codebase-onboarding
Pass
Audited by Gen Agent Trust Hub on Aug 5, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill has an inherent surface for indirect prompt injection due to its primary function.
- Ingestion points: The agent reads various files throughout the codebase in SKILL.md (Phases 1 and 3) and specifically reads existing
CLAUDE.mdfiles (Phase 4). - Boundary markers: There are no specific instructions to use delimiters or ignore blocks when processing the content of these files.
- Capability inventory: The skill involves writing to the local file system (generating or updating the
CLAUDE.mdartifact). - Sanitization: No explicit sanitization or escaping of harvested content is mandated before interpolation into output artifacts. However, the skill instructs the agent to verify findings against actual code and keep outputs concise, which provides some mitigation.
- [DATA_EXPOSURE]: The skill reconnaissance phase targets project metadata and configuration files such as
package.json,.env.example, and framework configs. This information is used locally for documentation and no exfiltration patterns were identified.
Audit Metadata