Skip to main content

A shared to-do list for AI agents. Dependency-aware task queue, zero dependencies, pure Python.

Project description

agentplan

Asana for AI agents — a task board that any AI tool can drive.

PyPI version License: MIT PyPI downloads Python versions OpenSSF Scorecard Tests

agentplan is Asana for AI agents — a shared task board that AI tools can use as the source of truth for work.

  • Persistent project + ticket queue
  • Dependency tracking with automatic unblocking
  • Atomic ticket claiming (safe for concurrent agents)
  • Web dashboard for visibility
  • Local-first SQLite storage
  • Built-in plugins for Claude Code and Codex

What AgentPlan Is

AgentPlan is the coordination layer for AI work.

Use it to:

  • create a project for a repo
  • break work into small tickets
  • track dependencies and progress
  • let agents claim the next unblocked task
  • watch the work move in the dashboard

AgentPlan is not the main execution engine. Claude Code, Codex, and other tools do the planning and implementation work. AgentPlan keeps the shared backlog, ticket state, and history in one place.

Start in 3 steps

# 1) Install
pip install agentplan

# On macOS (Homebrew Python):
pipx install agentplan

# 2) Connect your AI tool
agentplan setup claude    # Claude Code
agentplan setup codex     # Codex CLI

# 3) Tell your AI to plan
# In Claude Code: /agentplan:plan
# Or just say: "plan a new project for this repo"

Typical workflow

  1. Open Claude Code, Codex, or another supported AI tool.
  2. Ask it to create an AgentPlan project for the current repo.
  3. Let it break the work into tickets with dependencies.
  4. Have the AI tool work ticket-by-ticket using agentplan next and agentplan claim.
  5. Use the dashboard to monitor progress and inspect ticket state.

Short version: plan in the AI tool, track in AgentPlan, execute through the AI tool's own loop or scheduled workflow.

Quickstart

# Create a project linked to the current repo
agentplan create "Ship v1" --dir .

# Add a few tickets
agentplan ticket add ship-v1 "Set up database" --priority high
agentplan ticket add ship-v1 "Implement API" --priority high
agentplan ticket add ship-v1 "Write tests" --priority medium

# Add dependencies where needed
agentplan depend ship-v1 2 --on 1
agentplan depend ship-v1 3 --on 2

# See what's ready to work on
agentplan next ship-v1

# Claim the next unblocked ticket
agentplan claim ship-v1

# Mark it done
agentplan ticket done ship-v1 1

# Check progress
agentplan status ship-v1

AI tool setup

The setup command installs plugins from the pip package, so the AI tool can understand the AgentPlan workflow without cloning anything extra:

# Claude Code — registers as a local marketplace plugin
agentplan setup claude

# Codex CLI — copies the skill into ~/.codex/skills/
agentplan setup codex

After setup, restart the AI tool.

Claude Code flow

  • /agentplan:plan creates a project and tickets from the conversation
  • /agentplan:status shows project progress
  • /agentplan:loop generates the prompt/instructions for Claude's own loop or cron-driven workflow

Codex flow

  • use the installed agentplan skill
  • create a project, add tickets, and work from agentplan next / agentplan claim
  • keep AgentPlan as the shared backlog if Claude and Codex are both working on the same repo

Core CLI commands

Project lifecycle

Command Description
agentplan create Create a project (with optional --ticket flags)
agentplan list List all projects
agentplan status <project> Show project progress and ticket states
agentplan close <project> Close a completed project
agentplan archive <project> Archive a project
agentplan remove <project> Permanently remove a project

Ticket workflow

Command Description
agentplan ticket add <project> "title" Add a ticket
agentplan ticket list <project> List tickets
agentplan ticket done <project> <num> Mark ticket done
agentplan ticket skip <project> <num> Skip a ticket
agentplan ticket block <project> <num> Block a ticket
agentplan ticket fail <project> <num> Mark ticket failed
agentplan ticket edit <project> <num> Edit ticket details
agentplan next <project> Show next unblocked tickets
agentplan claim <project> Atomically claim the next unblocked ticket
agentplan search <query> Search tickets across all projects

Dependencies, logs, and notes

Command Description
agentplan depend <project> <ticket> --on <dep> Add dependency
agentplan undepend <project> <ticket> --on <dep> Remove dependency
agentplan log <project> Add a log entry
agentplan note <project> Set a note on project or ticket
agentplan attach <project> Attach a file or URL
agentplan history <project> <ticket> Show ticket state transitions

Utilities

Command Description
agentplan setup [claude|codex] Install AI tool plugin
agentplan dashboard Launch web dashboard
agentplan completion Print shell completion script

Dashboard

# install with dashboard extras (requires Flask)
pip install "agentplan[dashboard]"

agentplan dashboard
# or run in background:
agentplan dashboard --background

Open http://127.0.0.1:5001 to view projects, ticket board, and activity. Create and edit tickets directly from the UI.

Advanced workflows

AgentPlan also includes advanced and power-user surfaces for orchestration, CI, and other automation-heavy workflows. Those are intentionally de-emphasized in the main UX.

If you are evaluating AgentPlan for the first time, start with:

  • project creation
  • tickets and dependencies
  • next and claim
  • dashboard visibility
  • Claude/Codex plugin flow

Security + docs

  • Security policy: docs/security/security.md
  • Privacy: docs/security/privacy.md
  • Canonical docs alignment baseline: docs/docs-alignment-canonical-story.md

License

MIT

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

agentplan-0.8.7.tar.gz (96.2 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

agentplan-0.8.7-py3-none-any.whl (102.0 kB view details)

Uploaded Python 3

File details

Details for the file agentplan-0.8.7.tar.gz.

File metadata

  • Download URL: agentplan-0.8.7.tar.gz
  • Upload date:
  • Size: 96.2 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.7

File hashes

Hashes for agentplan-0.8.7.tar.gz
Algorithm Hash digest
SHA256 6bc563204cd98d4d700b0e78ee895b22dc5575a4d59c4e4886b03219c18cb2a7
MD5 1341ea86bdd77ea6328cf12fb4b6ac8c
BLAKE2b-256 679040fb08242d1c3df9ae7d9364f99e824b3bfb80dce226ff28c3eb8ca8ad0b

See more details on using hashes here.

Provenance

The following attestation bundles were made for agentplan-0.8.7.tar.gz:

Publisher: publish.yml on fraction12/agentplan

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file agentplan-0.8.7-py3-none-any.whl.

File metadata

  • Download URL: agentplan-0.8.7-py3-none-any.whl
  • Upload date:
  • Size: 102.0 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.7

File hashes

Hashes for agentplan-0.8.7-py3-none-any.whl
Algorithm Hash digest
SHA256 0d16e15bd98a37794340ce897bb47f51d8cd74e4c1b49d6893e1e523d08d3b23
MD5 065973c701edde8538874c202f911ccb
BLAKE2b-256 413bc944248b1eaae20e9b3b172807e894cce6899aa0ab03706239b49da64e95

See more details on using hashes here.

Provenance

The following attestation bundles were made for agentplan-0.8.7-py3-none-any.whl:

Publisher: publish.yml on fraction12/agentplan

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page