# Agent Builder Hacker

> Use when you need to rapidly hack together working AI agents from existing parts.

- Skill: `harishkotra/agent-builder-hacker` (Agent Skill)
- Install (CLI): `npx skillmds@latest add harishkotra/agent-builder-hacker`
- Raw SKILL.md: https://api.skillmd.com/api/skills/harishkotra/agent-builder-hacker/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: harishkotra (https://skillmd.com/u/harishkotra)
- Updated: 2026-09-22
- Page: https://skillmd.com/skills/harishkotra/agent-builder-hacker

---


# Agent Builder Hacker
Rapidly assemble working AI agents from existing tools, models, and APIs.

## When to use
- You need a working agent fast, from off-the-shelf parts.
- You want to iterate quickly rather than build infrastructure from scratch.

## Procedure
1. Clarify the single job the agent must do.
2. Pick the simplest stack: an OpenAI-compatible model + the fewest tools needed.
3. Build the happy path first; get it working end to end before hardening.
4. Add error handling and a fallback only after the core works.
5. Test with real inputs, not just ideal ones.

## Pitfalls
- Don't over-engineer before the happy path works; speed is the point.
- A working-but-fragile agent is fine for a prototype; say what it's not ready for.

## Verification
- The agent completes its core job on real inputs end to end.