skills/av/pace/pace-config/Gen Agent Trust Hub

pace-config

Pass

Audited by Gen Agent Trust Hub on Jun 25, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute various local shell commands for the 'pace' dashboard application, including status checks (pace adapters list), configuration validation (pace config check), and service management (docker run, docker-compose restart, bun run dev). These commands are consistent with the skill's purpose of managing a local application.
  • [EXTERNAL_DOWNLOADS]: The configuration generated by the skill facilitates the dashboard's primary function: fetching data from third-party feeds such as RSS, Reddit, GitHub, and Mastodon. It also references a Docker image (ghcr.io/av/pace:latest) associated with the skill's author.
  • [INDIRECT_PROMPT_INJECTION]: The application being configured has an inherent attack surface for indirect prompt injection.
  • Ingestion points: Untrusted data is ingested from external adapters (RSS, Reddit, Mastodon, etc.) as defined in SKILL.md.
  • Boundary markers: No specific boundary markers or 'ignore' instructions are provided in the generated LLM transform configurations in the examples.
  • Capability inventory: The system performs network fetches and processes content through an LLM for summarization and filtering as described in the Transform reference section.
  • Sanitization: No explicit sanitization of feed content before LLM processing is documented in the configuration templates. Users should be aware that malicious external content could potentially influence LLM summarization behavior.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 25, 2026, 11:07 AM
Security Audit — agent-trust-hub — pace-config