<body> 1. cd to worktree 2. Read CLAUDE.md and relevant docs 3. Implement requirements 4. Commit with descriptive message 5. Push and create PR with "Closes #N" 6. Return PR URL ``` ## Milestone Priority 1. Foundation 2. Eval Setup (parallel) 3. Strava Integration 4. GraphHopper Integration 5. Route Synthesis 6. Place Search, Water Stops, Climb Integration, Weather Integration 7. Ride Preparation, Narrative Research 8. Research Quality, Route Refinement 9. Evaluation Framework ## Output **Status:** ``` ## Project Status ### Foundation (1/2) - [x] #2 SDK project structure - [ ] #3 Checkpoint system - READY ### Ready (2) - #3 Checkpoint system (Foundation) - #4 Promptfoo setup (Eval Setup) ``` **Work:** ``` ## Parallel Work Spawning: - #3 Checkpoint system → worktree created, agent launched - #4 Promptfoo setup → worktree created, agent launched Monitor with /tasks. PRs appear for review. ``` --- > Converted and distributed by [TomeVault](https://tomevault.io/claim/bendrucker) — claim your Tome and manage your conversions. <!-- tomevault:4.0:skill

# Bendrucker Route Agent Project > Project Management - Skill: `tomevault-io/bendrucker-route-agent-project` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add tomevault-io/bendrucker-route-agent-project` - Raw SKILL.md: https://api.skillmd.com/api/skills/tomevault-io/bendrucker-route-agent-project/raw - Safety review: pending (external: skill-scanner PASS, skillspector PASS) - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: AI & ML - Author: tomevault-io (https://skillmd.com/u/tomevault-io) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/tomevault-io/bendrucker-route-agent-project --- # Project Management Assess project status, find ready tasks, and orchestrate parallel work. ## Commands | Command | Action | |---------|--------| | `/project` | Show milestone progress and ready tasks | | `/project next` | Recommend single next task | | `/project work` | Spawn sub-agents for ready tasks (parallel) | | `/project work #N` | Work on specific issue | | `/project cleanup` | Remove worktrees for merged PRs | ## Quick Start ```bash # Milestone progress gh api repos/bendrucker/route-agent/milestones --jq '.[] | "\(.title): \(.closed_issues)/\(.open_issues + .closed_issues)"' # Open issues gh issue list --repo bendrucker/route-agent --state open --json number,title,milestone,body --limit 100 ``` ## Finding Ready Tasks **Ready** = No "Depends On" section, OR all dependencies closed. Parse from issue body: ```markdown ## Depends On - Issue title here ``` ## Parallel Work (`/project work`) 1. **Find ready tasks** - Parse dependencies, filter unblocked 2. **Create worktrees** - One per issue in `./worktrees//` (inside project, sandbox-allowed) 3. **Spawn sub-agents** - Each works in background, creates PR 4. **Report** - List spawned agents for monitoring See [references/worktrees.md](references/worktrees.md) for details. ### Sub-Agent Prompt ``` Work on issue #N in worktree at . Issue: <body> 1. cd to worktree 2. Read CLAUDE.md and relevant docs 3. Implement requirements 4. Commit with descriptive message 5. Push and create PR with "Closes #N" 6. Return PR URL ``` ## Milestone Priority 1. Foundation 2. Eval Setup (parallel) 3. Strava Integration 4. GraphHopper Integration 5. Route Synthesis 6. Place Search, Water Stops, Climb Integration, Weather Integration 7. Ride Preparation, Narrative Research 8. Research Quality, Route Refinement 9. Evaluation Framework ## Output **Status:** ``` ## Project Status ### Foundation (1/2) - [x] #2 SDK project structure - [ ] #3 Checkpoint system - READY ### Ready (2) - #3 Checkpoint system (Foundation) - #4 Promptfoo setup (Eval Setup) ``` **Work:** ``` ## Parallel Work Spawning: - #3 Checkpoint system → worktree created, agent launched - #4 Promptfoo setup → worktree created, agent launched Monitor with /tasks. PRs appear for review. ``` --- > Converted and distributed by [TomeVault](https://tomevault.io/claim/bendrucker) — claim your Tome and manage your conversions.