README.md

# Software & AI Copilot — v0.2.0

A **skills-only**, repository-first plugin for software, data and AI codebases. It does not bundle an MCP server, credentials, autonomous build infrastructure or deployment service.

## What changed
Reorganized the original single skill into 18 focused workflow skills, retained all six original technical references under `references/original/`, archived the original skill for audit, added domain references, templates, deterministic evidence gates and a scenario benchmark suite. Original artwork retained.

## Invocation
Skills are auto-discovered under `skills/` from the portable `plugin.json`. Start with `get-started`, or route directly to the appropriate specialized workflow. All skills share `references/workflow-contract.md` and follow repository-specific conventions.

## Runtime truth
Actual source inspection, code edits, tests, subagents, external lookups and deployment require tools available in ChatGPT Work/Codex or the user's own environment. Text instructions cannot create tooling. A benchmark specification is **not** a passed benchmark.

## Local checks
Run `python3 scripts/validate_package.py` and `python3 -m unittest discover -s tests -v`. For optional evidence reports: `python3 scripts/check_evidence.py examples/evidence-good.json`.

## Publication
Publisher: Tekris Labs. Website: https://tekrislabs.com/ . Support: https://tekrislabs.com/contact . Privacy: https://tekrislabs.com/privacy . Terms: https://tekrislabs.com/terms . Three starter prompts. Public identity unchanged: `Software & AI Copilot` (`software-data-ai-copilot`).