review-plan

Pass

Audited by Gen Agent Trust Hub on Sep 10, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute standard development tools including git rev-parse, git hash-object, and sha256sum for version tracking and integrity verification.\n- [COMMAND_EXECUTION]: It executes a local script bun scripts/pre-execution-snapshot.mjs to handle snapshot building and verification, which is core to the skill's review logic.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted project artifacts which presents an indirect prompt injection attack surface.\n
  • Ingestion points: Reads files such as SPEC.md, PLAN.md, TASKS.md, and ACCEPTANCE.md (detailed in SKILL.md Step 0).\n
  • Boundary markers: SKILL.md contains explicit instructions to treat all artifact content as data only and to report any embedded instructions as findings rather than executing them.\n
  • Capability inventory: The agent can run shell commands (git, bun) and write to files (progress.md, planning-findings.md).\n
  • Sanitization: Instructions mandate that directives found in reviewable artifacts be flagged as defects.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 10, 2026, 10:04 AM
Security Audit — agent-trust-hub — review-plan