The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes several shell commands to gather repository state and forge data. Specifically, it uses git to check status, branch history, and diffs (e.g., git status --porcelain, git diff --stat). It also leverages the gh CLI to retrieve pull request and issue information from GitHub. Furthermore, it invokes a local project script using bun scripts/pre-execution-snapshot.mjs verify to validate the integrity of specific workflow artifacts.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external and project-defined sources that could theoretically contain malicious instructions.
Ingestion points: The skill reads from local markdown files including ROADMAP.md, progress.md, and known-issues.md, as well as remote data via gh pr list and gh issue list.
Boundary markers: The skill does not employ explicit delimiters for all ingested text but strictly enforces a fixed JSON output envelope, limiting how the agent can interpret the processed data.
Capability inventory: The agent possesses filesystem access (via Git and documentation reads) and network capabilities (via the GitHub CLI).
Sanitization: The instructions include a robust mitigation strategy for urgency signals, mandating that the agent derive 'urgent' status exclusively from GitHub's labels object while explicitly ignoring free-text fields like titles, bodies, and comments to prevent content-based injection attacks.
[EXTERNAL_DOWNLOADS]: The skill communicates with GitHub to synchronize project state. It performs git fetch and uses the gh CLI to pull metadata from the forge. These operations target GitHub, which is a well-known and trusted service for development workflows.

workflow-status