GitHub Project
Manager
Invisible AI project manager for GitHub. Auto-discovers projects, creates rich issues, estimates work, syncs boards, and keeps your backlog alive.
$skillfish add johnefemer/skillfish/skills/github-project-manager
"Every AI coding agent can write code, but none of them manage the project around it. I wanted an invisible project manager that lives inside the AI agent — one that creates rich, AI-readable issues, keeps the board honest, breaks work down like a senior engineer, and estimates with the precision of someone who's actually read the codebase."
— John Efemer, creator of AI Coach
When to Use
Three workflows where the GitHub Project Manager shines.
Issue Creation
Create structured issues from PRDs, bug reports, or feature requests with AI-optimized templates.
Board Management
Sync GitHub Projects v2 boards, move issues through states, set priorities and estimates.
Sprint Planning
Plan sprints, triage backlogs, generate standup reports, and monitor epic health.
How It Works
Four steps from request to organized project board.
Discover
Auto-detects repo, project board, field IDs, and team members. Saves config to .github/project-config.json.
Create
Generates rich, AI-readable issues with scope, acceptance criteria, and technical notes from CLAUDE.md.
Organize
Adds to project board, sets Status/Priority/Size/Estimate, and links sub-issues to epics.
Track
Standup reports, epic health checks, sprint velocity, and estimation calibration to keep projects on track.
Issue Types
Six issue types, each with structured templates optimized for AI agents.
[Epic] Large Initiative
High-level objective spanning multiple tasks. Includes vision statement, success metrics, and sub-issue breakdown.
[Feature] New Capability
User-facing functionality. Includes acceptance criteria, technical approach, and UI/UX considerations.
[Task] Implementation Work
Concrete development task. Includes scope definition, implementation hints, and files to modify.
[Bug] Defect Report
Reproduction steps, expected vs actual behavior, and severity classification for efficient triage.
[Chore] Maintenance
Dependency updates, refactoring, CI fixes. Includes justification, risk assessment, and rollback plan.
[Spike] Research
Time-boxed investigation. Includes research questions, success criteria, and expected deliverables.
Advanced Workflows
Power features for teams that want full autopilot project management.
PRD-to-Issues Pipeline
Parse a PRD into epics and tasks automatically. Generates a full issue tree with estimates, priorities, and board placement.
Work Estimation Expert
Size issues by analyzing codebase complexity. Uses file count, dependency depth, and test coverage to calibrate story points.
Sprint Planning Autopilot
Propose sprint backlogs based on velocity, team capacity, and priority. Automatically balances load across assignees.
Epic Health Monitor
Track progress across sub-issues, flag blockers, calculate completion percentage, and project estimated delivery dates.
Sample Output
Here's what an epic breakdown looks like after the skill runs.
📋 Epic: [Epic] User Authentication Overhaul (#140)
Sub-Issues Created:
├── [Task] Migrate session store to Redis (#141) — S (2 pts)
├── [Task] Implement session timeout (#142) — M (3 pts)
├── [Task] Add refresh token rotation (#143) — M (3 pts)
├── [Task] Update auth middleware (#144) — S (2 pts)
└── [Task] Write integration tests (#145) — M (3 pts)
Total: 5 tasks, 13 story points
All added to project board with Status: Backlog Ready to try it?
Install the GitHub Project Manager skill and let AI handle your project board, issues, and sprint planning.
$skillfish add johnefemer/skillfish/skills/github-project-manager