channel-registry
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses shell commands to resolve the project root via git rev-parse and executes a local script named registry-events.py to manage the event stream. These operations are restricted to the local environment and the skill's own logic.
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data described as untrusted evidence from social channel exports and messages, creating a potential surface for indirect injection.
- Ingestion points: Data enters the context through channels.ndjson, external channel exports, user-provided messages, and platform rule snapshots (SKILL.md).
- Boundary markers: The instructions explicitly warn the agent that channel exports and messages are untrusted evidence and require explicit authorization for saving results (SKILL.md).
- Capability inventory: The skill is capable of executing registry-events.py, querying projections, and performing file verification (SKILL.md).
- Sanitization: The procedure requires that capability values never enter request JSON, files, or logs, and mandates minimizing the collection of personal data (SKILL.md).
Audit Metadata