The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from GitHub issues, pull requests, and local documentation to generate its recommendations.\n
Ingestion points: Reads issue and PR metadata (titles, labels) via the gh CLI and parses local documentation in docs/plans/ and docs/adr/.\n
Boundary markers: No explicit delimiters or instruction-bypass warnings are used when processing these external strings for the dashboard output.\n
Capability inventory: The skill uses Bash to execute git/gh commands and Write/Edit to manage local snapshot files and update configuration files like .gitignore and CLAUDE.md.\n
Sanitization: The skill uses gh --jq to extract specific metadata fields, which minimizes the ingestion of arbitrary text from issue and PR bodies.\n- [COMMAND_EXECUTION]: Performs extensive read-only shell operations using git and gh to collect project telemetry, such as branch state, commit history, and stash lists.\n- [SAFE]: File modifications are limited to a specific scratch directory (docs/status/) and appending non-executable policy declarations to existing configuration files with user approval.

statuskit