The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches unresolved issue data from the Sentry API (e.g., sentry.io) and retrieves pull request information from GitHub using the gh CLI. Both Sentry and GitHub are well-known services, and the skill uses them strictly for issue triaging and PR status checks.
[COMMAND_EXECUTION]: Executes a suite of local bash scripts (preflight.sh, fetch-issues.sh, filter-candidates.sh) to coordinate the triage workflow. It also invokes standard command-line utilities including git for commit log analysis, gh for PR management, jq for JSON processing, and curl for API communication.
[INDIRECT_PROMPT_INJECTION]: Processes data received from the Sentry API, specifically issue IDs and titles, which are used to filter candidates and passed as arguments to a downstream fixer skill.
Ingestion points: Issue metadata retrieved via Sentry REST API in fetch-issues.sh and filter-candidates.sh.
Boundary markers: Absent; the skill relies on the structure of Sentry identifiers.
Capability inventory: Shell execution of local scripts, network access to Sentry and GitHub APIs, and reading of local project metadata files (AGENTS.md, CLAUDE.md).
Sanitization: Employs jq for robust JSON parsing and utilizes git log with the --fixed-strings flag to ensure issue identifiers are treated as literal strings during history searches.
[CREDENTIALS_UNSAFE]: The skill requires the SENTRY_AUTH_TOKEN environment variable to authenticate with the Sentry API. It performs a pre-flight check in preflight.sh to verify the token's presence, ensuring that credentials are managed via the environment rather than being hardcoded in scripts or instructions.

sentry-top-issue