plan-prompts
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external source documents (untrusted data) to generate development plans and execution prompts, which creates a vulnerability surface for instructions embedded in those documents.\n
- Ingestion points: Phase 0 specifies that the skill ingests documents from a user-provided folder path or an explicit file list.\n
- Boundary markers: The skill lacks explicit instructions to use delimiters or 'ignore embedded instructions' warnings for the input source documents, although it uses specific markers like '> ASSUMPTION:' and '> GAP:' in its own output.\n
- Capability inventory: The skill's workflow involves reading local documents, writing multiple planning files to the '.docs' directory, executing 'git check-ignore', and potentially creating or modifying a '.gitignore' file.\n
- Sanitization: The instructions require identifying contradictions and gaps in requirements but do not provide for escaping or filtering of external content interpolated into the prompts.\n- [COMMAND_EXECUTION]: The skill uses basic shell commands for environment discovery.\n
- Evidence: The Phase 1 workflow specifies using the command 'git check-ignore' to verify that the history ledger file is correctly ignored by the repository's configuration.
Audit Metadata