The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes various command-line utilities to manage repository state and interact with the GitHub API through the gh CLI. Evidence includes the use of git status, git log, gh issue list, and gh pr view for automated triage and review workflows.- [REMOTE_CODE_EXECUTION]: The agent is designed to execute locally-defined test scripts and implementation plans to verify code quality. Evidence is found in the scripts/run-tests.sh file, which auto-detects and triggers standard test runners such as npm test, cargo test, make test, or pytest.- [EXTERNAL_DOWNLOADS]: Standard package management tools used by the skill may initiate network requests to download dependencies or binaries from official registries during the verification process. Evidence includes the use of npx and other package manager commands within the testing scripts.- [PROMPT_INJECTION]: The skill has a surface for indirect prompt injection as it ingests and acts upon untrusted data from repository sources. Ingestion points: The skill reads git diff output, gh issue/pr content, and project files like README and AGENTS.md. Boundary markers: No explicit delimiters are used to isolate untrusted data from instructions. Capability inventory: The skill can execute shell commands, modify files (via commits), and perform GitHub API actions. Sanitization: No explicit sanitization or filtering of external content is documented in the skill instructions.

check