AI Dev Tasks
https://github.com/snarktank/ai-dev-tasks- Category
- AI Agents
- Rank
- No. 990Tools index
Previous survey · No. 975 ·
- Pricing
- Open Source
- Type
- TOOL
- Builder
- snarktank
- GitHub
- 7.8k stars
- Date
About
Simple task-management system for managing AI dev agents — track what your agents are doing, when they stalled, and what is next.
What it does
AI Dev Tasks is a pair of reusable Markdown prompts for staged feature development. One prompt turns a feature idea into a clarified product requirements document. The other converts requirements into parent tasks, pauses for approval, then expands them into actionable subtasks. Developers review each step and mark completed work in the generated checklist.
Why it's ranked here
The workflow is small but thoughtfully constrained. It asks critical questions before planning, separates high-level approval from detailed decomposition, and targets instructions at junior developers. Its value comes from disciplined prompting rather than software automation, so results still depend on the coding assistant and the context you provide.
What's good
The prompts create useful human checkpoints. Requirements include goals, user stories, scope exclusions, success metrics, and open questions. Task generation first exposes the parent plan, then waits for confirmation before adding subtasks. The final checklist identifies relevant code and test areas, and requires completion marks after every subtask.
Tradeoffs
This is a prescribed Markdown workflow, not an agent runtime or project-management service. It does not execute tasks, inspect progress automatically, or verify completed code. Task guidance assumes feature branches and suggests Jest conventions, which may require adaptation. The PRD prompt also caps clarification at a small set of critical questions, leaving deeper discovery to the user.
How to use it well
Use it for medium or large features where a developer wants explicit scope, a reviewable plan, and small implementation checkpoints. Give the assistant concrete product context, answer its clarification questions, inspect parent tasks carefully, then approve detailed decomposition. Adapt branch and testing guidance to your repository. Choose another tool for automated orchestration, live agent monitoring, or execution control.
Technical notes+
README.md documents a clone-or-download workflow built around create-prd.md and generate-tasks.md. create-prd.md requires 3 to 5 essential multiple-choice clarification questions, then writes /tasks/prd-[feature-name].md with nine prescribed sections. generate-tasks.md creates /tasks/tasks-[feature-name].md, inserts task 0.0 for feature-branch creation unless declined, pauses after parent tasks until the user replies “Go,” then adds subtasks, relevant files, testing notes, and checkbox progress instructions. LICENSE contains the Apache License 2.0. The supplied repository text shows Markdown prompt assets rather than a package manifest, executable, API, or runtime implementation.
Observed
- License
- Apache License 2.0
- Primary format
- Markdown prompt documents
- Install surface
- Clone or download the repository and place the Markdown prompts where an AI assistant can access them
- Interface
- Structured prompts consumed through AI-powered IDEs and CLIs
- Platform support
- Designed for any AI coding assistant; the README names Amp, Claude Code, and Windsurf
- Repository structure
- Two operational prompt documents, plus README documentation and a license
Read from README.md, LICENSE, create-prd.md, generate-tasks.md.
What it can do
Track AI agent activities
AI agent execution data → Real-time status of what agents are doing
Detect when agents stall
Agent performance metrics and execution logs → Alerts and notifications when agents stop progressing
Identify next tasks for agents
Current agent status and task queue → Prioritized list of upcoming tasks
Create development tasks
Task descriptions and requirements → Structured task entries in the system
Monitor task progress
Task execution data from AI agents → Progress reports and completion status
Organize agent workflows
Multiple AI agent tasks and dependencies → Structured workflow with task sequences
Tags
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.