export-session-as-markdown
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes a Python script using shell command interpolation:
python3 "${CLAUDE_PLUGIN_ROOT}/scripts/reduce-transcript.py" "<transcript_path>" "<output_path>" --markdown. - Evidence: Found in Step 3 of SKILL.md.
- Risk: If the
<transcript_path>or<output_path>variables (which can be user-supplied) contain shell metacharacters, it could lead to arbitrary command execution. - [INDIRECT_PROMPT_INJECTION]: The skill processes session transcripts (
.jsonlfiles) which contain the history of agent interactions, including potentially untrusted data from tool outputs or external web pages. - Ingestion points: The
transcript_pathfile read in SKILL.md. - Boundary markers: None identified to separate instructions from transcript content.
- Capability inventory: Shell command execution via
python3for script processing. - Sanitization: No sanitization or validation of the transcript content is mentioned before it is processed by the script.
Audit Metadata