Browse / Software development / agentic-ai-projects-task-manager
agentic-ai-projects-task-manager
savedpixel
Local-first workspace for planning, coordinating, and verifying AI-assisted software tasks.
Open savedpixel/agentic-ai-projects-task-manager → Does not match the four rules.
aielectronlocal-firstproject-managementpwatask-manager
Works with any agent
ships AGENTS.md, the cross-tool instruction file.
In its own words
An open-source, local-first workspace for planning, coordinating, and verifying AI-assisted software tasks.
It turns Markdown task plans into a focused operational interface for queues, decisions, lifecycle transitions, agent activity, project registries, lessons, observations, and user verification. Your task data stays in project folders you choose; the application has no hosted account requirement and no built-in telemetry.
What it does
- Aggregates task-system folders from multiple local projects.
- Presents queue, plan-detail, question, verification, activity, agent, lesson, observation, project, and settings views.
- Writes lifecycle changes and user-verification feedback back to the project’s Markdown/JSON task records.
- Runs as a PWA, a local writable web app, or an unsigned Electron desktop build.
- Uses isolated public app data under
Agentic AI Projects Task Manager; it does not import another product’s registry automatically. - Works with any coding agent or human workflow that follows the documented task-folder contract.
This project may feel familiar to people who use Notion to organize work, but its scope is narrower: it is a filesystem-backed software-task workspace, not a collaborative document/database suite. It is not affiliated with, endorsed by, or feature-compatible with Notion Labs, Inc., and it does not claim Notion feature parity.
Quick start
Requirements: Node.js 20 or newer, npm, and Python 3.11 or newer.
Quoting 43% of the README. Read the rest on GitHub ↗
Read from the repository's git tree on 2026-08-17.