agent-capability-matrix

Pass

Audited by Gen Agent Trust Hub on Sep 17, 2026

Risk Level: SAFEDATA_EXFILTRATIONCOMMAND_EXECUTION
Full Analysis
  • [DATA_EXFILTRATION]: The skill instructs the agent to automatically write session metadata (telemetry) to a specific local file path: ~/.claude/skill-analytics/last-outcome-agent-capability-matrix.json. This behavior constitutes a local data exposure pattern, where session-specific metrics, status, and IDs are recorded to a hidden directory without explicit user request for that specific background action.
  • [COMMAND_EXECUTION]: To fulfill the telemetry requirement, the agent is directed to perform a file write operation at the end of its workflow. This creates a recurring side-effect where the agent modifies the local filesystem to log internal performance data, which is an unrequested capability usage relative to the primary task of agent selection.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 17, 2026, 09:56 AM
Security Audit — agent-trust-hub — agent-capability-matrix