planning
Pass
Audited by Gen Agent Trust Hub on Nov 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted content from the user's repository, including source code, documentation (ADRs, RFCs), and Git logs, to generate implementation plans. This creates a surface where malicious instructions hidden in code comments or documentation could influence agent behavior.
- Ingestion points: Ingestion of repository content occurs as specified in
SKILL.md(Workflow step 2),references/doc-grounding.md(Where to find docs), andreferences/claim-verification.md(Workflow step 2). - Boundary markers: The skill does not define specific delimiters or warnings to ignore instructions embedded in the processed repository files.
- Capability inventory: The skill utilizes file system reads (
grep,find, file reading), shell command execution (wc,pnpm test,pytest,tsc), and network operations (curl) as described inreferences/claim-verification.md. - Sanitization: There are no instructions for sanitizing or escaping the content read from the repository before it is incorporated into the planning process or prompt.
- [COMMAND_EXECUTION]: The skill instructs the agent to execute various shell commands to verify claims made in implementation plans.
- Evidence:
references/claim-verification.mdlists commands such asgrep -r,find,wc -l,git log,pnpm test,curl, andtsc --noEmitas tools for evidence collection. While these are standard developer tools, they represent a capability that acts on the environment based on potentially untrusted plan claims.
Audit Metadata