The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes repository-owned files and logs to generate or repair CI/CD pipelines, creating a surface for indirect instructions. Ingestion points: Jenkinsfiles, application manifests, lockfiles, and build logs. Boundary markers: Mandatory human approval gates for all durable decisions and repository edits, supported by a structured Compact Decision Proposal (ADR). Capability inventory: Repository file writes, remote build triggers, and remote API-based validation. Sanitization: Explicit constraints against secret exposure and instructions to use sanitized log data. \n- [EXTERNAL_DOWNLOADS]: The skill includes instructions to discover and install specialized sub-skills from the author's repository space using system-level commands like npx. \n- [DYNAMIC_EXECUTION]: Implements a 'Specialist Contract' that enables the runtime loading of stack-specific sub-skills based on the detected runtime stack of the repository. \n- [COMMAND_EXECUTION]: The skill requires executing various build, test, and deployment commands (e.g., pnpm, cargo, composer, docker) within the agent's workspace to validate or implement pipeline stages.

jenkins-devops-acrazie