Skip to content
Open
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Original file line number Diff line number Diff line change
Expand Up @@ -4,7 +4,7 @@ title: Storing models in mocks — how close laqi gets to being a source of trut

# Storing models in mocks — how close laqi gets to being a source of truth

**Status:** open — argues against [ADR-0013](/decisions/0013-mocks-remember-their-model/), which is written but unmerged (PR #61)
**Status:** resolved — see "Outcome" below. Argued against [ADR-0013](/decisions/0013-mocks-remember-their-model/) (PR #61), which merged and was then superseded by the [JSON Schema adapters spec](/design/json-schema-adapters/)
**Date:** 2026-09-06
**Trigger:** "it would end up being a source of truth for laqi, when it should only deliver mock APIs"

Expand Down Expand Up @@ -207,6 +207,27 @@ kept for the thirty-second case.

Everything else here follows from that answer.

## Outcome

Resolved by [Plan 14](/plans/2026-09-07-14-json-schema-adapters/), whose
answer to the open question is **no**: the paste box is a convenience, and
project references and schema imports are the durable path (its product
decisions 1, 8 and 9).

- ADR-0013's `generatedFrom` merged in PR #61 and was then superseded. A
response no longer stores the pasted TypeScript source.
- What a response keeps is a JSON Schema (draft 2020-12) snapshot of the rules,
not the source: option B's "store the recipe, never the definition", with
JSON Schema as the compact, reviewable encoding this analysis said B lacked.
- The snapshot records imported rules and certifies nothing about the body or
the real backend; laqi never validates a body against it.
- Option C landed beside it: schemas can come from JSON Schema, OpenAPI, or the
project's own code through Standard JSON Schema.

The contracts are in the [JSON Schema adapters spec](/design/json-schema-adapters/),
merged in PRs #62, #64 and #65 and shipped in `laqi` 2.1.0. The ledger below is
kept as it was written.

## Ledger

Not analysed yet, and deliberately not guessed at:
Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -4,7 +4,7 @@ title: JSON Schema and adapters — design spec

# JSON Schema and adapters — design spec

**Status:** phase 1 implemented; phases 2-6 not started. Derived from [Plan 14](/plans/2026-09-07-14-json-schema-adapters/) and its twelve-point review
**Status:** all six phases implemented and merged — phases 1–2 in [#64](https://github.com/csdev19/laqi/pull/64), phases 3–6 in [#65](https://github.com/csdev19/laqi/pull/65) (2026-09-10), shipped in `laqi` 2.1.0. Derived from [Plan 14](/plans/2026-09-07-14-json-schema-adapters/) and its twelve-point review
**Date:** 2026-09-07, revised 2026-09-09 by what phase 1 measured
**Supersedes:** the section "One source of truth: the data. Models are never persisted" in [data-generators](/design/data-generators/), and the `generatedFrom` experiment of [ADR-0013](/decisions/0013-mocks-remember-their-model/) (PR #61) and the compact-recipe experiment that followed it

Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -5,7 +5,7 @@ title: Plan 14 — JSON Schema and adapters
# Plan 14 — JSON Schema and adapters

**Date:** 2026-09-07
**Status:** Planned — specification decisions revised after the twelve-point review on 2026-09-07.
**Status:** Merged — all six phases of the [design spec](/design/json-schema-adapters/) are on `main`: the spec in [#62](https://github.com/csdev19/laqi/pull/62), phases 1–2 in [#64](https://github.com/csdev19/laqi/pull/64) (phase 1 was first reviewed as [#63](https://github.com/csdev19/laqi/pull/63)), phases 3–6 in [#65](https://github.com/csdev19/laqi/pull/65), shipped in `laqi` 2.1.0. Specification decisions were revised after the twelve-point review on 2026-09-07.
**Scope:** Replace proprietary generation metadata with JSON Schema and introduce
explicit input/output adapters across the panel, local API, and MCP.

Expand Down
Original file line number Diff line number Diff line change
Expand Up @@ -5,7 +5,7 @@ title: Plan 15 — laqi.dev correctness fixes
# Plan 15 — laqi.dev correctness fixes

**Date:** 2026-10-01
**Status:** All eight tasks are implemented on `fix/laqi-dev-page-claims`.
**Status:** Merged in [#68](https://github.com/csdev19/laqi/pull/68) (2026-10-05). All eight tasks were implemented on `fix/laqi-dev-page-claims`.
Task 5 was resolved by removing the version badge rather than by any of the
three options below; task 1's removal was then reversed, deliberately, once
the recordings were about to be made. Both are recorded in their tasks.
Expand Down
Original file line number Diff line number Diff line change
@@ -0,0 +1,93 @@
---
title: Plan 16 — Docs housekeeping
---

# Plan 16 — Docs housekeeping

**Date:** 2026-10-07
**Linear:** [NIW2-103](https://linear.app/niway/issue/NIW2-103)
**Reference:** `origin/main` at `ee34302`

Bring the internal docs in line with `main`. The roadmap was last reviewed on
2026-09-02, the plans index still listed Plans 11–14 as **Planned**, and the
JSON Schema spec header still said phases 2–6 had not started.

Every status below was checked against `gh pr list --state all` and
`git log origin/main` on 2026-10-07, not against the plan documents.

## What the PRs say

| Work | PR(s) | State on 2026-10-07 |
| ---------------------------------------------- | ------------------------------- | ------------------- |
| Plan 11 — response scaffolding | #54 | Merged 2026-09-03 |
| Plan 12 — package-manager toggle | #55 | Open |
| Plan 13 — terminal output, stage 2 | #56 | Open |
| ADR-0013 — mocks remember their model | #61 | Merged, superseded |
| Plan 14 — JSON Schema spec | #62 | Merged 2026-09-09 |
| Plan 14 — phases 1–2 (goldens, compiler) | #63 (into the spec branch), #64 | Merged 2026-09-09 |
| Plan 14 — phases 3–6 (stored schema, adapters) | #65 | Merged 2026-09-10 |
| Plan 15 — laqi.dev fixes and the v3 home | #68 | Merged 2026-10-05 |
| Hero demo, getting ready to record | #71 | Merged 2026-10-05 |
| Data-type fidelity (numbers, dates, > 2^53) | #73 | Merged 2026-10-06 |
| Hero demo recording published on laqi.dev | #74 | Merged 2026-10-07 |
| Releases: `laqi` 2.1.0, site 0.2.0 and 0.3.0 | #50, #48, #75 | Merged |
| Editor saved vs live state | #66 | Open |

## Tasks

- [x] `product/roadmap.md`: new review date, the shipped table extended
(Plan 11, Plan 14, stored schema, v3 home, hero demo, data-type
fidelity, 2.1.0), In flight lists the three open PRs, and the
"Next" sections that have since shipped or got a PR say so.
- [x] `plans/index.md`: status column fixed for 11–14, Plan 14 and Plan 16
added to the table, the header paragraph no longer calls 14 and 15
proposals.
- [x] Plan 14's own status line, and Plan 15's.
- [x] `design/json-schema-adapters.md`: status header says all six phases are
implemented, with the PRs.
- [x] `adversarial/storing-models-in-mocks.md`: marked resolved, with a short
"Outcome" section pointing at what Plan 14 decided.
- [x] Duplicate keys within one JSON file: documented on laqi.dev's
[mock files](https://laqi.dev/docs/mock-files/) page (the place a user
reads the file format), and the Plan 1 audit item marked done.
- [x] The durable fix for the hand-maintained status column: proposed below.
Not built in this plan.

## Durable fix — proposal: derive plan status from PR state

The status column has drifted twice (fixed 2026-09-02, again here). Every
time, the plan documents were right about _what_ and wrong about _whether_,
because "merged" is a fact GitHub already holds and the docs copy it by hand.

**Proposal.** Each plan declares the PRs that implement it in frontmatter, and
the status is computed from them, never written:

```yaml
---
title: Plan 11 — Response scaffolding on create
prs: [54]
---
```

1. A script, `scripts/docs/plan-status.ts`, reads every `plans/*.md`
frontmatter, asks `gh pr view <n> --json state,mergedAt` for each PR, and
derives one status per plan: **Merged** (every PR merged), **In review**
(at least one open), **Planned** (no PRs listed), **Abandoned** (all
closed unmerged).
2. It writes `plans/status.json` (committed), and the plans index renders its
status column from that file through a small Astro component, so the
Markdown table no longer has a status column to forget.
3. It runs in two places: by hand (`bun scripts/docs/plan-status.ts`), and in a
scheduled workflow that opens a PR when `status.json` changes. It never
runs in `validate.yml`: a PR gate that depends on the GitHub API state of
_other_ PRs would go red for reasons the PR under review did not cause.
4. `--check` mode exits non-zero when a plan's `**Status:**` prose line
contradicts the derived state, so the prose cannot drift either.

**Why not now.** It touches the docs app's rendering, needs a workflow with a
token that can open PRs, and changes the format of every plan's frontmatter.
That is a feature, not housekeeping, and deserves its own review.

**Open question for Cristian.** Whether the derived status should also feed the
roadmap's "Shipped" table, or the roadmap stays curated prose (it groups by
feature, not by plan, so it is not a one-to-one mapping).
Loading
Loading