consensus-review
Pass
Audited by Gen Agent Trust Hub on Aug 26, 2026
Risk Level: SAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [DATA_EXPOSURE_AND_EXFILTRATION]: The skill accesses sensitive local file paths to retrieve data for review.\n
- Evidence: Accesses
~/.claude/plans/*.mdto resolve plan documents and~/.codex/consensus.jsonfor configuration settings, as defined inSKILL.mdandreferences/command.md.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data (documents, architecture proposals, and user descriptions) and interpolates them into model prompts.\n - Ingestion points: Processes content from
FILE_PATH,~/.claude/plans/*.md, and{USER DESCRIPTION, VERBATIM}inreferences/command.md.\n - Boundary markers: Absent. The content is interpolated directly into prompts without delimiters or instructions to ignore embedded instructions.\n
- Capability inventory: Reading local files and interacting with external model providers.\n
- Sanitization: No evidence of sanitization or filtering for the ingested content.
Audit Metadata