What is ProAgents?

ProAgents is an open-source agentic CLI + Agent Skill that equips existing coding agents with professional expertise, methods, skills, rules, tools and verification practices — a portable, structured, versioned Professional Agent Profile.

It does not replace Claude Code, Codex, OpenCode, Gemini CLI, Cursor or any other coding-agent harness. It gives them a profession.

Existing Coding Agent
        +
Professional Agent Profile
        ↓
Professional Agent

The core abstraction

A generic coding agent can write code. A professional agent knows how a professional in a particular discipline approaches the work: what to check first, what never to do, what evidence to produce before claiming completion.

A Professional Agent Profile defines the professional layer an agent operates under:

Professional Profile
│
├── Identity
├── Expertise
├── Knowledge
├── Methods
├── Skills
├── Rules
├── Policies
├── Standards
├── Tool requirements
└── Verification

A profile is not a prompt. It is a structured, composable and versioned definition of how an agent should operate within a profession:

proagent equip security-engineer

The underlying model does not change. The agent's professional capabilities and operating discipline do.

The architecture

ProAgents sits above existing coding-agent harnesses:

                       HUMAN
                         │
                         ▼
                    PROAGENTS
                         │
                         ▼
              PROFESSIONAL PROFILE
                         │
        ┌────────────────┼────────────────┐
        ▼                ▼                ▼
    KNOWLEDGE          METHODS           RULES
        │                │                │
        └────────────────┼────────────────┘
                         ▼
                       SKILLS
                         │
                       TOOLS
                         │
                   VERIFICATION
                         │
                         ▼
                 PROFILE COMPILER
                         │
          ┌──────────────┼──────────────┐
          ▼              ▼              ▼
     Claude Code       Codex        OpenCode
          │              │              │
          └──────────────┼──────────────┘
                         ▼
                  CODING AGENT

The canonical profile is provider-agnostic and harness-agnostic. The profile compiler (src/adapters/) decides how a target harness expresses it using that harness's own mechanisms: project instructions, skills directories, rules and hooks.

Two ways to work

Path	When	How
Equip a profile	Your coding agent is fine, but it should operate like a professional in a discipline	proagent detect → proagent equip security-engineer
Build a crew spec	The professional system you need doesn't exist yet — as a spec your harness executes	proagent init --intent "…" → progressive interview → spec/validate/build

Both paths share the same primitives: skills, rules, verification, context frameworks, validation codes, and the registry.

Design non-negotiables

• Provider-agnostic — the canonical profile never depends on a harness; adapters compile per target
• Harness-aware — compilation uses the strongest mechanism the target supports and reports limitations honestly where it can't
• Rules are enforced, not suggested — where the harness supports enforcement (hooks, policies), ProAgents compiles into it; otherwise it says so
• Composition is validated — conflicting rules, incompatible tools, capability gaps: serious conflicts are never silently ignored
• JSON-first — every operation has deterministic --json output
• Local and inspectable — profiles are plain JSON; equipped output is plain files in your repo

Next steps

• Getting started — detect, list, equip in five minutes
• Professional profiles — the schema, composition and validation codes
• Agent operating guide — how another AI agent drives the whole flow
• Context frameworks — builtin adapters and writing your own