Documentation
From installation to a useful first result.
Start with normal use. Python validation and development tests are optional and documented separately.
Installation
Install the skill in Codex, Claude Code, or Claude.ai.
Five-minute quick start
Run the fictional first example and understand the result.
Examples and outputs
See the complete fictional before-and-after result.
Safety and limitations
Understand the human-review and commercial boundaries.
Configuration
Change labels, drafting preferences, tone, and approval roles.
Troubleshooting
Resolve discovery, source, confidence, and optional Python issues.
Changelog
See the current version, included behavior, and known limitations.
How the source roles work
Baseline: the SOW, proposal, or other confirmed document used for comparison.
Client request: the new work being evaluated.
Supporting sources: optional project status, amendments, briefs, rate cards, or policies that add relevant facts.
How results should be used
Every report is draft operational analysis. A qualified agency reviewer confirms the sources, evidence, classification, pricing, schedule impact, and client wording before any commitment is made.