docs: GEO/README pass (llms.txt + accuracy)

README.md

@@ -8,7 +8,7 @@
 **Git repository archaeology framework.** Mine commit history, detect development signals, run 6 analysis vectors, and generate engineering narrative reports — from any git repository, fully local, no external services.
 
 ```bash
-pip install devarch-framework
+git clone https://github.com/KyaniteLabs/devarch-framework.git && cd devarch-framework && pip install -e .
 ```
 
 DevArch treats your git history as structured data. It extracts commits into a queryable SQLite database, runs heuristic signal detection (gaps, velocity shifts, author changes), executes specialized analysis vectors, and generates interactive HTML visualizations and markdown reports. Built for engineers, researchers, and AI agents that need to understand how a codebase evolved.

llms.txt

@@ -1,110 +1,12 @@
-# DevArch Framework
+# devarch-framework
 
-> Git repository archaeology framework — mine commit history, detect signals, run 6 analysis vectors, and generate engineering reports.
+> Git repository archaeology framework — mine commit history into evidence-backed engineering reports.
 
-## What is DevArch?
+## What it is
+A Python framework + CLI that mines a repo's git history, runs 6 analysis vectors, detects signals, and generates engineering/learning reports. ICM-aligned.
 
-DevArch Framework is a Python CLI tool that treats git history as a primary data source. It mines commits into a structured SQLite database, detects development signals (velocity shifts, gaps, scope changes), runs specialized analysis vectors, and generates markdown/HTML reports. It is designed for engineers, researchers, and AI agents that need to understand how a codebase evolved.
-
-## Who is it for?
-
-- **Engineers** auditing a codebase before joining or acquiring a project
-- **Researchers** studying software development patterns
-- **AI agents** running repository reviews with structured context
-- **Teams** wanting to understand their own development history
-
-## How do I install DevArch?
-
-```bash
-pip install devarch-framework
-```
-
-Or from source:
-
-```bash
-git clone https://github.com/KyaniteLabs/devarch-framework.git
-cd devarch-framework
-pip install -e .
-```
-
-## How do I use DevArch?
-
-```bash
-# Initialize a project
-devarch init my-project --repo-url https://github.com/user/repo
-
-# Mine git history
-devarch mine /path/to/repo --project my-project
-
-# Build SQLite database
-devarch build-db my-project
-
-# Detect development signals
-devarch signals my-project
-
-# Run all 6 analysis vectors
-devarch analyze my-project
-
-# Generate HTML visualization
-devarch visualize my-project
-
-# Export full report
-devarch export-report my-project --format markdown
-```
-
-## What are the 6 analysis vectors?
-
-1. **SDLC Gap Finder** — identify gaps in software development lifecycle practices
-2. **ML Pattern Mapper** — detect machine learning patterns and adoption
-3. **Agentic Workflow Analyzer** — identify AI-agent and automation workflows
-4. **Formal Terms Mapper** — track formal methods and terminology usage
-5. **Source Archaeologist** — deep code evolution and change tracking
-6. **YouTube Correlator** — correlate commit patterns with external timeline data
-
-## What signals does DevArch detect?
-
-- Commit velocity shifts (sudden acceleration or slowdown)
-- Long gaps in development activity
-- Author changes and contributor patterns
-- Scope changes (file-count jumps, language additions)
-- Supplementary data correlations
-
-## What is the era system?
-
-DevArch can scan commits for distinct development phases ("eras") — periods where the project's character changed significantly. It cascades era labels across files and maps era boundaries, giving you a timeline of the project's life.
-
-## What output does DevArch produce?
-
-- SQLite database (`archaeology.db`) — queryable via Datasette
-- `detected-signals.json` — structured signal findings
-- `analysis-<vector>.json` — per-vector analysis results
-- Interactive HTML visualization with Chart.js charts
-- Markdown and HTML reports
-- Public case study export (sanitized for sharing)
-
-## Key files
-
-- `README.md` — full documentation and quick start
-- `archaeology/cli.py` — CLI entry point (20+ commands)
-- `analysis-vectors/` — 6 analysis vector definitions (markdown prompts)
-- `stages/` — 9-stage pipeline contracts (ICM L2)
-- `shared/concepts.md` — framework concepts (signals, vectors, eras)
-- `docs/decisions.md` — architecture decision records
-- `setup.py` — package metadata
-
-## Requirements
-
-- Python 3.10+
-- A git repository with commit history
-- No external services required — fully local
-
-## License
-
-MIT — https://github.com/KyaniteLabs/devarch-framework/blob/main/LICENSE
+## Install / use
+git clone the repo, then `pip install -e .` (not yet on PyPI).
 
 ## Links
-
-- GitHub: https://github.com/KyaniteLabs/devarch-framework
-- PyPI: https://pypi.org/project/devarch-framework/
-- Homepage: https://kyanitelabs.tech
-- Related: https://github.com/KyaniteLabs/dev-learning-archaeologist
+- Repository: https://github.com/KyaniteLabs/devarch-framework