The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from GitHub Pull Request descriptions and diffs, which could contain instructions meant to override agent behavior.
Ingestion points: The skill retrieves PR bodies and code changes using gh pr view and gh pr diff in SKILL.md.
Boundary markers: The skill lacks explicit technical delimiters (such as XML tags or JSON structures) for separating ingested content from instructions, relying instead on natural language guidance.
Capability inventory: The skill has access to Bash, Read, and Grep tools, enabling it to execute local shell commands and read files.
Sanitization: The skill explicitly instructs the agent to ignore directives in the PR body, use the code diff as the authoritative source for classification, and redact any secrets found in diffs using placeholders.
[COMMAND_EXECUTION]: The skill uses the Bash tool to execute gh (GitHub CLI) commands to list, view, and diff Pull Requests within the grafana/tempo repository. This behavior is consistent with the skill's stated purpose of PR assessment.

docs-pr-check