The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external pull request sources, creating a surface for indirect instructions.\n
Ingestion points: SKILL.md instructions specify reading the pull request title, body, linked issues, and review threads.\n
Boundary markers: Absent; the skill does not instruct the agent to ignore or delimit embedded instructions in the ingested data.\n
Capability inventory: The skill uses git for rebasing and pushing, and executes repository checks via subprocess.\n
Sanitization: Absent; no sanitization of PR content is mentioned before it is processed by the agent.\n
Multi-step chains: The skill chains with leaving-pr-comments, potentially passing injected content into another skill's output.\n- [COMMAND_EXECUTION]: The skill performs shell commands to interact with the repository and git history.\n
Evidence: Executes git fetch, git rebase, git range-diff, and git push --force-with-lease. It also runs "repository's focused checks", which typically involves running tests or build scripts that can execute arbitrary code within the repository's context.

modernize