The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a custom CLI tool using node:child_process.spawnSync. Command arguments are handled as arrays to mitigate shell injection risks.
[EXTERNAL_DOWNLOADS]: The library scripts/lib/cli.mjs includes logic to download and run the @catalyst-cloud/catalyst-skills package via npx if the tool is not already installed locally.
[DYNAMIC_EXECUTION]: The is-it-mergeable.mjs script dynamically constructs SQL queries to analyze pull request review threads stored in a local replica database.
[INDIRECT_PROMPT_INJECTION]:
Ingestion points: Data from GitHub pull requests, including titles, bodies, and review comments, is ingested through scripts/read-pr.mjs, scripts/is-it-mergeable.mjs, and scripts/lib/pull.mjs.
Boundary markers: The instructions do not define explicit delimiters for untrusted content retrieved from external PRs.
Capability inventory: The skill has the ability to execute shell commands via the Bash tool in SKILL.md, though it is scoped to specific vendor commands.
Sanitization: While SQL queries in scripts/is-it-mergeable.mjs are sanitized using quote-escaping, other PR metadata is passed directly to the agent without specialized filtering.

catalyst-github