awesome-agents
CLAUDE.md0 starsChanged 3 months agoMITView on GitHub
A production-grade Claude Code control centre: 18 engineering teams, 135 specialist agents, each written to a top 0.1% professional standard. Before doing domain work yourself, check whether an agent owns it. Roster and name values: .claude/teams/README.md. Delegation chain: architect decides, engineer implements, quality verifies, release ships. Do not skip a link. .claude/agents/*.md and .claude/teams/*.md are generated. Edit scripts/agents_data.py and run python3 scripts/generate.py. Hand edits to generated files are lost. /team route work · /design architecture pass + ADR · /review…
In this repository16 other files it gives its agents.
- AGENTS.mdAGENTS.md
- .claude/commands/agents.mdAGENTS.md
- .opencode/AGENTS.mdAGENTS.md
- .github/copilot-instructions.mdCopilot instructions
- .cursor/rules/caveman.mdcCursor rule
- adrSkill
- agent-registrySkill
- cost-reviewSkill
- data-contractSkill
- delegation-protocolSkill
- marketplaceSkill
- onboardingSkill
- orchestrationSkill
- postmortemSkill
- runbookSkill
- threat-modelSkill
The file
We split this CLAUDE.md into the rules it gives an agent
3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 8
Before doing domain work yourself, check whether an agent owns it.
- Distinctive: line 31
Do not skip a link.
- Distinctive: line 68
No secrets in code, logs, or prompts.
≈ “Common” is an estimate: at least 1 in 5 of the 27,641 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.
Files
awesome-agents
A production-grade Claude Code control centre: 18 engineering teams, 135 specialist agents, each written to a top 0.1% professional standard.
Delegate, do not improvise
Before doing domain work yourself, check whether an agent owns it.
Roster and name values: .claude/teams/README.md.
- Architecture Team (
.claude/teams/architecture.md) — Own system-wide technical direction: boundaries, contracts, trade-offs, and long-term evolvability. - Backend Engineering Team (
.claude/teams/backend.md) — Build correct, observable, horizontally scalable services and the data layer beneath them. - Frontend Engineering Team (
.claude/teams/frontend.md) — Deliver fast, accessible, resilient user interfaces backed by a coherent design system. - Mobile Engineering Team (
.claude/teams/mobile.md) — Ship native-quality mobile experiences that survive poor networks, old devices, and store review. - Data & AI Engineering Team (
.claude/teams/data-ai.md) — Turn raw data into trustworthy products and ship AI systems that are evaluated, not vibed. - Platform Engineering Team (
.claude/teams/platform.md) — Provide paved roads that make the secure, reliable path the fastest path for product teams. - Quality Engineering Team (
.claude/teams/quality.md) — Prevent defects from reaching users through risk-based testing, automation, and fast feedback. - Security Engineering Team (
.claude/teams/security.md) — Make the system secure by default and prove it with threat models, tests, and controls evidence. - Release & Reliability Team (
.claude/teams/release.md) — Deliver change safely and keep production healthy against explicit SLOs and error budgets. - Product Strategy Team (
.claude/teams/product.md) — Decide what to build and why, with evidence, sequencing, and measurable outcomes. - Engineering Excellence Team (
.claude/teams/excellence.md) — Raise the floor of engineering practice through documentation, standards, and measurement. - IT Operations & Infrastructure Team (
.claude/teams/itops.md) — Run the corporate and production infrastructure that everything else stands on: servers, network, identity, endpoints, and recovery. - MLOps & Model Operations Team (
.claude/teams/mlops.md) — Take models and prompts from a notebook to reliable, monitored, reproducible production systems. - FinOps & Cost Engineering Team (
.claude/teams/finops.md) — Make technology spend visible, attributable, and efficient without slowing delivery down. - Enterprise Applications Team (
.claude/teams/entapps.md) — Deliver and integrate the internal systems the business runs on: ERP, CRM, workflow, and low-code platforms. - Governance, Risk & Privacy Team (
.claude/teams/grc.md) — Keep the organisation defensible: privacy by design, governed data, tracked risk, and audit evidence generated by systems. - Design Team (
.claude/teams/design.md) — Decide what the interface looks like and how it behaves, before a line of component code is written. - Leadership Team (
.claude/teams/leadership.md) — Decide across teams what no single team can decide alone, and own the outcome when they disagree.
Delegation chain: architect decides, engineer implements, quality verifies, release ships. Do not skip a link.
Generated files
.claude/agents/*.md and .claude/teams/*.md are generated. Edit scripts/agents_data.py
and run python3 scripts/generate.py. Hand edits to generated files are lost.
Standing rules
.claude/rules/engineering-standard.md— baseline for all code..claude/rules/agent-authoring.md— how to change agents..claude/rules/output-discipline.md— how results are reported.
Commands
/team route work · /design architecture pass + ADR · /review multi-lens diff review ·
/ship readiness gate · /audit security audit · /incident incident command ·
/threat-model STRIDE · /cost cost review · /runbook on-call runbook ·
/agents find the owner · /onboard explain this system.
Plugins
This repo is a Claude Code marketplace: /plugin marketplace add bakhod1r/awesome-agents.
For vendor tooling and live data access, use the marketplace skill — install, never vendor.
Output compression
Caveman is always on in this repository — see AGENTS.md for the rule other IDE agents load.
Prose is compressed; code, commands, file paths, and error strings are reproduced verbatim.
Drop it for security warnings, irreversible-action confirmations, and anything where a
fragment would be ambiguous, then resume. Commits, PRs, and documentation are written normally.
Install: /plugin marketplace add juliusbrussee/caveman, then /caveman ultra.
Non-negotiable
Verified over asserted. Measured over estimated. Reversible over clever. No secrets in code, logs, or prompts. No change without tests, observability, and a rollback path.
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
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.

