cc-skill-continuous-learning

Pass

Audited by Gen Agent Trust Hub on Oct 31, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The evaluate-session.sh script executes local shell commands including grep, jq, mkdir, and sed to process configuration files and session metadata.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from session transcripts (via CLAUDE_TRANSCRIPT_PATH). While it lacks sanitization, the processing is limited to counting message markers ("type":"user") using grep -c, which does not involve interpreting or executing the transcript content as instructions.
  • [DATA_EXPOSURE]: The script reads sensitive session transcripts to calculate session length. This data is handled locally and is not transmitted over the network or stored in insecure locations.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 31, 2026, 07:57 PM