Agent Engineering
Engineering discipline for AI coding agents: four always-on principles, TDD, verification gates, architecture playbooks, a product manager, docs/ project tracking and Copilot porting.
Engineering discipline for AI coding agents: four always-on principles, TDD, verification gates, architecture playbooks, a product manager, docs/ project tracking and Copilot porting.
What is inside
- Four Principles (skill). Four always-on rules for your coding agent: think before coding, keep it simple, change only what was asked, and loop until a check passes.
- Test-Driven Development (skill). Makes Claude write the failing test first, watch it fail, then write the minimum code to pass. Includes counters for the usual excuses to skip a test.
- Verification Before Completion (skill). Stops Claude from claiming work is done, fixed or passing until it has run the check and read the output. Evidence before assertions.
- System Architecture (skill). Pick the architecture pattern and the technology on purpose, before framework details: pattern trade-offs, design workflows and decision guides for databases, queues and APIs.
- PM: Scope-Disciplined Product Manager (skill). A product manager for your agent team: build-ready PRDs with testable gates, a roadmap, brainstorming, standups and a single launch or no-launch verdict, with a scope guard that cites the spec.
- Docs Project Management (skill). Manage a whole project inside the repo's docs/ folder: roadmap, PRDs, epics, user stories, test cases and bugs as linked Markdown files with a status lifecycle. No issue tracker needed.
- Convert to Copilot (skill). Port a Claude Code agent, with its command and skills, to one GitHub Copilot agent. Maps tools, models and sub-agent calls, validates the result and ships scripts for both directions.
- Golden Rules (rule). Two short rules for any Claude Code project: don't repeat yourself, and never guess: always answer with a link to the evidence.
Install
Open any item above to download it and see its own install steps.
Or install the agent-engineering plugin from the Has Brains agent kit in Claude Code:
/plugin marketplace add endorphin-ai/hasbrains-agent-kit
/plugin install agent-engineering@hasbrains-agent-kit
The plugin does not include: Golden Rules. Download those from their pages.
Source: endorphin-ai/hasbrains-agent-kit. License: MIT.
-
assets/maps/references/scripts/workflows/
A product manager for your agent team: build-ready PRDs with testable gates, a roadmap, brainstorming, standups and a single launch or no-launch verdict, with a scope guard that cites the spec.