Providers

Integrations start with providers.

Providers are AtlasClaw's integration boundary. Each provider packages the target-system contract, one or more Skills, implementation scripts, auth context, and the reference material needed to keep enterprise AI integrations understandable.

Runtime architecture

See how a provider carries runtime execution.

This diagram shows how a provider carries auth, skills, scripts, and downstream system execution.

AtlasClaw provider runtime overview for Skills, scripts, auth, and enterprise system execution Click to enlarge

Provider runtime overview

The provider keeps auth, skills, scripts, and downstream execution inside one explicit boundary.

Package structure

Provider package shape.

providers/<provider-name>/
├── PROVIDER.md
├── README.md
└── skills/
    ├── <skill-a>/
    │   ├── SKILL.md
    │   ├── scripts/
    │   └── references/
    └── <skill-b>/
Key rules

Keep boundaries explicit and scripts narrow.

  • The provider owns how AtlasClaw identity becomes provider-native target-system auth.
  • Skills stay human-readable while still binding to concrete entrypoints.
  • Provider-qualified naming avoids collisions across integrations.
  • Scripts should stay narrow, predictable, and safe with secrets and error mapping.
Naming

Provider-qualified skills.

Provider prefixes prevent skill collisions across systems.

{
  "providers_root": "../atlasclaw-providers/providers",
  "webhook": {
    "systems": [
      {
        "allowed_skills": ["jira:jira-issue"]
      }
    ]
  }
}
Examples

Study the references, then extend.