food-waste
Warn
Audited by Gen Agent Trust Hub on Mar 23, 2026
Risk Level: MEDIUMDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The 'Self-Evolution Telemetry' section instructs the agent to search for and write to '~/.claude/projects/'. This directory is an internal configuration path for the Claude Code agent containing project metadata and history. Accessing and appending to files in this location constitutes a persistence mechanism and a risk for exposing internal agent state.
- [PROMPT_INJECTION]: The skill is highly susceptible to indirect prompt injection (Category 8) because it is designed to ingest and interpret entire codebases and configuration files (Phase 1, Step 1.1).
- Ingestion points: Reads backend, database, and ML configuration files.
- Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present in the prompt.
- Capability inventory: The skill has the ability to read files across the system and write telemetry data to sensitive internal directories.
- Sanitization: No sanitization or validation of the ingested code or data is specified before it is processed by the agent to generate assessments.
Audit Metadata