The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill is instructed to read nx.json at the workspace root to verify the presence of nxCloudId or nxCloudAccessToken for authentication.
[EXTERNAL_DOWNLOADS]: The skill performs remote operations including:
Executing package manager installation commands (pnpm install, yarn install, npm install) to resolve lockfile issues.
Using nx apply-locally <shortLink> to fetch and apply patches from the Nx Cloud service.
[REMOTE_CODE_EXECUTION]: Applying remote patches via nx apply-locally modifies the local source code, which is then executed via nx run <taskId> or committed and pushed to CI.
[COMMAND_EXECUTION]: The skill has extensive command-line capabilities including:
Git operations: git branch, git status, git commit, and git push to the remote repository.
Task execution: Running arbitrary Nx tasks using <pm> nx run <taskId>.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection (Category 8) due to its processing of external data:
Ingestion points: Data enters the agent context via the ci-watcher subagent, specifically through taskOutputSummary, suggestedFixReasoning, and failedTaskIds which are derived from external CI logs.
Boundary markers: No explicit delimiters or boundary markers are defined to isolate untrusted CI output from the agent's core logic.
Capability inventory: The agent possesses high-impact capabilities including git push, pnpm/npm/yarn install, and nx run (subprocess execution).
Sanitization: There is no evidence of sanitization or filtering of the CI logs before they are analyzed for code generation or enhancement.

nx-ci-monitor