Open Source·Claude Code·Codex·Cursor·Gemini

Project management for the agentic engineering era.

mymir stores tasks, decisions, and dependencies as a project graph. Claude Code, Cursor, Codex, and Gemini use that graph in every session, so you stop re-explaining what was built and what comes next.

Get Started See how it works

AGPL v3·Runs Locally·MCP Native·6 Tools, 4 Agents

claude

brainstorm

The Problem

Coding agents can write the code. The hard part is giving them the project context behind it. When tasks live in flat lists, chat threads, and old notes, every session turns into rediscovery: what was planned, what was decided, what changed, and what depends on what.

The project graph should carry the context forward.

How It Works

From idea to implementation

mymir manages the full project lifecycle. Brainstorm the idea, decompose it into tasks with dependency edges, figure out what is ready next, and hand your coding agent the exact context it needs for whatever it is building right now.

Brainstorm

Describe your idea in plain language. A specialized agent explores it with you, scoping features, surfacing edge cases, and shaping the vision before a single task exists.

Decompose

Break the idea into implementable tasks with typed dependency edges. Each task gets a spec, category, and position in the context network.

Refine

Sharpen specs with acceptance criteria, architectural decisions, and file paths. Every refinement is captured in the context network for downstream tasks.

Plan

Plannable tasks get implementation blueprints with build sequences. Your agent receives upstream decisions, prerequisite specs, and related work automatically.

Execute

Ready tasks get full execution context: upstream decisions, file paths, and acceptance criteria. Your agent gets the constraints before it starts changing code.

Track

Execution records capture what was built, what changed, and why. Downstream tasks inherit this knowledge. No manual handoff, no context lost between sessions.

Core Primitives

Two primitives. One project graph.

Instead of flat task lists and scattered notes, mymir stores project context as structured data agents can query directly. Solo or in a team, people and agents work from the same project state.

Core Concept

Context Network

A project graph that captures not just what was built, but why decisions were made, what was tried and abandoned, and how different parts of the codebase relate to each other.

Tasks with specs, acceptance criteria, and status lifecycle

Typed dependency edges between tasks (depends_on, relates_to)

Decisions captured at the point they're made

Execution records that document what was built and why

File paths and implementation details per task

Core Concept

Context Retrieval Interface

The layer that lets agents query and use project knowledge at the right moment, so each session starts with the story so far.

Stage aware. Plannable tasks get specs, ready tasks get full execution context

Upstream decisions, downstream dependencies, file paths, and acceptance criteria bundled automatically into the context

Compact context blocks your agents reason from directly

Critical path analysis to find unblocked tasks on the shortest path

Capabilities

How it ships

6 MCP Tools

project, task, edge, query, context, analyze. Native Model Context Protocol.

4 Specialized Agents

Brainstorm scopes new ideas. Onboarding maps existing codebases. Decompose creates the task graph. Manage tracks progress.

Automatic Invocation

The /mymir skill detects project intent. Your agent knows when to use context.

Works with your CLI

Native plugins for Claude Code, Codex, Cursor, and Gemini CLI. Structure and Graph views in the browser.

The Workspace

Two views of the same project.

A workspace where your task list lives alongside full detail views. Refine specs, track progress, and review execution records, all without switching context.

StructureGraph

Progress29%

4 done2 active6 planned14 total

Foundation3/3

Database SchemaDONE

Auth & RBACDONE

CI/CD PipelineDONE

Data Pipeline1/3

Ingestion APIDONE

Stream ProcessorIN PROGRESS

Metrics StorePLANNED

Query & Alerting0/3

Query EnginePLANNED

Alert Rules EnginePLANNED

Notification SvcPLANNED

API & Realtime0/2

REST APIIN PROGRESS

WebSocket ServerPLANNED

Frontend0/2

Dashboard UIDRAFT

Alert Config UIDRAFT

Quality0/1

Load TestingPLANNED

Task/Metrics Platform

Stream Processor

Data Pipeline

kafkastreaming

DRAFT

PLANNED

IN PROGRESS

DONE

Description

Implement a stream processing pipeline that consumes events from the Ingestion API, applies transformations and aggregations, and writes to the Metrics Store.

Acceptance Criteria

Consume events from Ingestion API message queue

Apply configurable window aggregations (1m, 5m, 1h)

Write aggregated metrics to Metrics Store

Handle backpressure and dead letter queue

Decisions

Use Kafka Streams over Flink for lower operational overhead

Apr 06

Tumbling windows instead of sliding. Simpler semantics, sufficient for dashboard resolution

Apr 05

DLQ writes to separate Kafka topic, not PostgreSQL. Avoids coupling to main DB

Apr 06

Relationships

depends_onIngestion APIdone

blocksMetrics Storeplanned

blocksQuery Engineplanned

Files

src/services/stream/

src/services/stream/consumer.ts

src/services/stream/aggregator.ts

src/services/stream/dlq.ts

src/config/kafka.ts

Get Started

Up and running in minutes.

Local setupClone, configure, and serve mymir on your machine

setup

$ git clone [email protected]:FrkAk/mymir.git

$ cd mymir && bun install

$ cp .env.local.example .env.local

↳ set DATABASE_URL + BETTER_AUTH_SECRET (no API keys needed)

$ bun run db:setup

$ bun run dev

↳ web UI + plugin endpoint at http://localhost:3000

Install for your CLIPick the one you use. Each plugin is vendor-native.

Plugin via marketplace · OAuth on first /mcp

claude code plugin

$ claude plugin marketplace add ./plugins/claude-code

$ claude plugin install mymir@mymir-local

// inside claude code

› /mcp

↳ select mymir and sign in

Every install loads the Mymir MCP server, the /mymir skill, and the brainstorm, onboarding, decompose, and manage agents, auto-activated by description. One-time setup per machine.

FAQ

Common questions from developers.

What is mymir?

mymir is open-source project management for coding agents. It stores tasks, decisions, dependencies, acceptance criteria, and execution records as a project graph.

Which coding agents does mymir support?

mymir ships MCP plugins for Claude Code, Cursor, Codex, and Gemini CLI, plus a browser-based Structure view and Graph view.

How does mymir help agents keep context?

mymir returns context bundles scoped to the task lifecycle. Draft tasks pull specs and prerequisites, ready tasks pull implementation plans and upstream records, and completed tasks write execution records for downstream work.

Can teams use mymir?

Yes. Multiple people and agents can collaborate on the same project across sessions while the shared project graph keeps scope, plans, decisions, and progress in one place.

Is mymir self-hosted?

Yes. mymir is AGPL licensed, self-hosted, and runs on Postgres. A hosted version is planned for teams that want collaboration without running the infrastructure.

Open Source, Forever

Open source for the agentic engineering era.

Everyone should have access to tools that help them build better things. You can run mymir yourself. Always free, always will be. A hosted version is coming for teams who want to collaborate without the setup.

Star on GitHub Read the docs

AGPL v3Runs LocallyMCP NativeClaude CodeCodexCursorGemini CLI