project-implement
Start a new project from a request — explores, interviews, plans, creates milestones, and implements with agent swarms. Use this skill whenever the user wants to build something from scratch, implement a feature end-to-end, or kick off a multi-phase project. Triggers on phrases like "build me...", "implement...", "create a project for...", "start a new project", or any substantial implementation request that would benefit from structured orchestration.
What this skill does
# Project Runner: Full Implementation Orchestration
You are now operating as the **Project Runner Orchestrator**. You will take the user's request
and drive it from zero to 100% complete through a multi-phase workflow using specialized agents.
The user's request: **"$ARGUMENTS"**
Follow every phase below in exact order. Do NOT skip phases. Do NOT move to the next phase until
the current phase is fully complete. This is expected to be a long-running process.
---
## PHASE 1: Project Setup
Create the project workspace. Do this directly (no agent needed).
1. **Check for existing projects:** Use Glob to find `projects/*/` directories. Find the highest
existing project number. If no projects exist, start at 001.
2. **Generate the project slug:** Take the user's request, extract 3-5 key words, lowercase them,
join with hyphens. Example: "build a REST API for a todo app" → "rest-api-todo-app"
3. **Create the project directory:**
```
projects/NNN-slug/
```
4. **Write `request.md`:** Save the user's full request exactly as provided:
```markdown
# Project Request
$ARGUMENTS
```
5. **Write `status.md`:** Initialize the project status tracker:
```markdown
# Project Status
## Current Phase
exploration
## Progress
- [x] Phase 1: Project Setup
- [ ] Phase 2: Exploration
- [ ] Phase 3: Interview
- [ ] Phase 4: Planning
- [ ] Phase 5: Milestone Creation
- [ ] Phase 6: Milestone Verification
- [ ] Phase 7: Task Creation
- [ ] Phase 8-10: Implementation Swarm
## Project Path
projects/NNN-slug/
```
6. **Tell the user** the project has been created and what happens next.
**IMPORTANT — Resumability:** Before creating a new project, check if the user is referencing
an existing project. If `status.md` exists in a project folder and shows an incomplete phase,
ask the user if they want to resume that project instead of starting a new one.
---
## PHASE 2: Exploration
Launch the **pr-explorer** agent to explore the repository.
1. Use the Task tool to spawn the **pr-explorer** agent with this prompt:
```
You are the exploration agent for a project-runner workflow.
Project folder: projects/NNN-slug/
Read the request from: projects/NNN-slug/request.md
Explore the repository thoroughly and write your findings to:
projects/NNN-slug/initial-findings.md
Follow the exploration checklist in your system prompt. Be thorough.
```
2. Wait for the explorer to complete.
3. Update `status.md`: mark Phase 2 complete, set current phase to "interview".
4. Tell the user: "Exploration complete. Starting requirements interview..."
---
## PHASE 3: Interview
Launch the **pr-interviewer** agent as a **FOREGROUND** agent (it needs AskUserQuestion).
1. Use the Task tool to spawn the **pr-interviewer** agent as a foreground agent with this prompt:
```
You are the interview agent for a project-runner workflow.
Project folder: projects/NNN-slug/
Read the request from: projects/NNN-slug/request.md
Read the findings from: projects/NNN-slug/initial-findings.md
Conduct a thorough requirements interview with the user. Write the transcript to:
projects/NNN-slug/interview.md
IMPORTANT: Batch your questions in groups of 4 using a single AskUserQuestion call per
batch. Plan 8-16 questions total, then send them 4 at a time. Adapt between batches
based on the user's answers. Every question must include "Skip this question" and
"Stop interview" options. Mark one option per question as "(Recommended)".
```
2. Wait for the interviewer to complete.
3. Update `status.md`: mark Phase 3 complete, set current phase to "planning".
4. Tell the user: "Interview complete. Creating the high-level plan..."
---
## PHASE 4: Planning
Launch the **pr-planner** agent to create the strategic plan.
1. Use the Task tool to spawn the **pr-planner** agent with this prompt:
```
You are the planning agent for a project-runner workflow.
Project folder: projects/NNN-slug/
Read all context from:
- projects/NNN-slug/request.md
- projects/NNN-slug/initial-findings.md
- projects/NNN-slug/interview.md
Create a comprehensive high-level plan and write it to:
projects/NNN-slug/plan.md
Include: architecture, technology decisions, component breakdown, data model,
testing strategy, deployment approach. Do NOT create milestones or tasks yet.
```
2. Wait for the planner to complete.
3. Update `status.md`: mark Phase 4 complete, set current phase to "milestone-creation".
4. Tell the user: "Plan created. Breaking down into milestones and tasks..."
---
## PHASE 5: Milestone Creation
Launch the **pr-milestone-creator** agent to decompose the plan into milestones.
1. Use the Task tool to spawn the **pr-milestone-creator** agent with this prompt:
```
You are the milestone creator agent for a project-runner workflow.
Project folder: projects/NNN-slug/
Read all context from:
- projects/NNN-slug/request.md
- projects/NNN-slug/initial-findings.md
- projects/NNN-slug/interview.md
- projects/NNN-slug/plan.md
Create individual milestone files in the project folder:
- projects/NNN-slug/001-milestone-slug.md
- projects/NNN-slug/002-milestone-slug.md
- etc.
Each milestone file contains all its tasks inline. Follow the exact format from your
system prompt. Be exhaustive — every requirement must be covered. Leave no gaps.
Everything must be production-ready.
```
2. Wait for the milestone creator to complete.
3. Update `status.md`: mark Phase 5 complete, set current phase to "milestone-verification".
4. Tell the user: "Milestones created. Verifying completeness..."
---
## PHASE 6: Milestone Verification (loops until pass)
Launch the **pr-milestone-verifier** agent. If it finds issues, it fixes them and re-verifies.
1. Use the Task tool to spawn the **pr-milestone-verifier** agent with this prompt:
```
You are the milestone verifier agent for a project-runner workflow.
Project folder: projects/NNN-slug/
Read all project files including all milestone files (NNN-*.md pattern).
Verify completeness, correctness, and consistency. If you find ANY issues:
1. Fix them directly in the milestone files
2. Re-run your full verification checklist
3. Repeat until ALL checks pass
Write your verification report to: projects/NNN-slug/verification.md
The final status in verification.md MUST be "PASS".
```
2. Wait for the verifier to complete.
3. Read `verification.md` and confirm it says "PASS". If it doesn't, re-launch the verifier.
4. Update `status.md`: mark Phase 6 complete, set current phase to "task-creation".
5. Tell the user: "Milestones verified. Creating task list..."
---
## PHASE 7: Task Creation
Parse the milestone files and create tasks using the TaskCreate tool. Do this directly.
1. **Read all milestone files** from the project folder (use Glob for `projects/NNN-slug/[0-9]*.md`).
2. **Parse each milestone** to extract tasks. For each task found:
- Call `TaskCreate` with:
- `subject`: "Task N.M: [Task Title]"
- `description`: The full task description and acceptance criteria
- `activeForm`: "[Task Title] — implementing"
3. **Set up dependencies** using `TaskUpdate`:
- For each task that has dependencies, use `addBlockedBy` to reference the blocking task IDs
- Map milestone task IDs (e.g., "Task 2.3") to the TaskCreate IDs
4. **Update `status.md`:**
- Record total task count
- Mark Phase 7 complete
- Set current phase to "implementation"
5. Tell the user: "N tasks created with dependencies. Starting the implementation swarm..."
---
## PHASE 8-10: Implementation Swarm
This is the main implementation loop. Choose the swarm mode:
### Check for Agent Teams Support
Check if the environment has Agent Teams enabled. You can tell if you have access to the
TeammateTool or if the user has `CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS` enablRelated in AI Agents
skill-development
IncludedComprehensive meta-skill for creating, managing, validating, auditing, and distributing Claude Code skills and slash commands (unified in v2.1.3+). Provides skill templates, creation workflows, validation patterns, audit checklists, naming conventions, YAML frontmatter guidance, progressive disclosure examples, and best practices lookup. Use when creating new skills, validating existing skills, auditing skill quality, understanding skill architecture, needing skill templates, learning about YAML frontmatter requirements, progressive disclosure patterns, tool restrictions (allowed-tools), skill composition, skill naming conventions, troubleshooting skill activation issues, creating custom slash commands, configuring command frontmatter, using command arguments ($ARGUMENTS, $1, $2), bash execution in commands, file references in commands, command namespacing, plugin commands, MCP slash commands, Skill tool configuration, or deciding between skills vs slash commands. Delegates to docs-management skill for official documentation.
reprompter
IncludedTransform messy prompts into well-structured, effective prompts — single or multi-agent. Use when: "reprompt", "reprompt this", "clean up this prompt", "structure my prompt", rough text needing XML tags and best practices, "reprompter teams", "repromptception", "run with quality", "smart run", "smart agents", multi-agent tasks, audits, parallel work, anything going to agent teams. Don't use when: simple Q&A, pure chat, immediate execution-only tasks. See "Don't Use When" section for details. Outputs: Structured XML/Markdown prompt, quality score (before/after), optional team brief + per-agent sub-prompts, agent team output files. Success criteria: Single mode quality score ≥ 7/10; Repromptception per-agent prompt quality score 8+/10; all required sections present, actionable and specific.
adaptive-compaction
IncludedAdaptive add-on policy and recovery layer that decides WHEN to compact, prune, snapshot, or fork -- replacing fixed-percent auto-compaction across Claude Code, Codex, and MCP-capable hosts. Trigger on auto-compact timing or damage: "when should I compact", "is it safe to compact now or start a fresh session", "auto-compact fires too early/mid-task", "switching to an unrelated task but the window still has space", "context rot", "answers get worse the longer the session runs", "the agent forgot the plan or my decisions after it summarized", "add a layer on top that manages context without changing the agent", raising autoCompactWindow to give the policy room, or installing/tuning a cross-tool compaction policy or PreCompact hook -- even when "compaction" is never said but the problem is context-window pressure or post-summarization memory loss. Do NOT use to summarize a conversation, build RAG, write a summarization prompt (decides WHEN not HOW), or answer max-context-length trivia.
agent-skill-creator
IncludedCreate cross-platform agent skills from workflow descriptions. Activates when users ask to create an agent, automate a repetitive workflow, create a custom skill, or need advanced agent creation. Triggers on phrases like create agent for, automate workflow, create skill for, every day I have to, daily I need to, turn process into agent, need to automate, create a cross-platform skill, validate this skill, export this skill, migrate this skill. Supports single skills, multi-agent suites, transcript processing, template-based creation, interactive configuration, cross-platform export, and spec validation.
llm-wiki
IncludedUse when building or maintaining a persistent personal knowledge base (second brain) in Obsidian where an LLM incrementally ingests sources, updates entity/concept pages, maintains cross-references, and keeps a synthesis current. Triggers include "second brain", "Obsidian wiki", "personal knowledge management", "ingest this paper/article/book", "build a research wiki", "compound knowledge", "Memex", or whenever the user wants knowledge to accumulate across sessions instead of being re-derived by RAG on every query.
skill-master
IncludedAgent Skills authoring, evaluation, and optimization. Create, edit, validate, benchmark, and improve skills following the agentskills.io specification. Use when designing SKILL.md files, structuring skill folders (references, scripts, assets), ingesting external documentation into skills, running trigger evals, benchmarking skill quality, optimizing descriptions, or performing blind A/B comparisons. Keywords: agentskills.io, SKILL.md, skill authoring, eval, benchmark, trigger optimization.