The backlog your AI coding agent reads, picks up, and closes itself.

Agent-native model

Every ticket carries an agent slug (which Claude Code agent persona handles it) and a model field (opus/sonnet/haiku). The backlog encodes how work gets done, not just what exists.

MCP server

runloq-mcp lets any MCP-capable agent read, create, update and close tickets without a UI. Works with Claude Code, Cursor, and Codex out of the box.

Local-first, offline-first

One SQLite file. No cloud dependency. No seat pricing. Private, instant, and works on a plane. pipx install runloq and you’re done.

Scheduled & recurring tickets

Auto-spawn the next iteration on close. Scheduled tickets surface at the right time via the SessionStart hook — your agent picks them up automatically.

What makes runloq different from every other issue tracker?

Every other issue tracker was designed for humans. runloq was designed for AI agents.

When you open Linear or GitHub Issues, you see a list of tasks a person will eventually click through and work on. runloq’s backlog is executable: an agent reads it, knows exactly which specialist picks up the ticket and which LLM tier runs it, executes the work, and closes it with a full audit trail — autonomously.

Install in 60 seconds

pipx install runloq
cd your-project
runloq init          # scaffolds runloq.config.toml
runloq serve         # starts the React dashboard

Then connect your agent

# MCP server (Claude Code / Cursor / Codex)
runloq-mcp

# Claude Code integration kit — one-command install
bash integrations/claude-code/install.sh

Who is runloq for?

Solo developers using Claude Code or Cursor who want a structured backlog their agent can actually drive — not a GitHub Issues board they have to manage manually.

AI-first teams (early) building agent-driven workflows where task routing, model selection, and audit trails matter.

If you paste tasks into a chat window and hope the agent remembers context between sessions, runloq is the structure you’re missing.

Compare to alternatives

Feature	runloq	GitHub Issues	Linear
agent field (which AI specialist)	✅	❌	❌
model field (opus/sonnet/haiku)	✅	❌	❌
MCP server	✅	❌	❌
Local-first SQLite	✅	❌	❌
Scheduled + recurring tickets	✅	❌	Partial
Append-only audit log	✅	Partial	Partial
Free forever (local)	✅	✅	❌

→ Why not GitHub Issues? · Why not Linear?