source-command-cowork-update-releases

Pass

Audited by Gen Agent Trust Hub on Aug 26, 2026

Risk Level: SAFEREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests data from a local CHANGELOG.md (Step 3) and external WebSearch results (Step 4) to generate content.
  • Ingestion points: /Users/florianbruniaux/Sites/perso/Codex-cowork-guide/CHANGELOG.md and Anthropic announcement search results.
  • Boundary markers: No explicit delimiters are defined to separate the source content from the agent's instructions during the parsing process.
  • Capability inventory: The skill has the capability to write directly to /Users/florianbruniaux/Sites/perso/Codex-cowork-guide-landing/src/data/releases.ts and /Users/florianbruniaux/Sites/perso/Codex-cowork-guide/machine-readable/Codex-cowork-releases.yaml.
  • Sanitization: The skill performs heuristic-based condensing and HTML formatting but does not mention specific validation or escaping of characters that could break the TypeScript structure.
  • [REMOTE_CODE_EXECUTION]: The skill generates a TypeScript source file (releases.ts) using data derived from external or local text files (Step 7). As this file is intended to be used in a build process (pnpm build mentioned in Step 9), maliciously crafted input in the source documents could lead to code injection in the final application.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 26, 2026, 03:29 AM
Security Audit — agent-trust-hub — source-command-cowork-update-releases