pastlight-agents-pointer
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's core purpose is to inject a 'Canonical snippet' into project-level agent instruction files like AGENTS.md or CLAUDE.md. This snippet provides instructions that will be followed by agents in future sessions.
- Ingestion points: The agent reads existing repo instruction files (AGENTS.md, CLAUDE.md) as specified in SKILL.md.
- Boundary markers: No specific delimiters or 'ignore embedded instructions' warnings are included in the snippet, though it is placed as a standalone section.
- Capability inventory: The injected instructions describe the use of the 'search_agent_session_history' tool and CLI commands 'pastlight index' and 'pastlight index --rebuild'. The latter includes explicit double-confirmation instructions to prevent accidental long-running operations.
- Sanitization: The skill replaces existing text with a controlled, static snippet without additional sanitization of the surrounding file content.
Audit Metadata