The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted issue descriptions and titles which could contain malicious instructions to bias the agent's triage verdicts.
Ingestion points: Issue bodies and titles are fetched from the GitHub API using gh project item-list or GraphQL as described in references/board-queries.md.
Boundary markers: The instructions do not define explicit delimiters or 'ignore' instructions for the processed issue text.
Capability inventory: The skill possesses write capabilities through gh api graphql (mutations) and gh issue edit for updating board fields and labels.
Sanitization: No sanitization or filtering of issue content is implemented before the agent processes the text for triage categorization.
[DYNAMIC_EXECUTION]: The skill uses a small inline Python script to perform integrity checks on JSON data during board enumeration.
Evidence: references/board-queries.md contains a shell block that executes python3 with a heredoc script to validate that the number of retrieved items matches the total count reported by GitHub.
[COMMAND_EXECUTION]: The skill makes extensive use of the GitHub CLI and standard Unix utilities to interact with external services and manage data.
Evidence: Multiple bash blocks in references/board-queries.md implement logic using gh, jq, awk, and grep to resolve project identifiers and verify writes.

triage-board