IDKMesh

Beginner comprehension and first-run documentation review (2026-10-11)

Owner question / goal

Can a person with limited software or AI-agent experience immediately tell what IDKMesh does, what works today, and how to try it? Improve the public documentation and preserve the result in this repository.

Findings from current main

Decision and work in this change

  1. Lead the public README with a concrete job-to-be-done and direct beginner paths instead of requiring people to understand the research program first.
  2. Rewrite the beginner guide around a verifiable first result: install the CLI in a local virtual environment, audit the committed vote matrix, and inspect readable output.
  3. Explicitly separate the available verdict-auditing path from the broader intended task/worker/verifier/decision framework cycle; no single-step end-to-end automation is implied.
  4. Include a worked mistake from the real fixture, explain ground_truth vs reviewer verdicts, and define terms in non-specialist language.
  5. Adjust the published quickstart’s first-screen explanation and make the link to the beginner guide visible.
  6. Preserve explicit limits: synthetic evidence is not real observation; no AI key is needed for local demos; control-tower cannot dispatch, approve, or merge; external-project bootstrap remains dry-run planning.

Scope and evidence

Base main reviewed: bee77a6125dbccb58f166ec363e978cd960f43f3 (2026-10-11). Relevant sources inspected: README.md, AGENTS.md, PROJECT_RULES.md, docs/BEGINNER_GUIDE.md, docs/GETTING_STARTED.md, docs/WHAT_IS_IDKMESH.md, docs/CAPABILITY_MATRIX.md, docs/PROJECT_ADOPTION_GUIDE.md, docs/RUN_LIFECYCLE_TUTORIAL.md, docs/start.html, pyproject.toml, idkmesh/cli.py, scripts/demo.py, and examples/gate-audit/panel-votes.example.json.

This is a documentation and wording improvement, not a runtime capability or scientific evidence promotion. Tests or live user research are not claimed unless their results are reported in the PR. A useful subsequent check is a first-time-user walkthrough by someone who has never seen IDKMesh: ask them what the tool accepts, what it produces, and what it does not automate, before and after following the guide.

Community impact

Makes the shortest hands-on path, evidence caveats, and onward routes obvious to new readers without requiring them to learn the architecture or join the research program first.

AI/tool provenance

Drafted with ChatGPT GPT-6 using live GitHub-connected repository data. No independent human usability study is represented.

Integration note (same documentation branch)

While this work was being prepared, main advanced from bee77a6125dbccb58f166ec363e978cd960f43f3 to 12242ba12e4823c025c76dce2b471b830afbdf7c with new command/API/library inventory documentation. The final documentation integrates those additions: the README is patched onto the later main text, and the beginner guide adds a direct link and introductory help commands for docs/EXECUTABLE_CAPABILITIES.md. No newer main changes are intentionally removed.