honcho/mcp
ajspig bd6163d9e3
docs: adding honcho-memory skill (#784)
* docs: adding honcho-memory skill

* fix: skills to point at llm friendly content

* docs: split honcho-mcp skill out of honcho-memory; address PR review

Restructure honcho-memory into a concepts/strategy hub that routes to
per-connection path skills, and add a dedicated honcho-mcp skill holding
the MCP-tool mechanics that previously lived inline.

Addresses review feedback on #784:
- honcho-memory step 2 now leads with fast context reads, with chat as
  the slower escalation
- honcho-mcp adds a "Speed: reads vs reasoning" section, describes what
  each context call returns, and a reasoning-levels table
- get_representation framed as a contextualized snapshot insertable into
  a system prompt
- drop schedule_dream from the tool table (manual escape hatch, not
  routine guidance)
- prune queue-status references from honcho-cli; document honcho-mcp in
  vibecoding skill registry

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* refactor(skills): move skills to canonical top-level skills/ with .claude symlink

Establish a single source of truth for agent skills. The real files now live in
the top-level skills/ directory (the publishing convention used by Vercel,
Supabase, and Cloudflare, and the tree Honcho's `npx skills add` distributes).
.claude/skills becomes a symlink to ../skills so Claude Code discovery keeps
working off the one tree — eliminating the parallel-copy sync burden.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

* fix: splitting context into references & verifying content is consistent.

* docs: fixing core language

* docs: fixing core language

* fix: language about observe_others

* chore: adding .agents folder for codex

* fix: add instructions.md into the mcp server & delete mcp skill in favor of including it in honcho-memory.

* chore: remove migrate docs (can be found on older versions)

---------

Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
2026-08-10 15:06:07 -04:00
..
src docs: adding honcho-memory skill (#784) 2026-08-10 15:06:07 -04:00
.dev.vars.example feat(mcp): read HONCHO_API_URL env var to support self-hosted Honcho (#575) 2026-04-20 16:13:16 -04:00
.gitignore feat(mcp): read HONCHO_API_URL env var to support self-hosted Honcho (#575) 2026-04-20 16:13:16 -04:00
README.md feat: support OAuth for MCP clients. (#923) 2026-07-21 15:31:09 -04:00
bun.lock refactor: MCP server improvements (#379) 2026-03-31 17:34:36 -04:00
instructions.md docs: adding honcho-memory skill (#784) 2026-08-10 15:06:07 -04:00
package.json refactor: MCP server improvements (#379) 2026-03-31 17:34:36 -04:00
tsconfig.json refactor: MCP server improvements (#379) 2026-03-31 17:34:36 -04:00
wrangler.toml docs: adding honcho-memory skill (#784) 2026-08-10 15:06:07 -04:00

README.md

Honcho MCP Server

A Cloudflare Worker that implements the Model Context Protocol (MCP) for Honcho, providing AI memory and personalization tools to LLM clients like Claude Desktop.

Quickstart: Use the Hosted Server

  1. Get an API key at https://app.honcho.dev
  2. Add Honcho to your Claude Desktop config:
{
  "mcpServers": {
    "honcho": {
      "command": "bunx",
      "args": [
        "mcp-remote",
        "https://mcp.honcho.dev",
        "--header",
        "Authorization:${AUTH_HEADER}"
      ],
      "env": {
        "AUTH_HEADER": "Bearer <your-honcho-key>"
      }
    }
  }
}

Optional Headers

Header Default Description
X-Honcho-Workspace-ID "default" Workspace to operate in

Available Tools

Workspace: inspect_workspace (aggregates metadata, configuration, and peer/session IDs), list_workspaces (enumerates accessible workspaces), search (semantic search scoped by optional peer/session params), get_metadata, set_metadata

Peers: create_peer, list_peers, chat, get_peer_card, set_peer_card, get_peer_context, get_representation

Sessions: create_session, list_sessions, delete_session, clone_session, add_peers_to_session, remove_peers_from_session, get_session_peers, inspect_session, add_messages_to_session, get_session_messages, get_session_message, get_session_context

Conclusions: list_conclusions, query_conclusions, create_conclusions, delete_conclusion

System: schedule_dream, get_queue_status

Architecture

src/
  index.ts              # Worker entry point — parse config, delegate to MCP handler
  server.ts             # createServer() — registers all tools on an McpServer
  config.ts             # HonchoConfig, parseConfig(), createClient()
  types.ts              # ToolContext, result helpers
  tools/
    workspace.ts        # inspect, list, search, metadata
    peers.ts            # CRUD, chat, card, context, representation
    sessions.ts         # CRUD, peers, messages, inspect, context, clone
    conclusions.ts      # list, query, create, delete
    system.ts           # dream, queue status

Built on:

Self-Hosted Honcho

If you run Honcho yourself (for privacy, latency, or offline use), deploy the MCP Worker alongside your instance and set HONCHO_API_URL in its environment.

Local dev (bun run dev): create mcp/.dev.vars:

HONCHO_API_URL=http://127.0.0.1:28000

Deployed Worker:

wrangler secret put HONCHO_API_URL
# paste your URL when prompted

When HONCHO_API_URL is unset the Worker routes to https://api.honcho.dev, so this change is backward-compatible.

Development

Setup

bun install

Local dev

bun dev

Type-check

bun run tsc --noEmit

Test locally

bunx mcp-remote http://localhost:8787 \
  --header "Authorization:Bearer <key>"

Deploy

bun run deploy              # production
bun run deploy:staging      # staging