Give coding agents a live map of your repo.

agentmako — local-first MCP server for Claude Code, Codex CLI, Cursor, and Cline

agentmako is a local-first MCP server that keeps code, routes, schema, diagnostics, project rules, findings, and freshness state queryable for Claude Code, Codex CLI, Cursor, Cline, and other MCP clients.

Read the docs →
local-first · no hosted service node ≥ 20 typed tools apache-2.0
~/projects/your-app — agentmako
reef engine

Reef keeps the map alive.

The index is only the start. Reef is Mako's durable fact and finding layer: it remembers what changed, what failed, what was reviewed, and which evidence is fresh enough for an agent to trust.

01

Durable facts

Routes, imports, symbols, schema, instructions, and rule-pack hits become typed local evidence.

02

Findings and acks

Known issues, cross-file risks, open loops, and reviewed decisions survive across sessions.

03

Background freshness

The daemon watches edits, refreshes scoped diagnostics, and labels stale or contradicted evidence.

04

Agent workflows

context_packet, file_preflight, and reef_diff_impact turn Reef into one-call guidance.

side-by-side

Vanilla agent vs. mako-backed agent.

Same model, same prompts. The difference is the context layer underneath.

~ vs.diff — same prompt, two agents vanilla / mako
files
−greps and guesses
+queries an indexed graph
freshness
−assumes the first read is still true
+labels freshness and refreshes changed files in the background
memory
−findings die with the chat
+reef remembers findings, acks, runs, and conventions
location
−hosted vector store you can't inspect
+local sqlite, no hosted service
quickstart

Three commands. No hosted service.

Requires Node ≥ 20. Everything else runs locally — index, store, MCP transport.

~ — install agentmako (≈ 60 seconds) node ≥ 20
01
$npm install -g agentmako added agentmako@0.3.0 · 1 binary in $PATH
02
$cd /path/to/your/project $agentmako connect . --no-db indexed 482 files · 6,213 symbols · 38 routes reef snapshot ready · 0 open loops
03
$agentmako doctor config · api service · index · all green
04
$agentmako mcp stdio MCP server ready · listening on agent
integrations

Drop into any MCP client.

Same stdio config everywhere. Tools surface as mcp__mako-ai__*.

Claude Code
CLI · plugin
$ claude mcp add mako-ai \
    agentmako mcp
  • + /mako-ai plugin with 8 bundled skills
  • config: ~/.claude.json
  • verify: /mcp inside Claude Code
Claude Code MCP setup guide ↗
Codex CLI
TOML
; ~/.codex/config.toml
[mcp_servers.mako-ai]
command = "agentmako"
args    = ["mcp"]
  • works with codex + codex exec
  • stdio transport, zero extra config
  • tools surface as mako-ai__*
Codex CLI MCP setup guide ↗
Cursor
JSON · per-project
// .cursor/mcp.json
{
  "mcpServers": {
    "mako-ai": { "command": "agentmako",
      "args": ["mcp"] }
  }
}
  • Composer (⌘ I) + Agent mode
  • Project Rules to enforce Mako-first
  • or global ~/.cursor/mcp.json
Cursor MCP setup guide ↗
agents.md

Teach your agent the playbook.

Tools are only as useful as the agent's choice to call them. Drop our CLAUDE.md / AGENTS.md into your project root so the agent knows when to reach for Mako.

CLAUDE.md · AGENTS.md — preview current template 
# Mako MCP Usage

This file is the short agent-facing guide for projects that have Mako
registered as `mako-ai`:

```json
{
  "mako-ai": {
    "command": "agentmako",
    "args": ["mcp"]
  }
}
```

In Claude Code, tools usually appear as `mcp__mako-ai__<toolName>`. Examples below use bare names. For complete CLI/tool details, use `agentmako --help`, `agentmako tool --help`, MCP `tools/list`, or `mako_help`.

## Default Workflow

When unsure, call `mako_help` first. It returns an ordered recipe with suggested args and verification steps.

For broad or vague work, start with `context_packet` using `mode: "explore" | "plan" | "implement" | "review"`. Before editing a risky file, call `file_preflight`. Mid-edit or before review, call `reef_diff_impact` on changed files.

## Freshness

In long-running MCP sessions, Mako's watcher refreshes changed files and runs scoped diagnostics in the background. Normal edits should not require manual `project_index_refresh` or `diagnostic_refresh`.

If Mako and live files/tests disagree, trust live reads and test output, then refresh Mako if indexed context should catch up.
drops into your project root as CLAUDE.md or AGENTS.md
CLAUDE.md AGENTS.md
tool catalog

Typed tools. One MCP server.

Six clusters cover the whole agent loop — from "what's relevant?" to "did it ship?".

Context
Ranked context for vague tasks.
mako_help context_packet
Code intel
Repo map, AST search, import graph.
cross_search imports_impact
Reef Engine
Durable facts, findings, freshness.
file_preflight reef_diff_impact
Database
Live Postgres / Supabase awareness.
db_table_schema db_rls
Trace & flow
Routes, RPCs, auth paths, graph walks.
route_trace auth_path
Diagnostics
TSC, ESLint, Biome, pre-commit gates.
lint_files git_precommit_check
Browse the full catalog →
common questions

Quick answers about MCP, Mako, and your agent.

The most-asked questions from devs first wiring up an MCP server. Full bank lives at the agentmako FAQ.

What is an MCP server?

An open standard from Anthropic that lets AI coding agents call external tools over a stdio interface. An MCP server exposes tools — search, schema, diagnostics — that any compliant client (Claude Code, Codex CLI, Cursor, Cline) can call.

More on MCP →

Why does my agent rediscover the codebase every prompt?

It has no memory across turns. Each prompt it greps and re-reads files it touched 30 seconds ago. Mako indexes the repo once and serves typed context packets on demand.

What is a context packet? →

Is it free? Does it send my code anywhere?

Free, Apache-2.0, zero telemetry. Everything runs locally; data sits in .mako-ai/ under each project. Outbound traffic only when you wire a model provider or connect Postgres / Supabase.

Privacy policy →

How is this different from a vector RAG database?

RAG fuzzy-matches text chunks. Mako stores a typed graph — files, symbols, routes, schema, RLS — and does deterministic lookups. Lower tokens, higher accuracy when fresh, and freshness itself is labeled.

What is the Reef Engine? →

How do I get my agent to actually use Mako?

Tools are only as useful as the agent's choice to call them. Drop the CLAUDE.md / AGENTS.md template into your project root — it tells the agent when to reach for which Mako tool. Without it, agents default to grep.

Read the full FAQ →