session-trends
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local Python scripts (
compute-metrics.py) and shell commands to process metric data from the.claudedirectory.\n- [DYNAMIC_EXECUTION]: Utilizespython3 -cto run inline Python snippets for querying and aggregating session metrics, which constitutes dynamic code execution from predefined templates.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes session data that could be influenced by external inputs, creating a surface for indirect prompt injection.\n - Ingestion points: Reads session metrics from
.claude/session-metrics/metrics.jsonland project memory fromMEMORY.md.\n - Boundary markers: No explicit boundary markers or instructions to ignore embedded content are present in the processing pipeline.\n
- Capability inventory: The skill can execute Python scripts, write trend data to
.claude/session-metrics/trends.json, and output formatted reports to the user.\n - Sanitization: The skill relies on standard
json.loadsparsing, which provides structure but does not sanitize the content of the metrics for potential downstream injection attacks.\n- [SAFE]: Includes a reference to a GitHub repository (github.com/earendil-works/pi) for methodological inspiration, which is documented as a neutral reference.
Audit Metadata