ai-readiness-assessment

TechTalk AI Readiness Assessment

Product page: techtalk.at/ai-readiness-assessment

A self-contained AI readiness assessment, packaged as a plugin for GitHub Copilot (and any other tool that consumes the same .claude-plugin/plugin.json + commands/ + skills/ layout — Claude Code, Cursor via wrapper, Windsurf via wrapper).

Drop into any repository, invoke once — get a scan, a score, and a plan:

• A timestamped assessment at assessments/YYYY-MM-DD-assessment.md.
• A full placement on the Agentic Experience 5-Level Habitat Maturity Model — all fourteen dimensions, L1–L5, reported with the model's own verbs — and a headline Habitat Maturity Level.
• A position on the six-level cognitive ladder from The Sovereign Engineer (Russ Miles, Habitat-Thinking) — leanpub.com/thesovereignengineer — L0 Aware of the landscape through L5 Sovereign engineering — and the Habitat Build Gap between the two reads.
• A gap-anchored reading path into the book.
• Recommendations on next steps matched to the weakest of the three disciplines: Context Engineering, Architectural Constraints, or Guardrail Design.

The assessment is fully self-contained. It does not depend on any external plugin, skill, agent, or service. The model, the framework, scoring heuristic, and evidence checklist all live inside this repo. The instrument itself is inspired by the ai-literacy-superpowers plugin — see Inspired by below.

Full documentation: techtalk.github.io/ai-readiness-assessment — a Diátaxis-structured site with tutorials, how-to guides, reference, and explanation.

Contributing or working on the plugin itself? Start with ONBOARDING.md.

Prerequisites

To run the assessment you need:

• One supported AI coding tool, with an active account or subscription — either:
  • GitHub Copilot CLI — Node.js 18+ (npm install -g @github/copilot) and a signed-in GitHub account with Copilot access; or
  • Claude Code — the claude CLI installed (code.claude.com) and a signed-in Anthropic account.
  • Cursor and Windsurf also work via their convention-file wrappers.
• The plugin installed from the techtalkai marketplace — two commands, see Install below.
• A repository to assess. Open the tool with that repository as the working directory / project root, so the scan can see its instruction files, CI config, specs, and other habitat signals. A git repo is recommended but not required.
• Write access to the working directory — the assessment saves its report to assessments/YYYY-MM-DD-assessment.md in the repo.
• Network access the first time, to register the marketplace from GitHub.

The assessment itself runs entirely inside the AI tool — there is no language runtime, build step, or service to set up. Budget around ten minutes, and expect to answer 3–5 short clarifying questions interactively. (The Python in tests/ is only for contributors verifying the instrument — not needed to run an assessment.)

Install

This repo is the techtalkai plugin marketplace. The plugin format (.claude-plugin/plugin.json + commands/ + skills/) is consumed by both GitHub Copilot CLI and Claude Code, so the install flow is the same two steps in either tool — register the marketplace, then install the plugin.

GitHub Copilot CLI

First install the Copilot CLI (if you haven't already) and sign in: