MCP 服务器

@git-fabric/chat

An AI conversation management layer that enables creating chat sessions, persisting message history to GitHub, and performing semantic searches over past interactions. It supports multi-turn threading and context injection to integrate external memory sources into Claude conversations.

README

@git-fabric/chat

Chat fabric app — AI conversation sessions, semantic history search, and context threading as a composable MCP layer.

Part of the git-fabric ecosystem.

What it is

A fabric app for AI conversation management — create and manage chat sessions with Claude, persist conversation history to Qdrant Cloud (semantic search over past conversations), support multi-turn threading, and provide context injection from external memory sources (e.g. Aiana).

This is the "conversation plane" of the fabric. Consumers (cortex agents, Claude Desktop, Claude Code via git-steer) use these tools to interact with Claude across sessions.

Tools

Tool	Description
`chat_session_create`	Create a new chat session with optional system prompt, project, model, and title
`chat_session_list`	List recent sessions, filtered by project and state
`chat_session_get`	Get full session with message history
`chat_session_archive`	Mark a session as archived
`chat_session_delete`	Permanently delete a session and all its messages
`chat_message_send`	Send a message and get a Claude response (full multi-turn context)
`chat_message_list`	List messages in a session with pagination
`chat_search`	Semantic search over all stored conversation content
`chat_context_inject`	Inject external context (e.g. Aiana memory recall) into a session
`chat_status`	Aggregate stats: total sessions, messages, tokens today
`chat_health`	Ping Anthropic and Qdrant, returns latency for each
`chat_thread_fork`	Fork a session at a message point to explore an alternative branch

Architecture

Follows the git-fabric layered pattern:

Detection / Query  →  layers/sessions.ts, layers/search.ts (reads)
Action             →  layers/messages.ts, layers/sessions.ts (effectful)
Adapter            →  adapters/env.ts (Anthropic + OpenAI + Qdrant + GitHub)
Surface            →  app.ts (FabricApp factory)

State storage

Sessions + messages → GitHub repo ry-ops/git-steer-state (same state repo as git-steer)
- Session metadata: chat/sessions/{sessionId}.json
- Message history: chat/sessions/{sessionId}/messages.jsonl (JSONL, one message per line)
- Fast listing index: chat/index.json
Semantic vectors → Qdrant Cloud collection chat_fabric__messages__v1 (1536-dim, text-embedding-3-small)
Completions → Anthropic API (claude-sonnet-4-6 default, configurable per session)

Usage

Via gateway (recommended)

# gateway.yaml
apps:
  - name: "@git-fabric/chat"
    enabled: true

Standalone MCP server

ANTHROPIC_API_KEY=sk-ant-... \
OPENAI_API_KEY=sk-... \
QDRANT_URL=https://your-cluster.qdrant.io \
QDRANT_API_KEY=... \
GITHUB_TOKEN=ghp_... \
npx @git-fabric/chat

Programmatic

import { createApp } from "@git-fabric/chat";

const app = createApp();
// app.tools, app.health(), etc.

Environment Variables

Variable	Required	Description
`ANTHROPIC_API_KEY`	Yes	Anthropic API key for Claude completions
`OPENAI_API_KEY`	Yes	OpenAI API key for text-embedding-3-small
`QDRANT_URL`	Yes	Qdrant Cloud cluster URL
`QDRANT_API_KEY`	Yes	Qdrant Cloud API key
`GITHUB_TOKEN`	Yes	GitHub PAT for state repo read/write
`GITHUB_STATE_REPO`	No	State repo (default: `ry-ops/git-steer-state`)