RBraga01RBraga01/builder-ai/CLAUDE.md

builder-ai

CLAUDE.md2 starsChanged 5 months agoMITView on GitHub

A drop-in skill and agent pack for teams building LLM-powered products. Three hard gates block incomplete work. Five workflow skills cover the full build cycle. Five specialist agents handle prompt engineering, evaluation, RAG, model selection, and safety. Works standalone. Works alongside A Team. When an agent encounters a task that requires a hard gate, it must: 1. Read the relevant skill file 2. Follow the process 3. Produce the required evidence (eval results, versioned prompt, fallback tests) 4. Only then…

In this repository9 other files it gives its agents.
- AGENTS.mdAGENTS.md
- ai-cost-auditSkill
- ai-safety-reviewSkill
- context-optimizationSkill
- eval-before-shipSkill
- fallback-requiredSkill
- model-benchmarkingSkill
- prompt-versioningSkill
- rag-pipeline-designSkill

The file

We split this CLAUDE.md into the rules it gives an agent

3 rules. About 0 are common, 3 are distinctive, and 0 are very specific to this project.

≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.

Common
Distinctive
Very specific

Distinctive:
When an agent encounters a task that requires a hard gate, it must:
line 56
Distinctive:
Read the relevant skill file
line 57
Distinctive:
Only then mark the task complete
line 60

≈ “Common” is an estimate: at least 1 in 5 of the 26,980 CLAUDE.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.

Files

CLAUDE.md

builder-ai — Build LLM Products That Don't Fail Silently v1.0.0

Works standalone. Works alongside A Team.

Skills in This Pack

Hard Gates (mandatory before shipping)

eval-before-ship — no LLM feature ships without documented eval results
prompt-versioning — all prompts must be versioned in prompts/
fallback-required — every LLM call must have a fallback path

Workflow Skills

rag-pipeline-design — chunking, embedding, retrieval, reranking, generation
model-benchmarking — structured benchmarking before committing to a model
context-optimization — reduce cost and latency without quality loss
ai-cost-audit — audit and project LLM API spend
ai-safety-review — injection, hallucination, abuse, and agentic scope review

Agents in This Pack

Agent	Trigger
`prompt-engineer`	Writing, iterating, or debugging prompts
`eval-designer`	Designing evaluation suites for LLM outputs
`rag-architect`	Designing or debugging retrieval pipelines
`model-selector`	Selecting the cost-optimal model for a task
`ai-safety-reviewer`	Safety and abuse review before user-facing features ship

Expected Directory Layout After Installation

your-project/
├── prompts/              ← versioned prompts (created by prompt-engineer)
│   └── feature-name/
│       ├── v1.0.0.md
│       └── CHANGELOG.md
├── evals/                ← eval suites and results (created by eval-designer)
│   └── feature-name/
│       ├── test-set.jsonl
│       └── results-<date>.md
├── benchmarks/           ← model comparison results (created by model-selector)
├── cost-audit/           ← cost audit reports (created by ai-cost-audit skill)
├── safety-reviews/       ← safety review documents (created by ai-safety-reviewer)
├── skills/               ← this pack's skills
└── .claude/agents/       ← this pack's agents

Using Hard Gates

When an agent encounters a task that requires a hard gate, it must:

Read the relevant skill file
Follow the process
Produce the required evidence (eval results, versioned prompt, fallback tests)
Only then mark the task complete

"Looks good" is not evidence. The skill files specify exactly what counts.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

It worked
Partly
It didn't

What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free