execute
Execute tasks with velocity and quality. Use when ready to implement after clarity and prioritization are complete. This is the fourth system in the 5-system framework.
Best use case
execute is best used when you need a repeatable AI agent workflow instead of a one-off prompt.
Execute tasks with velocity and quality. Use when ready to implement after clarity and prioritization are complete. This is the fourth system in the 5-system framework.
Teams using execute should expect a more consistent output, faster repeated execution, less prompt rewriting.
When to use this skill
- You want a reusable workflow that can be run more than once with consistent structure.
When not to use this skill
- You only need a quick one-off answer and do not need a reusable workflow.
- You cannot install or maintain the underlying files, dependencies, or repository context.
Installation
Claude Code / Cursor / Codex
Manual Installation
- Download SKILL.md from GitHub
- Place it in
.claude/skills/4-execute/SKILL.mdinside your project - Restart your AI agent — it will auto-discover the skill
How execute Compares
| Feature / Agent | execute | Standard Approach |
|---|---|---|
| Platform Support | Not specified | Limited / Varies |
| Context Awareness | High | Baseline |
| Installation Complexity | Unknown | N/A |
Frequently Asked Questions
What does this skill do?
Execute tasks with velocity and quality. Use when ready to implement after clarity and prioritization are complete. This is the fourth system in the 5-system framework.
Where can I find the source code?
You can find the source code on GitHub using the link provided at the top of the page.
SKILL.md Source
# Execution System > **Purpose:** Do the work at high quality with velocity. > **When to trigger:** After clarity is established and priorities are set. ## Pre-Flight Checklist Before writing any code, verify: - [ ] **Clarity exists** - active-context.md is current with clear success criteria - [ ] **Task is prioritized** - This is the right thing to work on now - [ ] **Success criteria defined** - I know exactly what "done" looks like - [ ] **Boundaries set** - I know what NOT to do If any are missing → Return to appropriate system. ## Execution Protocol ### 1. Track Progress Visibly Use TodoWrite at the start: ``` 1. [ ] Step one of implementation 2. [ ] Step two of implementation 3. [ ] Validation step ``` Update status as you go. The user should always know where you are. ### 2. Make the Smallest Viable Change - Don't over-engineer - Don't add features that weren't requested - Don't refactor unrelated code - Don't add "nice to haves" - Solve the exact problem, nothing more ### 3. Validate Continuously After every significant change: ```bash npx tsc --noEmit # Catch type errors immediately ``` Don't let errors accumulate. Fix as you go. ### 4. One Thing at a Time - Complete one task before starting another - Don't context-switch mid-implementation - If blocked, acknowledge it rather than switching - Mark complete immediately when done ### 5. Parallel When Independent Use subagents (Task tool) for truly independent work: - Research while implementing - Multiple unrelated fixes - Exploration while writing But NOT for dependent steps - those must be sequential. ## Quality Gates Before marking any task complete: ### Code Quality - [ ] Types pass (`npx tsc --noEmit`) - [ ] No lint errors - [ ] Follows existing patterns in codebase - [ ] No unintended side effects ### Functionality - [ ] Works for the happy path - [ ] Edge cases considered - [ ] Error states handled - [ ] Loading states present (if UI) ### Integration - [ ] Doesn't break existing functionality - [ ] Works with existing data - [ ] API contracts maintained ### Documentation (if applicable) - [ ] Complex logic commented - [ ] API changes noted in SOURCE_OF_TRUTH.md - [ ] Breaking changes flagged ## Velocity Principles **Do:** - Edit existing files over creating new ones - Use existing patterns and utilities - Ask early if blocked (don't spin) - Ship incrementally - Commit working states frequently **Don't:** - Invent new patterns when existing ones work - Gold-plate solutions - Wait until everything is perfect - Make changes outside the task scope ## Output Tracking During execution, maintain: 1. **Todos** - Current task breakdown with status 2. **Active context** - Updated if scope changes 3. **Issues** - Any new problems discovered → Identity System ## Completion Criteria A task is ONLY complete when: 1. All acceptance criteria from clarity are met 2. All quality gates pass 3. Type check passes 4. Functionality verified 5. No regressions introduced ## Handling Blocks If you hit a wall during execution: 1. **STOP** - Don't keep trying the same failing approach 2. **DOCUMENT** - What was attempted, what failed 3. **TRANSITION** - Go to Reset System if blocked, or ask user Never mark something complete that isn't actually done. ## Transition After execution: - Success → Mark complete, update SOURCE_OF_TRUTH.md if significant - Partial success → Document what remains, continue or handoff - Failure → Proceed to **Reset System** - New issues discovered → Log to **Identity System** **Capture Learning:** If something important was learned (new pattern, gotcha, or insight), add one line to `learnings.md`. Keep it brief - this is memory, not documentation. --- *This is System 4 of 5: Clarity → Identity → Priority → Execution → Reset*
Related Skills
52-execute-refactor-150
[52] EXECUTE. Three-stage refactoring workflow: (1) iterative research of refactor/modularization options, (2) plan + risk/edge-case analysis + Scope150 validation, then implement with tests after user confirmation, and (3) apply Scout105 cleanup protocol. Use when asked to refactor, modularize, or restructure code safely.
50-execute-gated-150
[50] EXECUTE. Execute plans step-by-step with confirmation gates. Each step requires user approval before proceeding. Includes change management lifecycle (Pre-Change → During → Post-Change → Rollback). Use when implementing approved plans, deploying changes, or any multi-step execution requiring control and reversibility.
51-execute-quality-150
[51] EXECUTE. Commitment to maximum quality work with 150% coverage. Use when you need the highest quality output for critical tasks, complex problems, important decisions, or when standard work isn't enough. Triggers on "maximum quality", "150% mode", "full quality", "critical task", or when you explicitly want AI to work at its best.
thor-skills
An entry point and router for AI agents to manage various THOR-related cybersecurity tasks, including running scans, analyzing logs, troubleshooting, and maintenance.
tech-blog
Generates comprehensive technical blog posts, offering detailed explanations of system internals, architecture, and implementation, either through source code analysis or document-driven research.
chrome-debug
This skill empowers AI agents to debug web applications and inspect browser behavior using the Chrome DevTools Protocol (CDP), offering both collaborative (headful) and automated (headless) modes.
ux
This AI agent skill provides comprehensive guidance for creating professional and insightful User Experience (UX) designs, covering user research, information architecture, interaction design, visual guidance, and usability evaluation. It aims to produce actionable, user-centered solutions that avoid generic AI aesthetics.
astro
This skill provides essential Astro framework patterns, focusing on server-side rendering (SSR), static site generation (SSG), middleware, and TypeScript best practices. It helps AI agents implement secure authentication, manage API routes, and debug rendering behaviors within Astro projects.
ontopo
An AI agent skill to search for Israeli restaurants, check table availability, view menus, and retrieve booking links via the Ontopo platform, acting as an unofficial interface to its data.
grail-miner
This skill assists in setting up, managing, and optimizing Grail miners on Bittensor Subnet 81, handling tasks like environment configuration, R2 storage, model checkpoint management, and performance tuning.
modal-deployment
Run Python code in the cloud with serverless containers, GPUs, and autoscaling using Modal. This skill enables agents to generate code for deploying ML models, running batch jobs, serving APIs, and scaling compute-intensive workloads.
whisper-transcribe
Transcribes audio and video files to text using OpenAI's Whisper CLI, enhanced with contextual grounding from local markdown files for improved accuracy.