Skip to content

Tooling in scaffold skill, commit-per-task law, Task 1 opens with add-data-source - #3

Merged
chumleesockson merged 1 commit into
mainfrom
scaffold-tooling-and-commit-law
Jun 27, 2026
Merged

Tooling in scaffold skill, commit-per-task law, Task 1 opens with add-data-source#3
chumleesockson merged 1 commit into
mainfrom
scaffold-tooling-and-commit-law

Conversation

@chumleesockson

Copy link
Copy Markdown
Owner

Aligns the harness with the confirmed live run-through: 3 setup skills, then 4 build-verify tasks (no separate dataverse-add-data-source step). Two of the changes answer "shouldn't the AI do that?" — yes, and now the harness directs it to.

Changes

  1. scaffold-code-app skill (both .agents and .claude mirrors) — new step 5: set up the quality tooling (Vitest + RTL + jsdom, ESLint + typescript-eslint + react-hooks, the npm run verify script). This makes the app verifiable before Task 1's gate runs, instead of Task 1 having to configure tooling live. Push step renumbered to 6.
  2. AGENTS.md — new Architect-mode law: commit each task once its gate is green, one task per commit. The AI commits, directed by the harness; the git log becomes a live "I directed it" artifact you can scroll on stage.
  3. specs/tasks/task-1-data-layer.md — Task 1 now opens by running dataverse-add-data-source (it is the first build task, not a fourth setup skill), and the smoke-test bullet assumes the runner was already set up by scaffold-code-app.
  4. DEMO-PLAN.md — rewritten from the old 4-line skill list into the full run sheet: setup skills → task loop → deploy → teaching points → dry-run risks.

The run-through this encodes

Setup: /dataverse-scaffold-tables/dataverse-create-records/scaffold-code-app
Build: Complete Task 1 → 2 → 3 → 4 (each: read AGENTS.md + task, build, npm run verify, commit when green, stop)
Deploy: npm run buildpower-apps push → live sign-in + real write.

Note for maintenance

The skills are mirrored in .agents/skills/ and .claude/skills/; this PR keeps both identical. Any future skill edit has to touch both or they drift.

🤖 Generated with Claude Code

… with add-data-source

Aligns the harness with the confirmed live run-through (3 setup skills,
then 4 build-verify tasks).

- scaffold-code-app skill (both mirrored copies): new step 5 sets up the
  quality tooling (Vitest + RTL + jsdom, ESLint + react-hooks, the
  `npm run verify` script) so the gate works before Task 1; push step
  renumbered to 6.
- AGENTS.md: new Architect-mode law to commit each task once its gate is
  green, one task per commit. The git log becomes the build record.
- task-1-data-layer.md: opens by running dataverse-add-data-source (no
  longer a separate setup skill); the smoke-test bullet now assumes the
  runner was set up by scaffold-code-app.
- DEMO-PLAN.md: rewritten as the confirmed run sheet (setup skills, task
  loop, deploy, teaching points, dry-run risks).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Copilot AI review requested due to automatic review settings June 27, 2026 09:09
@chumleesockson
chumleesockson merged commit c054e8a into main Jun 27, 2026
1 check passed
@chumleesockson
chumleesockson deleted the scaffold-tooling-and-commit-law branch June 27, 2026 09:11

Copilot AI left a comment

Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Pull request overview

Updates the repo’s “harness” documentation and skill guidance to match a live run-through flow: 3 setup skills (including scaffolding quality tooling) followed by 4 build/verify tasks, with Task 1 starting by adding Dataverse data sources and each task committed after a green verify gate.

Changes:

  • Adds a new “quality tooling” step to the scaffold-code-app skill so npm run verify is available before Task 1.
  • Introduces an Architect-mode rule to commit after each task’s green gate, one task per commit.
  • Adjusts Task 1 to begin by running dataverse-add-data-source and updates the demo run sheet to the new end-to-end flow.

Reviewed changes

Copilot reviewed 5 out of 5 changed files in this pull request and generated 4 comments.

Show a summary per file
File Description
specs/tasks/task-1-data-layer.md Adds “Start here” guidance to run add-data-source at the beginning of Task 1 and clarifies tooling expectations.
DEMO-PLAN.md Expands from a short list into a full demo run sheet (setup → task loop → deploy → teaching points → risks).
AGENTS.md Adds “commit per task once gate is green” as an Architect-mode law.
.claude/skills/scaffold-code-app/SKILL.md Adds a new step to set up testing/linting/verify scripts before feature work; renumbers push step.
.agents/skills/scaffold-code-app/SKILL.md Same as .claude/ mirror: adds tooling step and renumbers push step.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.


**Goal:** the two Dataverse tables become typed data sources the app can read and write; no UI yet.

> **Start here:** run the `dataverse-add-data-source` skill to add `eppc_session` then `eppc_feedback`
Comment thread DEMO-PLAN.md

### Deploy + the live proof

- `npm run build`, then `power-apps push` into the same solution.
Comment on lines +91 to +93
See **AGENTS.md (repo root) → Stack notes** for the exact pins. After this, `npm run verify` runs
clean on the empty app, so the first task gate (Task 1) has a working harness to run against rather
than being set up live.
Comment on lines +91 to +93
See **AGENTS.md (repo root) → Stack notes** for the exact pins. After this, `npm run verify` runs
clean on the empty app, so the first task gate (Task 1) has a working harness to run against rather
than being set up live.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants