The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it ingests untrusted data from external sources.
Ingestion points: Processes pull request reviews, inline threads, conversation comments, and GitHub Actions CI logs as specified in SKILL.md.
Boundary markers: Includes a 'Retrieved Evidence Boundary' section in SKILL.md that explicitly instructs the agent to treat external content as data rather than instructions.
Capability inventory: Performs file writes to the .tigerkit/ directory, executes git push to remote repositories, and makes GitHub API calls to resolve threads and post comments.
Sanitization: Instructions require summary bodies to be prepared as literal text in files or structured arguments to avoid shell expansion risks.
[DYNAMIC_EXECUTION]: The skill uses a Test-Driven Development (TDD) and Software Design Document (SDD) workflow to address feedback.
It dynamically generates code fixes and verification tests based on the pull request context.
It executes these generated tests at runtime to ensure the regression is fixed before committing or pushing changes, as detailed in references/testing.md and references/sdd.md.
[COMMAND_EXECUTION]: The skill performs various shell-based operations for repository management.
It uses standard git and gh (GitHub CLI) commands to interact with pull requests and branches.
It references and executes a local Python script scripts/durable_artifacts.py to manage durable artifact storage locations in references/artifact-paths.md.

tk-pr-respond