A self-hosted HTTP registry that stores agent capability declarations and answers routing queries: which agents handle a given intent, accept a conversation kind, or follow a specific agent in a sequencing chain. Part of the clagentic suite.
- Capability registry. Each agent declares its intents, conversation kinds, sequencing constraints, trust labels, and return format in a versioned YAML entry.
- Routing queries. Three query surfaces:
FindByCapability(intent...),FindByConversationKind(kind),FindBySequencing(afterAgent). - Two registry backends.
--registry-source filefor local directories (with fsnotify hot-reload);--registry-source gitfor a tracked git repo (polled). - Strict vocabulary validation.
schema_version: 2entries are validated against a closed vocabulary. Extend it per-deployment with--vocabulary-extensions. - Self-build mechanisms. Three opt-in mechanisms observe running agents and write
proposed_changes/entries for operator review: MCP discovery, engram watch, usage-driven inference.
No runtime dependencies. Drop-in binary or container. Operators own the registry; agents never write to it directly.
# Build
go build -o clagentic-directory ./cmd/clagentic-directory/
# Run against the example registry
./clagentic-directory \
--registry-source file \
--registry-dir ./examples/registry/ \
--listen :8444
# Query
curl -s localhost:8444/v1/agents | jq '.[].name'
curl -s localhost:8444/v1/agents/find?intent=code-review | jq
curl -s localhost:8444/readyzgo install github.com/clagentic/clagentic-directory/cmd/clagentic-directory@latestOr build from source:
git clone https://github.com/clagentic/clagentic-directory
cd clagentic-directory
go build -o clagentic-directory ./cmd/clagentic-directory/Requirements: Go 1.25+.
-
inspect— one-shot MCP introspection for an agent, writes aproposed_changes/entry without starting the service:clagentic-directory inspect --agent <name> --mcp-url <http://localhost:PORT/mcp>
See docs/SELF-BUILD.md for full flag reference and operator workflow.
| Document | Contents |
|---|---|
| docs/DESIGN.md | Architecture, store interface, data flow |
| docs/SCHEMA.md | Agent entry schema reference (v1 and v2) |
| docs/VOCABULARY.md | Canonical vocabulary for v2 entries |
| docs/DEPLOY.md | Operator deployment guide |
| docs/SELF-BUILD.md | Self-build mechanisms (MCP, engram watch, usage inference) |
| docs/A2A-COMPAT.md | A2A AgentCard compatibility |
| docs/SECURITY.md | Security and auth notes |
If clagentic:directory is useful to you: ko-fi.com/clagentic
Not affiliated with Anthropic or OpenAI. Claude is a trademark of Anthropic. Codex is a trademark of OpenAI. Provided "as is" without warranty. Users are responsible for complying with their AI provider's terms of service.
FSL-1.1-MIT — Functional Source License 1.1, with MIT as the Change License.
Free for personal, internal-business, evaluation, research, and non-commercial use. Not free for offering this tool (or a substantial fork) as a competing commercial product. Each release auto-converts to MIT on its second anniversary.