unvibecode 0.3.0

Local repository code map and business workflow risk review

ShipReadyV2 0.3.0

This project packages the authoritative ShipReady sources with the complete Customer MVP Runner V9 experience. Version 0.3.0 uses ShipReady's authenticated hosted model gateway, so customers do not provide an OpenAI API key.

Customer outputs:

1. Connected Code Map for LLMs.
2. Complete repository context ZIP.
3. Business Workflow Map.
4. Business Risk Findings, or a scoped no-findings review.

The terminal shows the V9 customer dashboard: real Fast Lane milestones, fine-grained business-review stages, reviewed-area counts, ETA or elapsed time, large-repository context-only messaging, and customer-ready output paths. Technical engine output is retained in each run's technical_logs folder.

One-time development setup

Open this project folder in VS Code and run:

Set-ExecutionPolicy -Scope Process -ExecutionPolicy Bypass
.\scripts\setup.ps1

This creates .venv and installs the project in editable mode. You do not need to recreate a source ZIP after changes.

Activate once

& ".\.venv\Scripts\shipready.exe" activate

Paste the ShipReady customer access key at the private prompt. The key is verified without uploading repository code and stored in the operating system's credential manager.

Run a complete review

& ".\.venv\Scripts\shipready.exe" review `
  --repository "D:\upstox" `
  --runs-root "D:\shipreadyv2\customer_runs"

Add --no-open-browser to create reports without opening browser tabs. No customer .env file or OpenAI API key is used.

For a normal repository, the Connected Code Map, Workflow Map and Risk Findings open only after the full review finishes; Risk Findings opens last. For a repository above approximately two million tokens, only the Connected Code Map and complete context ZIP are produced and the CLI displays one customer result rather than a confusing skipped second result.

If no candidate meets the evidence threshold, Step 06 is skipped. The review completes successfully and publishes a customer report explaining that no evidence-supported business risks were identified within the completed scope.

Permanent source workflow

Canonical package code:

src\shipreadyv2\customer_runner.py
src\shipreadyv2\cli.py
src\shipreadyv2\engine_sources\

customer_runner.py is the exact supplied V9 runner. The files under engine_sources are the supplied current analysis sources. Their SHA-256 hashes are recorded in SOURCE_MANIFEST.json. Version 0.3.0 keeps all 16 authoritative engine and runner sources byte-identical to the validated 0.2.1 baseline; only the package activation and network adapter changed.

For each change:

git switch -c "change/short-description"
# Edit files under src\shipreadyv2
& ".\.venv\Scripts\python.exe" -m pytest
git add .
git commit -m "Describe the change"

Build a release:

.\scripts\release.ps1 -Python "$PWD\.venv\Scripts\python.exe"

The wheel is written to dist, and RELEASE_CHECKSUMS.txt is refreshed. Do not edit an installed wheel or files under site-packages.

Hosted-service boundary

The customer package performs graph construction, context preparation, deterministic selection and report rendering locally. The two existing structured model requests use:

https://shipready-api.alphashots.ai/v1

The package passes a ShipReady customer credential, never an OpenAI key. The hosted service allows only ShipReady's two established structured request schemas and does not log or persist customer request bodies. A private temporary engine directory is removed after every review.

For non-interactive automation, SHIPREADY_ACCESS_KEY may be supplied in the process environment instead of using shipready activate.