holistic-analysis
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from the analyzed codebase, creating a surface for potential instruction injection via code comments or strings.
- Ingestion Points: Phase 1 and Context Gathering read source code, tests, and git logs into the agent's context.
- Capability Inventory: The agent can read files, execute git and codexray commands, and write to a persistent knowledge store (memory.write).
- Boundary Markers: The skill lacks explicit delimiters or warnings to treat ingested code as data rather than instructions, potentially allowing embedded malicious comments to influence the agent's reasoning.
- Sanitization: While a privacy pre-flight check is implemented for writing to the knowledge store, there is no evidence of sanitization for the ingested codebase content itself.
- [EXTERNAL_DOWNLOADS]: The skill includes an optional accelerator that utilizes the codexray tool from the author's GitHub repository.
- Evidence: References github.com/mthines/codexray in rules/call-graph-map.md. The skill checks for the tool's presence on the system path before use.
- [COMMAND_EXECUTION]: The skill performs project analysis using local tools.
- Evidence: Execution of git log, git diff, and codexray to build a mental model of the codebase.
Audit Metadata