README.md

# LLM & Agent Builder Copilot v0.2.0

An upgraded skills-only Tekris Labs plugin for designing and implementing reliable LLM and tool-using agent workflows. Preserves the original package identity, icon, one-skill instruction set and seven original references in the archive. The current workflow is split into 18 focused skills and 17 new reference guides, with original references also available under `references/original/`.

## Entry points
- Use `skills/get-started/SKILL.md` for routing.
- Use `skills/choose-system-complexity` before assuming multi-agent requirements.
- Use `skills/implement-agent-workflow` only when code work is actually requested and host coding tools exist.
- Use `skills/design-agent-evals` and `skills/prepare-agent-release` to set evidence-backed release criteria.

## Architecture and execution limits
Skills-only. No bundled remote MCP server, hosted backend, secrets, agent runtime, or guaranteed code execution. Subagent orchestration happens only where the host exposes genuine subagent tools. The deterministic script under `scripts/` audits synthetic trace fixtures; it cannot validate real production agent behavior independently.

## Validation
From the root, run `python scripts/validate_package.py` and `python -m unittest discover -s tests -v`. Package tests do **not** execute the 28 prompt-level benchmark scenarios in an installed plugin.

## Publisher and submission
Tekris Labs: https://tekrislabs.com/ · support https://tekrislabs.com/contact · privacy https://tekrislabs.com/privacy · terms https://tekrislabs.com/terms. Publish through the existing OpenAI plugin update flow after review; this package alone does not update a published listing.