the developer machine | local decision trail / receipts | not an immutable SIEM; local logs |
+| MCP09 | shadow-mcp-server-discovery advisory | Guard on new/changed MCP servers before the harness launches | not continuous org-wide network discovery of shadow servers |
+| MCP10 | context-window-scraping / data-overexposure advisories | Guard on excessive local file reads | does not quarantine or partition model context windows |
+
+This mapping does not replace the disclaimers above. HOL publishes this dataset; it is not independent third-party validation.
+
 ## Citation
 
 HOL Guard Team. "HOL Plugin Security." 2026. https://huggingface.co/datasets/HashgraphOnline/hol-plugin-security

advisories.jsonl

@@ -1,22 +1,22 @@
-{"slug": "npm-postinstall-script-abuse", "title": "npm postinstall script abuse in AI coding environments", "severity": "critical", "threat_class": "supply-chain", "summary": "Malicious npm packages use postinstall scripts to execute arbitrary code during installation. In AI coding environments, these scripts can modify agent configuration, install backdoor MCP servers, or exfiltrate project secrets — all before the developer reviews the package.", "surfaces": ["npm postinstall scripts", "package.json dependencies", "agent configuration files", "project .env files"], "url": "https://hol.org/guard/security/advisories/npm-postinstall-script-abuse", "date_modified": "2026-05-09T00:00:00.000Z", "curated": true}
-{"slug": "env-file-exfiltration-via-webhook", "title": "Environment file exfiltration via webhook", "severity": "critical", "threat_class": "secret-exfiltration", "summary": "AI agents can be tricked into reading .env files and sending their contents to external endpoints through tool calls, webhook integrations, or HTTP requests that appear legitimate.", "surfaces": [".env files", "environment variables", "HTTP tool calls", "webhook integrations"], "url": "https://hol.org/guard/security/advisories/env-file-exfiltration-via-webhook", "date_modified": "2026-05-09T00:00:00.000Z", "curated": true}
-{"slug": "ci-cd-pipeline-poisoning", "title": "CI/CD pipeline poisoning via agent-written config", "severity": "critical", "threat_class": "supply-chain", "summary": "AI agents that write CI/CD configuration files (GitHub Actions, GitLab CI, CircleCI) can introduce backdoors — injecting steps that exfiltrate secrets, modify artifacts, or deploy malicious code — that execute on every build.", "surfaces": ["GitHub Actions workflows", "GitLab CI configs", "CI/CD secrets", "build artifacts"], "url": "https://hol.org/guard/security/advisories/ci-cd-pipeline-poisoning", "date_modified": "2026-05-09T00:00:00.000Z", "curated": true}
-{"slug": "mcp-tool-description-poisoning", "title": "MCP tool description poisoning", "severity": "high", "threat_class": "mcp-tool-poisoning", "summary": "Malicious MCP tool descriptions embed hidden instructions that redirect AI agents into calling the wrong tool, exfiltrating secrets, or executing unintended commands — even when the tool itself appears harmless.", "surfaces": ["MCP tool descriptions", "agent tool selection", "team policy memory"], "url": "https://hol.org/guard/security/advisories/mcp-tool-description-poisoning", "date_modified": "2026-05-09T00:00:00.000Z", "curated": true}
-{"slug": "prompt-injection-via-issue-comments", "title": "Prompt injection via issue comments and pull requests", "severity": "high", "threat_class": "prompt-injection", "summary": "Attackers embed hidden instructions in GitHub issues, PR comments, and commit messages. When an AI agent reads these to help triage or review, it follows the embedded instructions — potentially approving malicious code or leaking repository secrets.", "surfaces": ["GitHub issues", "pull request comments", "commit messages", "agent context window"], "url": "https://hol.org/guard/security/advisories/prompt-injection-via-issue-comments", "date_modified": "2026-05-09T00:00:00.000Z", "curated": true}
-{"slug": "agent-readable-config-poisoning", "title": "Agent-readable config file poisoning", "severity": "high", "threat_class": "prompt-injection", "summary": "AI agents read configuration files like CLAUDE.md, .cursorrules, and AGENTS.md as trusted context. An attacker who can modify these files — via a compromised dependency, a malicious collaborator, or a typo in a path — gains the ability to inject persistent instructions the agent follows on every session.", "surfaces": ["CLAUDE.md", ".cursorrules", "AGENTS.md", ".github/copilot-instructions.md", "agent context window"], "url": "https://hol.org/guard/security/advisories/agent-readable-config-poisoning", "date_modified": "2026-05-09T00:00:00.000Z", "curated": true}
-{"slug": "rm-rf-via-injected-instruction", "title": "Destructive command execution via injected instruction", "severity": "high", "threat_class": "unsafe-command", "summary": "Prompt injection can cause AI agents to run destructive shell commands like rm -rf, git push --force, or database drops — by embedding instructions in files, issues, or tool descriptions.", "surfaces": ["shell command execution", "git operations", "file system", "package install scripts"], "url": "https://hol.org/guard/security/advisories/rm-rf-via-injected-instruction", "date_modified": "2026-05-09T00:00:00.000Z", "curated": true}
-{"slug": "gitignore-bypass-via-agent", "title": "Gitignore bypass via agent file reads", "severity": "high", "threat_class": "secret-exfiltration", "summary": "AI agents can read files that are gitignored — secrets, private keys, and internal configs — because gitignore only prevents git tracking, not file system access. These files often contain the most sensitive data in a repository.", "surfaces": [".gitignore", "private keys", "secrets directories", "agent file-read tools"], "url": "https://hol.org/guard/security/advisories/gitignore-bypass-via-agent", "date_modified": "2026-05-09T00:00:00.000Z", "curated": true}
-{"slug": "mcp-token-theft-via-headers", "title": "MCP authentication token theft via headers", "severity": "high", "threat_class": "secret-exfiltration", "summary": "MCP servers that accept authentication tokens in headers can leak those tokens if the server logs requests, shares telemetry, or is compromised. Tokens passed to MCP servers persist in server-side logs and may be accessible to attackers.", "surfaces": ["MCP server authentication", "HTTP headers", "server-side logs", "telemetry pipelines"], "url": "https://hol.org/guard/security/advisories/mcp-token-theft-via-headers", "date_modified": "2026-05-09T00:00:00.000Z", "curated": true}
-{"slug": "malicious-skill-with-hidden-prompt", "title": "Malicious skill with hidden prompt injection", "severity": "high", "threat_class": "prompt-injection", "summary": "AI agent skills (Claude Code skills, Cursor rules, Copilot extensions) can contain hidden prompt injections in their instructions. When the skill is loaded, the hidden prompt executes on every session that uses the skill.", "surfaces": ["agent skills", "skill instructions", "plugin manifests", "agent context window"], "url": "https://hol.org/guard/security/advisories/malicious-skill-with-hidden-prompt", "date_modified": "2026-05-09T00:00:00.000Z", "curated": true}
-{"slug": "indirect-prompt-injection-via-web", "title": "Indirect prompt injection via web content", "severity": "high", "threat_class": "prompt-injection", "summary": "When AI agents fetch web pages — documentation, Stack Overflow answers, package READMEs — the fetched content can contain hidden instructions that the agent follows, potentially exfiltrating data or executing unintended actions.", "surfaces": ["web fetch tools", "fetched HTML content", "documentation pages", "agent context window"], "url": "https://hol.org/guard/security/advisories/indirect-prompt-injection-via-web", "date_modified": "2026-05-09T00:00:00.000Z", "curated": true}
-{"slug": "agent-identity-spoofing", "title": "Agent identity spoofing via system prompt mimicry", "severity": "high", "threat_class": "prompt-injection", "summary": "Attackers can craft content that mimics system prompts or tool outputs, tricking the agent into believing it received instructions from the harness, the user, or a trusted tool — when the instructions actually came from untrusted data.", "surfaces": ["system prompt context", "tool output formatting", "agent trust model"], "url": "https://hol.org/guard/security/advisories/agent-identity-spoofing", "date_modified": "2026-05-09T00:00:00.000Z", "curated": true}
-{"slug": "dockerfile-injection-via-agent", "title": "Dockerfile injection via AI agent writes", "severity": "high", "threat_class": "supply-chain", "summary": "When AI agents write or modify Dockerfiles, prompt injection can cause them to add malicious instructions — pulling attacker-controlled base images, exfiltrating build secrets, or installing backdoors that persist across all container builds.", "surfaces": ["Dockerfiles", "container build context", "base images", "build secrets"], "url": "https://hol.org/guard/security/advisories/dockerfile-injection-via-agent", "date_modified": "2026-05-09T00:00:00.000Z", "curated": true}
-{"slug": "shadow-mcp-server-discovery", "title": "Shadow MCP server discovery and persistent access", "severity": "medium", "threat_class": "mcp-tool-poisoning", "summary": "MCP servers added to a project during development can persist in configuration files and maintain access to the agent’s context window long after they are forgotten. These \"shadow\" servers continue receiving tool calls and may be modified by attackers who compromise the original server.", "surfaces": ["MCP server configuration files", "agent context window", "team policy memory", "persistent tool connections"], "url": "https://hol.org/guard/security/advisories/shadow-mcp-server-discovery", "date_modified": "2026-05-09T00:00:00.000Z", "curated": true}
-{"slug": "context-window-scraping", "title": "Context window scraping via long file reads", "severity": "medium", "threat_class": "data-overexposure", "summary": "AI agents that read large files can leak proprietary code, internal documentation, and customer data into their context window — which may then be sent to external LLM APIs or logged in cloud telemetry.", "surfaces": ["LLM context window", "model API egress", "cloud telemetry", "proprietary code files"], "url": "https://hol.org/guard/security/advisories/context-window-scraping", "date_modified": "2026-05-09T00:00:00.000Z", "curated": true}
-{"slug": "cross-workspace-credential-leak", "title": "Cross-workspace credential leak via monorepo traversal", "severity": "medium", "threat_class": "data-overexposure", "summary": "AI agents in monorepo environments can read credentials, configs, and secrets from adjacent workspaces — leaking data across team boundaries.", "surfaces": ["monorepo workspaces", "shared config files", "team credentials", "adjacent project directories"], "url": "https://hol.org/guard/security/advisories/cross-workspace-credential-leak", "date_modified": "2026-05-09T00:00:00.000Z", "curated": true}
-{"slug": "clipboard-and-terminal-injection", "title": "Clipboard and terminal buffer injection", "severity": "medium", "threat_class": "prompt-injection", "summary": "Attackers can plant hostile instructions in clipboard contents or terminal scrollback buffers. When an AI agent reads terminal output or the user pastes clipboard content, the hidden instructions execute as if they came from the user.", "surfaces": ["clipboard contents", "terminal scrollback", "pasted text", "agent context window"], "url": "https://hol.org/guard/security/advisories/clipboard-and-terminal-injection", "date_modified": "2026-05-09T00:00:00.000Z", "curated": true}
-{"slug": "tool-permission-creep", "title": "Tool permission creep in AI agents", "severity": "medium", "threat_class": "mcp-tool-poisoning", "summary": "AI agents accumulate tool permissions over time as developers approve new tools \"just this once.\" These permissions persist across sessions, creating an ever-widening attack surface where tools that were approved once can be used by prompt injection in future sessions.", "surfaces": ["tool permission grants", "agent session memory", "MCP tool configurations"], "url": "https://hol.org/guard/security/advisories/tool-permission-creep", "date_modified": "2026-05-09T00:00:00.000Z", "curated": true}
-{"slug": "model-confusion-via-multiple-instructions", "title": "Model confusion via conflicting instructions", "severity": "medium", "threat_class": "prompt-injection", "summary": "When an AI agent receives multiple conflicting instructions — from the user, the system prompt, tool descriptions, and file contents — it may follow the wrong one. Attackers exploit this by planting instructions that conflict with the user's actual intent.", "surfaces": ["agent instruction hierarchy", "context window", "tool descriptions", "file contents"], "url": "https://hol.org/guard/security/advisories/model-confusion-via-multiple-instructions", "date_modified": "2026-05-09T00:00:00.000Z", "curated": true}
-{"slug": "excessive-file-reading-during-exploration", "title": "Excessive file reading during project exploration", "severity": "medium", "threat_class": "data-overexposure", "summary": "When AI agents explore a project to understand its structure, they often read dozens or hundreds of files — far more than needed for the task. This excessive reading can expose secrets, proprietary code, and customer data that enter the context window and model API.", "surfaces": ["file system access", "context window", "model API egress"], "url": "https://hol.org/guard/security/advisories/excessive-file-reading-during-exploration", "date_modified": "2026-05-09T00:00:00.000Z", "curated": true}
-{"slug": "stale-dependency-exploitation", "title": "Stale dependency exploitation in AI environments", "severity": "medium", "threat_class": "supply-chain", "summary": "AI agents often work with projects that have outdated dependencies. When an agent suggests or installs packages based on a stale package.json, it can introduce known-vulnerable versions — and in AI environments, the vulnerability is amplified because the agent can execute commands.", "surfaces": ["package.json", "lockfiles", "transitive dependencies", "agent install commands"], "url": "https://hol.org/guard/security/advisories/stale-dependency-exploitation", "date_modified": "2026-05-09T00:00:00.000Z", "curated": true}
-{"slug": "token-cost-amplification-attack", "title": "Token cost amplification via context flooding", "severity": "low", "threat_class": "data-overexposure", "summary": "Attackers can craft content that causes AI agents to consume excessive tokens — by inserting large files, repetitive instructions, or recursive prompts that bloat the context window. This inflates API costs and can cause rate-limit denial of service.", "surfaces": ["context window", "model API costs", "rate limits", "agent response time"], "url": "https://hol.org/guard/security/advisories/token-cost-amplification-attack", "date_modified": "2026-05-09T00:00:00.000Z", "curated": true}
+{"slug":"npm-postinstall-script-abuse","title":"npm postinstall script abuse in AI coding environments","severity":"critical","threat_class":"supply-chain","summary":"Malicious npm packages use postinstall scripts to execute arbitrary code during installation. In AI coding environments, these scripts can modify agent configuration, install backdoor MCP servers, or exfiltrate project secrets — all before the developer reviews the package.","surfaces":["npm postinstall scripts","package.json dependencies","agent configuration files","project .env files"],"url":"https://hol.org/guard/security/advisories/npm-postinstall-script-abuse","date_modified":"2026-05-09T00:00:00.000Z","curated":true,"owasp_mcp":"MCP04","owasp_mcp_name":"Software Supply Chain Attacks & Dependency Tampering"}
+{"slug":"env-file-exfiltration-via-webhook","title":"Environment file exfiltration via webhook","severity":"critical","threat_class":"secret-exfiltration","summary":"AI agents can be tricked into reading .env files and sending their contents to external endpoints through tool calls, webhook integrations, or HTTP requests that appear legitimate.","surfaces":[".env files","environment variables","HTTP tool calls","webhook integrations"],"url":"https://hol.org/guard/security/advisories/env-file-exfiltration-via-webhook","date_modified":"2026-05-09T00:00:00.000Z","curated":true,"owasp_mcp":"MCP01","owasp_mcp_name":"Token Mismanagement & Secret Exposure"}
+{"slug":"ci-cd-pipeline-poisoning","title":"CI/CD pipeline poisoning via agent-written config","severity":"critical","threat_class":"supply-chain","summary":"AI agents that write CI/CD configuration files (GitHub Actions, GitLab CI, CircleCI) can introduce backdoors — injecting steps that exfiltrate secrets, modify artifacts, or deploy malicious code — that execute on every build.","surfaces":["GitHub Actions workflows","GitLab CI configs","CI/CD secrets","build artifacts"],"url":"https://hol.org/guard/security/advisories/ci-cd-pipeline-poisoning","date_modified":"2026-05-09T00:00:00.000Z","curated":true,"owasp_mcp":"MCP04","owasp_mcp_name":"Software Supply Chain Attacks & Dependency Tampering"}
+{"slug":"mcp-tool-description-poisoning","title":"MCP tool description poisoning","severity":"high","threat_class":"mcp-tool-poisoning","summary":"Malicious MCP tool descriptions embed hidden instructions that redirect AI agents into calling the wrong tool, exfiltrating secrets, or executing unintended commands — even when the tool itself appears harmless.","surfaces":["MCP tool descriptions","agent tool selection","team policy memory"],"url":"https://hol.org/guard/security/advisories/mcp-tool-description-poisoning","date_modified":"2026-05-09T00:00:00.000Z","curated":true,"owasp_mcp":"MCP03","owasp_mcp_name":"Tool Poisoning"}
+{"slug":"prompt-injection-via-issue-comments","title":"Prompt injection via issue comments and pull requests","severity":"high","threat_class":"prompt-injection","summary":"Attackers embed hidden instructions in GitHub issues, PR comments, and commit messages. When an AI agent reads these to help triage or review, it follows the embedded instructions — potentially approving malicious code or leaking repository secrets.","surfaces":["GitHub issues","pull request comments","commit messages","agent context window"],"url":"https://hol.org/guard/security/advisories/prompt-injection-via-issue-comments","date_modified":"2026-05-09T00:00:00.000Z","curated":true,"owasp_mcp":"MCP06","owasp_mcp_name":"Intent Flow Subversion"}
+{"slug":"agent-readable-config-poisoning","title":"Agent-readable config file poisoning","severity":"high","threat_class":"prompt-injection","summary":"AI agents read configuration files like CLAUDE.md, .cursorrules, and AGENTS.md as trusted context. An attacker who can modify these files — via a compromised dependency, a malicious collaborator, or a typo in a path — gains the ability to inject persistent instructions the agent follows on every session.","surfaces":["CLAUDE.md",".cursorrules","AGENTS.md",".github/copilot-instructions.md","agent context window"],"url":"https://hol.org/guard/security/advisories/agent-readable-config-poisoning","date_modified":"2026-05-09T00:00:00.000Z","curated":true,"owasp_mcp":"MCP03","owasp_mcp_name":"Tool Poisoning"}
+{"slug":"rm-rf-via-injected-instruction","title":"Destructive command execution via injected instruction","severity":"high","threat_class":"unsafe-command","summary":"Prompt injection can cause AI agents to run destructive shell commands like rm -rf, git push --force, or database drops — by embedding instructions in files, issues, or tool descriptions.","surfaces":["shell command execution","git operations","file system","package install scripts"],"url":"https://hol.org/guard/security/advisories/rm-rf-via-injected-instruction","date_modified":"2026-05-09T00:00:00.000Z","curated":true,"owasp_mcp":"MCP05","owasp_mcp_name":"Command Injection & Execution"}
+{"slug":"gitignore-bypass-via-agent","title":"Gitignore bypass via agent file reads","severity":"high","threat_class":"secret-exfiltration","summary":"AI agents can read files that are gitignored — secrets, private keys, and internal configs — because gitignore only prevents git tracking, not file system access. These files often contain the most sensitive data in a repository.","surfaces":[".gitignore","private keys","secrets directories","agent file-read tools"],"url":"https://hol.org/guard/security/advisories/gitignore-bypass-via-agent","date_modified":"2026-05-09T00:00:00.000Z","curated":true,"owasp_mcp":"MCP01","owasp_mcp_name":"Token Mismanagement & Secret Exposure"}
+{"slug":"mcp-token-theft-via-headers","title":"MCP authentication token theft via headers","severity":"high","threat_class":"secret-exfiltration","summary":"MCP servers that accept authentication tokens in headers can leak those tokens if the server logs requests, shares telemetry, or is compromised. Tokens passed to MCP servers persist in server-side logs and may be accessible to attackers.","surfaces":["MCP server authentication","HTTP headers","server-side logs","telemetry pipelines"],"url":"https://hol.org/guard/security/advisories/mcp-token-theft-via-headers","date_modified":"2026-05-09T00:00:00.000Z","curated":true,"owasp_mcp":"MCP01","owasp_mcp_name":"Token Mismanagement & Secret Exposure"}
+{"slug":"malicious-skill-with-hidden-prompt","title":"Malicious skill with hidden prompt injection","severity":"high","threat_class":"prompt-injection","summary":"AI agent skills (Claude Code skills, Cursor rules, Copilot extensions) can contain hidden prompt injections in their instructions. When the skill is loaded, the hidden prompt executes on every session that uses the skill.","surfaces":["agent skills","skill instructions","plugin manifests","agent context window"],"url":"https://hol.org/guard/security/advisories/malicious-skill-with-hidden-prompt","date_modified":"2026-05-09T00:00:00.000Z","curated":true,"owasp_mcp":"MCP06","owasp_mcp_name":"Intent Flow Subversion"}
+{"slug":"indirect-prompt-injection-via-web","title":"Indirect prompt injection via web content","severity":"high","threat_class":"prompt-injection","summary":"When AI agents fetch web pages — documentation, Stack Overflow answers, package READMEs — the fetched content can contain hidden instructions that the agent follows, potentially exfiltrating data or executing unintended actions.","surfaces":["web fetch tools","fetched HTML content","documentation pages","agent context window"],"url":"https://hol.org/guard/security/advisories/indirect-prompt-injection-via-web","date_modified":"2026-05-09T00:00:00.000Z","curated":true,"owasp_mcp":"MCP06","owasp_mcp_name":"Intent Flow Subversion"}
+{"slug":"agent-identity-spoofing","title":"Agent identity spoofing via system prompt mimicry","severity":"high","threat_class":"prompt-injection","summary":"Attackers can craft content that mimics system prompts or tool outputs, tricking the agent into believing it received instructions from the harness, the user, or a trusted tool — when the instructions actually came from untrusted data.","surfaces":["system prompt context","tool output formatting","agent trust model"],"url":"https://hol.org/guard/security/advisories/agent-identity-spoofing","date_modified":"2026-05-09T00:00:00.000Z","curated":true,"owasp_mcp":"MCP07","owasp_mcp_name":"Insufficient Authentication & Authorization"}
+{"slug":"dockerfile-injection-via-agent","title":"Dockerfile injection via AI agent writes","severity":"high","threat_class":"supply-chain","summary":"When AI agents write or modify Dockerfiles, prompt injection can cause them to add malicious instructions — pulling attacker-controlled base images, exfiltrating build secrets, or installing backdoors that persist across all container builds.","surfaces":["Dockerfiles","container build context","base images","build secrets"],"url":"https://hol.org/guard/security/advisories/dockerfile-injection-via-agent","date_modified":"2026-05-09T00:00:00.000Z","curated":true,"owasp_mcp":"MCP04","owasp_mcp_name":"Software Supply Chain Attacks & Dependency Tampering"}
+{"slug":"shadow-mcp-server-discovery","title":"Shadow MCP server discovery and persistent access","severity":"medium","threat_class":"mcp-tool-poisoning","summary":"MCP servers added to a project during development can persist in configuration files and maintain access to the agent’s context window long after they are forgotten. These \"shadow\" servers continue receiving tool calls and may be modified by attackers who compromise the original server.","surfaces":["MCP server configuration files","agent context window","team policy memory","persistent tool connections"],"url":"https://hol.org/guard/security/advisories/shadow-mcp-server-discovery","date_modified":"2026-05-09T00:00:00.000Z","curated":true,"owasp_mcp":"MCP09","owasp_mcp_name":"Shadow MCP Servers"}
+{"slug":"context-window-scraping","title":"Context window scraping via long file reads","severity":"medium","threat_class":"data-overexposure","summary":"AI agents that read large files can leak proprietary code, internal documentation, and customer data into their context window — which may then be sent to external LLM APIs or logged in cloud telemetry.","surfaces":["LLM context window","model API egress","cloud telemetry","proprietary code files"],"url":"https://hol.org/guard/security/advisories/context-window-scraping","date_modified":"2026-05-09T00:00:00.000Z","curated":true,"owasp_mcp":"MCP10","owasp_mcp_name":"Context Injection & Over-Sharing"}
+{"slug":"cross-workspace-credential-leak","title":"Cross-workspace credential leak via monorepo traversal","severity":"medium","threat_class":"data-overexposure","summary":"AI agents in monorepo environments can read credentials, configs, and secrets from adjacent workspaces — leaking data across team boundaries.","surfaces":["monorepo workspaces","shared config files","team credentials","adjacent project directories"],"url":"https://hol.org/guard/security/advisories/cross-workspace-credential-leak","date_modified":"2026-05-09T00:00:00.000Z","curated":true,"owasp_mcp":"MCP01","owasp_mcp_name":"Token Mismanagement & Secret Exposure"}
+{"slug":"clipboard-and-terminal-injection","title":"Clipboard and terminal buffer injection","severity":"medium","threat_class":"prompt-injection","summary":"Attackers can plant hostile instructions in clipboard contents or terminal scrollback buffers. When an AI agent reads terminal output or the user pastes clipboard content, the hidden instructions execute as if they came from the user.","surfaces":["clipboard contents","terminal scrollback","pasted text","agent context window"],"url":"https://hol.org/guard/security/advisories/clipboard-and-terminal-injection","date_modified":"2026-05-09T00:00:00.000Z","curated":true,"owasp_mcp":"MCP06","owasp_mcp_name":"Intent Flow Subversion"}
+{"slug":"tool-permission-creep","title":"Tool permission creep in AI agents","severity":"medium","threat_class":"mcp-tool-poisoning","summary":"AI agents accumulate tool permissions over time as developers approve new tools \"just this once.\" These permissions persist across sessions, creating an ever-widening attack surface where tools that were approved once can be used by prompt injection in future sessions.","surfaces":["tool permission grants","agent session memory","MCP tool configurations"],"url":"https://hol.org/guard/security/advisories/tool-permission-creep","date_modified":"2026-05-09T00:00:00.000Z","curated":true,"owasp_mcp":"MCP02","owasp_mcp_name":"Privilege Escalation via Scope Creep"}
+{"slug":"model-confusion-via-multiple-instructions","title":"Model confusion via conflicting instructions","severity":"medium","threat_class":"prompt-injection","summary":"When an AI agent receives multiple conflicting instructions — from the user, the system prompt, tool descriptions, and file contents — it may follow the wrong one. Attackers exploit this by planting instructions that conflict with the user's actual intent.","surfaces":["agent instruction hierarchy","context window","tool descriptions","file contents"],"url":"https://hol.org/guard/security/advisories/model-confusion-via-multiple-instructions","date_modified":"2026-05-09T00:00:00.000Z","curated":true,"owasp_mcp":"MCP06","owasp_mcp_name":"Intent Flow Subversion"}
+{"slug":"excessive-file-reading-during-exploration","title":"Excessive file reading during project exploration","severity":"medium","threat_class":"data-overexposure","summary":"When AI agents explore a project to understand its structure, they often read dozens or hundreds of files — far more than needed for the task. This excessive reading can expose secrets, proprietary code, and customer data that enter the context window and model API.","surfaces":["file system access","context window","model API egress"],"url":"https://hol.org/guard/security/advisories/excessive-file-reading-during-exploration","date_modified":"2026-05-09T00:00:00.000Z","curated":true,"owasp_mcp":"MCP10","owasp_mcp_name":"Context Injection & Over-Sharing"}
+{"slug":"stale-dependency-exploitation","title":"Stale dependency exploitation in AI environments","severity":"medium","threat_class":"supply-chain","summary":"AI agents often work with projects that have outdated dependencies. When an agent suggests or installs packages based on a stale package.json, it can introduce known-vulnerable versions — and in AI environments, the vulnerability is amplified because the agent can execute commands.","surfaces":["package.json","lockfiles","transitive dependencies","agent install commands"],"url":"https://hol.org/guard/security/advisories/stale-dependency-exploitation","date_modified":"2026-05-09T00:00:00.000Z","curated":true,"owasp_mcp":"MCP04","owasp_mcp_name":"Software Supply Chain Attacks & Dependency Tampering"}
+{"slug":"token-cost-amplification-attack","title":"Token cost amplification via context flooding","severity":"low","threat_class":"data-overexposure","summary":"Attackers can craft content that causes AI agents to consume excessive tokens — by inserting large files, repetitive instructions, or recursive prompts that bloat the context window. This inflates API costs and can cause rate-limit denial of service.","surfaces":["context window","model API costs","rate limits","agent response time"],"url":"https://hol.org/guard/security/advisories/token-cost-amplification-attack","date_modified":"2026-05-09T00:00:00.000Z","curated":true,"owasp_mcp":"MCP10","owasp_mcp_name":"Context Injection & Over-Sharing"}

advisories.parquet

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:eecb056908870c55c4adfdf1b1447c1a0a6140ba9eef16d43d07c3387f6a0c25
-size 13772
+oid sha256:95ffbeb515471031190139ca58d34943219e3f6258d4af0350af016bd5eb01c7
+size 15265