prd-quality-gate
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONREMOTE_CODE_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted input from external issue bodies and comments, which can be manipulated by attackers to influence agent behavior. \n
- Ingestion points: The skill reads "Live issue body/comments" (SKILL.md) and "current issue body" (references/execution-readiness.md) to perform validation and hash computation.\n
- Boundary markers: While the skill uses Markdown templates for structure, it lacks explicit safety directives or delimiters to prevent the agent from obeying instructions embedded within the ingested issue content.\n
- Capability inventory: The skill has the capability to inspect the repository, perform git operations, and execute local scripts (plan-header.mjs) based on the processed state.\n
- Sanitization: No sanitization of the input text from issues is specified before it is parsed and processed.\n- [REMOTE_CODE_EXECUTION]: The skill depends on a script from an external skill package to perform its primary validation tasks. \n
- Evidence: references/execution-readiness.md instructs the agent to execute "executing-plans skill's scripts/plan-header.mjs". This execution involves running a script external to the skill's own directory to compute digests and verify plan headers.\n- [COMMAND_EXECUTION]: The skill utilizes shell commands to manage repository state and validate implementation readiness. \n
- Evidence: The instructions describe using "git fetch", "git rev-parse origin/", and direct execution of "plan-header.mjs" to check for drift and validate plan integrity.
Audit Metadata