The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill provides workflow templates that interpolate untrusted GitHub event data directly into shell commands, which could be exploited if event properties (like tag names or issue numbers) are maliciously crafted.
Ingestion points: ${{ github.ref_name }}, ${{ github.event.issue.number }}, and ${{ github.event.number }} in SKILL.md.
Capability inventory: Shell command execution via gh CLI and npm scripts.
Boundary markers: None present in the provided templates to delimit untrusted data.
Sanitization: No explicit sanitization or escaping is shown in the shell snippets.
[COMMAND_EXECUTION]: The documentation contains multiple snippets demonstrating the use of gh CLI, npm, curl, and jq for interacting with GitHub services and managing project dependencies. These operations are consistent with the skill's educational purpose for CI/CD modernization.
[EXTERNAL_DOWNLOADS]: Demonstrates the use of npx pin-github-action to fetch a utility from the npm registry. This is presented as a method for improving security through SHA pinning of actions.

github-actions-2025