spec
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to use the
git hash-objectcommand to generate blob hashes for delta documents andprintfto append entries to a local log file. These are standard operations for the skill's file-tracking and logging functionality. - [INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it ingests untrusted or third-party project data to produce generated documents.
- Ingestion points: Reads from
log.md,seed.md,AGENTS.md, and project decision records (docs/adr/,docs/pdr/). - Boundary markers: The skill defines specific document formats and mandates an 'audit pass' where the agent compares the draft against the conversation history.
- Capability inventory: The skill writes multiple files to the repository (
spec.md, delta documents, and pending-decision bundles) and modifies the thread'slog.mdfile. - Sanitization: Content is primarily handled through Markdown formatting and LLM-driven auditing rather than explicit programmatic sanitization filters.
Audit Metadata