- Home
- Skills
- Agents & Orchestration
- wu5 dev flow
wu5 dev flow
A rigorous Python dev workflow enforcing SDD, TDD, and Git gates through an audited state machine.
Free
Works with the AI Agents you already use
wu5 dev flow
Example session with this skill installed
Initialize a new feature called 'user-auth' in this Python repo using the full flow. I need to add JWT verification to the existing login endpoint.
- Read your context and instructions
- Compiled the wu5 dev flow
I have initialized the user-auth change using the full SDD/TDD workflow. The state machine is now at drafting-spec, and I have generated the initial directory structure for this task.
Initialization Result
| Action | Path/Status | Description |
|---|---|---|
| Workspace | /spec/changes/user-auth/ | Created directory for feature specifications |
| State | drafting-spec | Current phase in state.json |
| Drafted | proposal.md | Initial requirement summary for JWT verification |
| Drafted | design.md | Placeholder for token validation logic and dependency injection |
| Script | wu5_flow.py status | System verified: No dirty workspace detected |
Proposed Specification Outline
I have drafted a proposal focusing on
- Dependency Injection: Moving secret keys out of the implementation into a config provider.
- Red-Green Plan: Writing a failing test for a 401 response on expired tokens before any implementation.
- Review Points: Specifically checking for hardcoded algorithms or salt values.
Next steps
- Review the generated
spec/changes/user-auth/proposal.mdand provide feedback. - Once satisfied, run
python scripts/wu5_flow.py approve specto lock the requirements. - I will then begin the
drafting-planphase to map specific tests to the JWT logic.
Connects securely to your tools. The creator never sees your data.
About this skill
Software projects often fail due to drift between specifications, tests, and implementation. This skill enforces a rigorous, state-machine driven workflow that prevents unverified code from reaching your main branch.
What it does
- Enforce SDD/TDD: Requires Software Design Documents and Test-Driven Development cycles for every functional change.
- Automated Gates: Uses a dedicated Python script to verify project state, approve baselines, and check commit readiness.
- Zero-Hardcode Policy: Scans and blocks implementation of hardcoded keys, environment-specific paths, or business thresholds.
- State Persistence: Tracks progress in
state.jsonandstatus.mdto maintain context across different sessions or agents. - Atomic Commits: Handles local git staging and commits only after rigorous verification and review cycles.
How it works
- Initialize: Run
wu5_flow.py initto scaffold the specification directory and baseline the existing codebase. - Define: Create a new change slug and draft the
proposal.mdandplan.mdfor user approval. - Develop: Execute the Red-Green-Refactor loop, ensuring tests fail as expected before writing business logic.
- Verify: Run automated checks for types, linting, and test coverage, followed by a two-stage manual review gate.
- Ship: Archive the change, update the feature documentation, and prepare the final PR summary.
Frameworks & tools
Specifically designed for Python environments using pytest, pyright, and ruff. It integrates directly with Git for workflow enforcement and utilizes custom Python scripts for state management.
Why this beats prompting it yourself
Prompting an agent to "follow TDD" often results in shortcuts or forgotten steps over long sessions. This skill uses a hard-coded state machine and external script validation that refuses to proceed if specifications are unapproved or tests are missing.
Use cases
- Initializing a new Python project with strict architectural standards.
- Executing high-stakes refactoring where regression is not an option.
- Onboarding agents to existing repositories while maintaining a verifiable audit trail.
- Managing complex feature additions that span multiple work sessions.
Known limitations
Does not handle database schema migrations or irreversible data transformations. Requires local Python environment to run the gatekeeper scripts.
Requires local Python environment to run wu5_flow.py. Excludes database migrations and data transformations. One active change per repository limit.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
1 use
Times used to date
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 1 month ago
- Free with an Agensi account
Requires