monitor-ci
Pass
Audited by Gen Agent Trust Hub on Apr 6, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The skill uses dynamic context injection placeholders to retrieve local repository metadata, such as the current branch and commit hash, to populate the agent's context. It also executes local git commands and Nx CLI tasks to verify and apply code fixes.
- [EXTERNAL_DOWNLOADS]: The skill communicates with Nx Cloud via specific MCP tools to fetch CI pipeline status and suggested fixes. This interaction is restricted to the services provided by the workspace vendor.
- [PROMPT_INJECTION]: The skill ingests external data from CI task failure summaries and AI-suggested fixes. While this represents a potential surface for indirect prompt injection, the risk is mitigated by the skill's structure, which includes local verification steps and budget gates for automated attempts.
- Ingestion points: CI information and suggested fixes retrieved through mcp__plugin_nx_nx-mcp__ci_information.
- Boundary markers: Not explicitly defined in the subagent prompts.
- Capability inventory: The agent can perform git operations and execute local build tasks via the Bash tool.
- Sanitization: The skill uses deterministic decision scripts to process status codes rather than directly interpreting raw external strings as instructions.
Audit Metadata