@qvac/core 0.3.0 → 0.3.1
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist-lib/lib/api/chat.d.ts +3 -1
- package/dist-lib/lib/api/chat.js +6 -2
- package/dist-lib/lib/api/index.d.ts +2 -0
- package/dist-lib/lib/api/index.js +3 -1
- package/dist-lib/lib/api/inference-calls.d.ts +18 -0
- package/dist-lib/lib/api/inference-calls.js +15 -0
- package/dist-lib/lib/api/rag.js +10 -2
- package/dist-lib/lib/assistant/harness.d.ts +1 -0
- package/dist-lib/lib/assistant-tree.d.ts +2 -0
- package/dist-lib/lib/assistant-tree.js +11 -2
- package/dist-lib/lib/assistant.d.ts +14 -1
- package/dist-lib/lib/assistant.js +107 -15
- package/dist-lib/lib/core/chunk.js +5 -0
- package/dist-lib/lib/core/files/rag/build.d.ts +1 -2
- package/dist-lib/lib/core/files/rag/build.js +1 -2
- package/dist-lib/lib/core/files/rag/extract/office-collabora.d.ts +1 -1
- package/dist-lib/lib/core/files/rag/index.d.ts +4 -24
- package/dist-lib/lib/core/files/rag/index.js +3 -17
- package/dist-lib/lib/core/files/rag/load.d.ts +1 -2
- package/dist-lib/lib/core/files/rag/load.js +1 -2
- package/dist-lib/lib/core/files/rag/produce.d.ts +1 -1
- package/dist-lib/lib/core/files/rag/produce.js +1 -3
- package/dist-lib/lib/core/files/rag/retrieval/full-text.d.ts +1 -1
- package/dist-lib/lib/core/files/rag/retrieval/full-text.js +1 -4
- package/dist-lib/lib/core/files/rag/similarity.d.ts +5 -0
- package/dist-lib/lib/core/files/rag/similarity.js +6 -0
- package/dist-lib/lib/core/files/upload.d.ts +1 -1
- package/dist-lib/lib/core/inference-calls-handoff.d.ts +20 -0
- package/dist-lib/lib/core/inference-calls-handoff.js +37 -0
- package/dist-lib/lib/core/inference-calls-receiver.d.ts +14 -0
- package/dist-lib/lib/core/inference-calls-receiver.js +39 -0
- package/dist-lib/lib/core/notification.js +5 -4
- package/dist-lib/lib/core.d.ts +8 -1
- package/dist-lib/lib/core.js +78 -4
- package/dist-lib/lib/inference-calls.d.ts +40 -0
- package/dist-lib/lib/inference-calls.js +266 -0
- package/dist-lib/schema/content.js +5 -1
- package/dist-lib/schema/engine.js +42 -1
- package/dist-lib/schema/qvac.js +4 -2
- package/dist-lib/spec/bind.js +3 -0
- package/dist-lib/spec/calls.js +1 -0
- package/dist-lib/spec/capabilities.d.ts +2 -0
- package/dist-lib/spec/hrpc/hrpc.json +13 -0
- package/dist-lib/spec/hrpc/index.js +15 -3
- package/dist-lib/spec/hrpc/messages.js +172 -49
- package/dist-lib/spec/hyperschema/index.js +172 -49
- package/dist-lib/spec/hyperschema/schema.json +127 -1
- package/dist-lib/spec/hyperschema/types.d.ts +34 -1
- package/dist-lib/spec/local/hyperdb/messages.js +64 -128
- package/dist-lib/spec/local/hyperschema/index.js +32 -64
- package/dist-lib/spec/local/hyperschema/schema.json +16 -0
- package/dist-lib/spec/mesh/hyperdb/messages.js +176 -311
- package/dist-lib/spec/mesh/hyperdispatch/messages.js +132 -236
- package/dist-lib/spec/mesh/hyperschema/index.js +132 -236
- package/dist-lib/spec/mesh/hyperschema/schema.json +70 -1
- package/dist-lib/spec/mesh/hyperschema/types.d.ts +3 -1
- package/package.json +14 -17
- package/tui/lib/knowledge-import.mjs +1 -1
- package/.claude/skills/asana/SKILL.md +0 -86
- package/.claude/skills/compaction/SKILL.md +0 -136
- package/.claude/skills/device-mesh-debug/SKILL.md +0 -328
- package/.claude/skills/gh-stack/SKILL.md +0 -181
- package/.claude/skills/gh-stack/references/commands.md +0 -177
- package/.claude/skills/gh-stack/references/stack-design.md +0 -97
- package/.claude/skills/gh-stack/references/troubleshooting.md +0 -159
- package/.claude/skills/js-house-style/SKILL.md +0 -671
- package/.claude/skills/pr-create/SKILL.md +0 -243
- package/.claude/skills/pr-review/SKILL.md +0 -428
- package/.claude/skills/release/SKILL.md +0 -122
- package/CLAUDE.md +0 -48
- package/dist-lib/lib/core/files/rag/artifact/codec.d.ts +0 -3
- package/dist-lib/lib/core/files/rag/artifact/codec.js +0 -172
- package/dist-lib/lib/core/files/rag/artifact/id.d.ts +0 -10
- package/dist-lib/lib/core/files/rag/artifact/id.js +0 -39
- package/dist-lib/lib/core/files/rag/artifact/types.d.ts +0 -24
- package/dist-lib/lib/core/files/rag/artifact/types.js +0 -41
- package/dist-lib/lib/core/files/rag/chunk/chunking.d.ts +0 -12
- package/dist-lib/lib/core/files/rag/chunk/chunking.js +0 -117
- package/dist-lib/lib/core/files/rag/chunk/metadata.d.ts +0 -30
- package/dist-lib/lib/core/files/rag/chunk/metadata.js +0 -89
- package/dist-lib/lib/core/files/rag/chunk/tokenizer.d.ts +0 -7
- package/dist-lib/lib/core/files/rag/chunk/tokenizer.js +0 -18
- package/dist-lib/lib/core/files/rag/extract/csv-text.d.ts +0 -1
- package/dist-lib/lib/core/files/rag/extract/csv-text.js +0 -117
- package/dist-lib/lib/core/files/rag/extract/document-types.d.ts +0 -22
- package/dist-lib/lib/core/files/rag/extract/document-types.js +0 -152
- package/dist-lib/lib/core/files/rag/extract/extract.d.ts +0 -17
- package/dist-lib/lib/core/files/rag/extract/extract.js +0 -81
- package/dist-lib/lib/core/files/rag/extract/html-table-text.d.ts +0 -1
- package/dist-lib/lib/core/files/rag/extract/html-table-text.js +0 -125
- package/dist-lib/lib/core/files/rag/extract/html-text.d.ts +0 -1
- package/dist-lib/lib/core/files/rag/extract/html-text.js +0 -132
- package/dist-lib/lib/core/files/rag/extract/pdf-platform.d.ts +0 -17
- package/dist-lib/lib/core/files/rag/extract/pdf-platform.js +0 -141
- package/dist-lib/lib/core/files/rag/extract/pdf-text.d.ts +0 -12
- package/dist-lib/lib/core/files/rag/extract/pdf-text.js +0 -43
- package/dist-lib/lib/core/files/rag/extract/rtf-text.d.ts +0 -1
- package/dist-lib/lib/core/files/rag/extract/rtf-text.js +0 -15
- package/dist-lib/lib/core/files/rag/extract/text-decoding.d.ts +0 -1
- package/dist-lib/lib/core/files/rag/extract/text-decoding.js +0 -91
- package/dist-lib/lib/core/files/rag/extract/types.d.ts +0 -36
- package/dist-lib/lib/core/files/rag/extract/types.js +0 -12
- package/dist-lib/lib/core/files/rag/extract/utf.d.ts +0 -2
- package/dist-lib/lib/core/files/rag/extract/utf.js +0 -18
- package/dist-lib/lib/core/files/rag/retrieval/citation-match.d.ts +0 -2
- package/dist-lib/lib/core/files/rag/retrieval/citation-match.js +0 -16
- package/dist-lib/lib/core/files/rag/retrieval/coverage.d.ts +0 -4
- package/dist-lib/lib/core/files/rag/retrieval/coverage.js +0 -48
- package/dist-lib/lib/core/files/rag/retrieval/diversity.d.ts +0 -10
- package/dist-lib/lib/core/files/rag/retrieval/diversity.js +0 -37
- package/dist-lib/lib/core/files/rag/retrieval/floor.d.ts +0 -5
- package/dist-lib/lib/core/files/rag/retrieval/floor.js +0 -21
- package/dist-lib/lib/core/files/rag/retrieval/format.d.ts +0 -6
- package/dist-lib/lib/core/files/rag/retrieval/format.js +0 -21
- package/dist-lib/lib/core/files/rag/retrieval/knowledge.d.ts +0 -2
- package/dist-lib/lib/core/files/rag/retrieval/knowledge.js +0 -54
- package/dist-lib/lib/core/files/rag/retrieval/notices.d.ts +0 -16
- package/dist-lib/lib/core/files/rag/retrieval/notices.js +0 -27
- package/dist-lib/lib/core/files/rag/retrieval/scoring.d.ts +0 -8
- package/dist-lib/lib/core/files/rag/retrieval/scoring.js +0 -82
- package/dist-lib/lib/core/files/rag/retrieval/search.d.ts +0 -3
- package/dist-lib/lib/core/files/rag/retrieval/search.js +0 -49
- package/dist-lib/lib/core/files/rag/retrieval/selection.d.ts +0 -11
- package/dist-lib/lib/core/files/rag/retrieval/selection.js +0 -93
- package/dist-lib/lib/core/files/rag/retrieval/source-tag-filter.d.ts +0 -6
- package/dist-lib/lib/core/files/rag/retrieval/source-tag-filter.js +0 -79
- package/dist-lib/lib/core/files/rag/retrieval/sources.d.ts +0 -13
- package/dist-lib/lib/core/files/rag/retrieval/sources.js +0 -20
- package/dist-lib/lib/core/files/rag/retrieval/target.d.ts +0 -13
- package/dist-lib/lib/core/files/rag/retrieval/target.js +0 -89
- package/dist-lib/lib/core/files/rag/retrieval/terms.d.ts +0 -8
- package/dist-lib/lib/core/files/rag/retrieval/terms.js +0 -70
- package/dist-lib/lib/core/files/rag/retrieval/types.d.ts +0 -29
- package/dist-lib/lib/core/files/rag/retrieval/types.js +0 -1
- package/dist-lib/lib/core/files/rag/retrieval/web-sources.d.ts +0 -8
- package/dist-lib/lib/core/files/rag/retrieval/web-sources.js +0 -41
- package/dist-lib/lib/core/files/rag/retrieval/whole-document-query.d.ts +0 -1
- package/dist-lib/lib/core/files/rag/retrieval/whole-document-query.js +0 -96
- package/dist-lib/lib/core/files/rag/vector/memory.d.ts +0 -15
- package/dist-lib/lib/core/files/rag/vector/memory.js +0 -135
- package/dist-lib/lib/core/files/rag/vector/types.d.ts +0 -28
- package/dist-lib/lib/core/files/rag/vector/types.js +0 -1
|
@@ -1,243 +0,0 @@
|
|
|
1
|
-
---
|
|
2
|
-
name: pr-create
|
|
3
|
-
description: Write the title, body and commits for a pull request in @qvac/core, and decide the conventional-commit type for a change — `fix` is reserved for user-facing defects in a released version. Use when opening a PR here, splitting work into a stack, or choosing between `fix` and `chore`.
|
|
4
|
-
---
|
|
5
|
-
|
|
6
|
-
# PR creation
|
|
7
|
-
|
|
8
|
-
`@qvac/core` is a published package. The log is the only durable record of intent,
|
|
9
|
-
and for this repo it is also the changelog a consumer reads before upgrading.
|
|
10
|
-
|
|
11
|
-
## Title
|
|
12
|
-
|
|
13
|
-
```
|
|
14
|
-
type(scope): subject
|
|
15
|
-
```
|
|
16
|
-
|
|
17
|
-
Imperative, lowercase after the colon, no trailing period. Scope optional, and
|
|
18
|
-
only when it narrows something real. What and why, not how.
|
|
19
|
-
|
|
20
|
-
```
|
|
21
|
-
feat: publish the sidecar entry, so spawning it needs no bundle
|
|
22
|
-
fix(publish): repoint dynamic imports at dist-lib too
|
|
23
|
-
refactor: own the schema codegen tooling
|
|
24
|
-
chore: regenerate the schema
|
|
25
|
-
```
|
|
26
|
-
|
|
27
|
-
There is no ticket prefix here, and no title validator in CI. That is not
|
|
28
|
-
licence to be loose: nothing will catch a bad title but a reviewer.
|
|
29
|
-
|
|
30
|
-
## Types
|
|
31
|
-
|
|
32
|
-
```
|
|
33
|
-
build chore ci docs style refactor perf test fix feat
|
|
34
|
-
```
|
|
35
|
-
|
|
36
|
-
**`fix` is reserved for a defect a consumer of a released version could hit.**
|
|
37
|
-
It is the line someone reads to decide whether to upgrade, so it claims real
|
|
38
|
-
breakage in something already published.
|
|
39
|
-
|
|
40
|
-
| The change | Type |
|
|
41
|
-
| ----------------------------------------------------------- | -------------------------------------------------- |
|
|
42
|
-
| Breaks a consumer on a published version | `fix` |
|
|
43
|
-
| Repairs code that has never shipped (same release cycle) | `chore` |
|
|
44
|
-
| Repairs the build, the bundle, or what lands in the tarball | `build` or `fix` if a published artifact was wrong |
|
|
45
|
-
| Repairs CI, a workflow, an action pin | `ci` |
|
|
46
|
-
| Repairs or adds a test | `test` |
|
|
47
|
-
| Repairs a lint/type error | `chore` |
|
|
48
|
-
| Pure move or rename, no behaviour change | `refactor` |
|
|
49
|
-
| New capability, new export, new flag | `feat` |
|
|
50
|
-
| README, comments, docs | `docs` |
|
|
51
|
-
|
|
52
|
-
When a repair is on unreleased code, check the last release tag: if the bug was
|
|
53
|
-
never in a published version, it is `chore`, not `fix`.
|
|
54
|
-
|
|
55
|
-
Never relabel a real defect as `chore` to make the log quieter, and never
|
|
56
|
-
inflate a cleanup into `fix` to make it look urgent.
|
|
57
|
-
|
|
58
|
-
## Scopes
|
|
59
|
-
|
|
60
|
-
Optional, and taken from what the diff actually touches — `lib`, `tui`, `schema`, `spec`, `scripts`, `test`, `ci`, `publish`, or a domain name (`mesh`, `sync`, `assistant`, `storage`).
|
|
61
|
-
A scope that restates the type (`fix(fix)`) or names the whole repo adds
|
|
62
|
-
nothing; leave it off.
|
|
63
|
-
|
|
64
|
-
## Examples
|
|
65
|
-
|
|
66
|
-
`examples/` is where a consumer meets this library: `mesh-peer.mjs` stands a
|
|
67
|
-
Core and a Client up from nothing, `mesh-lab.mjs` drives scenarios against it.
|
|
68
|
-
They are read before `lib/` is.
|
|
69
|
-
|
|
70
|
-
**A PR that adds public API ships an example with it.** Public API means a new
|
|
71
|
-
entry in `package.json#exports`, a new export from `index.ts` or `client.ts`,
|
|
72
|
-
or a new capability that reaches the `Client` surface through `schema/`. Extend
|
|
73
|
-
an existing script or add one, whichever shows the capability in the fewest
|
|
74
|
-
lines someone can actually run.
|
|
75
|
-
|
|
76
|
-
An example is a runnable `.mjs` under `examples/` that imports the package the
|
|
77
|
-
way a consumer does. Give it a `package.json#scripts` entry when it is a
|
|
78
|
-
scenario run by name, and name it in the PR body.
|
|
79
|
-
|
|
80
|
-
Nothing else needs one — an internal module, a `tui/` command, a refactor, a
|
|
81
|
-
fix. Do not pad `examples/` to clear a checklist: one script a reader can run
|
|
82
|
-
beats three that restate the types.
|
|
83
|
-
|
|
84
|
-
## API docs
|
|
85
|
-
|
|
86
|
-
`docs/api/` carries one file per api module: every route's signature, its input
|
|
87
|
-
fields with what each is for, the mesh scope its handler honours, and a
|
|
88
|
-
changelog.
|
|
89
|
-
|
|
90
|
-
**A PR that changes a route's behaviour updates that route's file in the same
|
|
91
|
-
commit**, including its `### [Unreleased]` entry under the Keep a Changelog
|
|
92
|
-
heading it fits — `Fixed` for a defect, `Changed` for a contract that moved,
|
|
93
|
-
`Added` for a new route. The entry says what a caller has to do differently, not
|
|
94
|
-
what the diff did.
|
|
95
|
-
|
|
96
|
-
This is not optional documentation. The scope line is the only place that records
|
|
97
|
-
whether a handler honours `meshId`, and the changelog is what a consumer reads
|
|
98
|
-
to decide whether an upgrade touches them. A route whose behaviour moved and
|
|
99
|
-
whose file did not is a review finding.
|
|
100
|
-
|
|
101
|
-
Nothing else triggers it. An internal refactor behind an unchanged route, a
|
|
102
|
-
`tui/` change, a test — no doc change, no finding.
|
|
103
|
-
|
|
104
|
-
## Decision records
|
|
105
|
-
|
|
106
|
-
`docs/decisions/` holds the durable, non-obvious choices — why a boundary is
|
|
107
|
-
where it is, why an approach was rejected. They are reference for a future
|
|
108
|
-
reader, **not rules, and not to be loaded by default**.
|
|
109
|
-
|
|
110
|
-
Add one when this PR settles something a later reader would otherwise have to
|
|
111
|
-
reverse-engineer from the diff: a runtime or storage boundary, a wire-format or
|
|
112
|
-
schema choice, a build or publish constraint, a rejected alternative that will
|
|
113
|
-
look obvious later. Do not add one for a bug fix, a rename, or anything the code
|
|
114
|
-
already says plainly.
|
|
115
|
-
|
|
116
|
-
Copy `docs/decisions/TEMPLATE.md`, give it a `status`, list it once in
|
|
117
|
-
`INDEX.md`, and name it in the PR's Decisions section. `node
|
|
118
|
-
scripts/docs-audit.cjs` checks all of that, and CI runs it.
|
|
119
|
-
|
|
120
|
-
An **active** record may cite only paths that exist here; a path in another
|
|
121
|
-
repository is written `@qvac/harness → lib/x.ts` so the audit treats it as
|
|
122
|
-
context rather than a local file. A superseded record is exempt — it describes
|
|
123
|
-
how things used to be.
|
|
124
|
-
|
|
125
|
-
### One PR, one type
|
|
126
|
-
|
|
127
|
-
The commits carry conventional-commit types, so the mix of types _is_ the scope
|
|
128
|
-
statement. `feat:`, `fix:` and `chore:` on one branch is not a well-rounded
|
|
129
|
-
change, it is three changes sharing a branch, and a reviewer has to hold all
|
|
130
|
-
three while judging any one of them.
|
|
131
|
-
|
|
132
|
-
List them before writing the body:
|
|
133
|
-
|
|
134
|
-
```bash
|
|
135
|
-
git log --format='%s' main..HEAD | grep -oE '^[a-z]+' | sort -u
|
|
136
|
-
```
|
|
137
|
-
|
|
138
|
-
- **One type** is the PR's type.
|
|
139
|
-
- **A dominant type plus its own tests or docs** (`feat` + `test`, `fix` +
|
|
140
|
-
`docs`) is fine; the dominant one wins.
|
|
141
|
-
- **Unrelated types** means split the PR, or stack it.
|
|
142
|
-
|
|
143
|
-
## Small PRs, stacked
|
|
144
|
-
|
|
145
|
-
A PR is read by a human, so it has to fit in one head. Default to one concern.
|
|
146
|
-
When a change decomposes into dependent steps, do not grow one PR to fit them —
|
|
147
|
-
stack them, so each step is reviewable alone and lands in order.
|
|
148
|
-
|
|
149
|
-
- A large refactor belongs on a `refactor/...` branch with `refactor:` commits.
|
|
150
|
-
- **Refactor commits never live inside a non-refactor branch.** A `feat/` branch
|
|
151
|
-
that also reshuffles unrelated code is two PRs. Behaviour changes and pure
|
|
152
|
-
moves are reviewed differently, and mixing them hides both.
|
|
153
|
-
- A pure rename or move lands _before_ the change that depends on it, as its own
|
|
154
|
-
commit or its own PR in the stack.
|
|
155
|
-
|
|
156
|
-
Branch names must never contain `codex`, `cursor`, or `claude` (case
|
|
157
|
-
insensitive). Use a neutral name before committing or pushing.
|
|
158
|
-
|
|
159
|
-
## Body
|
|
160
|
-
|
|
161
|
-
`.github/pull_request_template.md` pre-fills the headings. Keep its shape, fill
|
|
162
|
-
what applies, and delete the sections that would be empty — an absent section
|
|
163
|
-
says more than one saying "n/a". The template's own order is Why, What, Testing,
|
|
164
|
-
Decisions; the expanded form below is what each should contain.
|
|
165
|
-
|
|
166
|
-
```markdown
|
|
167
|
-
## What
|
|
168
|
-
|
|
169
|
-
One paragraph. The change, in the reviewer's terms.
|
|
170
|
-
|
|
171
|
-
## Why
|
|
172
|
-
|
|
173
|
-
The problem this solves, or the decision it implements. Link the issue,
|
|
174
|
-
PR, or upstream change that forced it.
|
|
175
|
-
|
|
176
|
-
## How
|
|
177
|
-
|
|
178
|
-
Only the parts a reader would not infer from the diff: the seam that moved,
|
|
179
|
-
the invariant being preserved, the thing that looks wrong but isn't.
|
|
180
|
-
|
|
181
|
-
## Testing
|
|
182
|
-
|
|
183
|
-
What you ran, and what it proved. ``bun run test` (node, bare and addon lanes)` at minimum.
|
|
184
|
-
Say plainly when something could not be tested locally and why.
|
|
185
|
-
```
|
|
186
|
-
|
|
187
|
-
For a **breaking change**, say so in the body under its own `## Breaking`
|
|
188
|
-
heading, with the migration a consumer has to perform. Published package —
|
|
189
|
-
this is the only warning they get.
|
|
190
|
-
|
|
191
|
-
### Length, and the shape it has to hold
|
|
192
|
-
|
|
193
|
-
Twenty lines. The bodies in this log run 17–28 (`#134`, `#133`, `#130`), and a
|
|
194
|
-
reviewer reads the diff, not the description.
|
|
195
|
-
|
|
196
|
-
- **Use the template's headings and only those.** Delete what does not apply;
|
|
197
|
-
do not invent `## How I decided`, `## What is unresolved`, `## Is this a
|
|
198
|
-
contract change`. A point that needs its own heading is usually a point that
|
|
199
|
-
belongs in the commit message.
|
|
200
|
-
- **Never hard-wrap.** GitHub renders a single newline as a line break in PR
|
|
201
|
-
bodies, so text wrapped at 80 columns displays as a narrow ragged column
|
|
202
|
-
instead of filling the width. One bullet is one line, however long.
|
|
203
|
-
- **No process narrative.** "Checked rather than assumed", "found by
|
|
204
|
-
intersecting two facts", "I have not established" — how you arrived at the
|
|
205
|
-
change is not what a reviewer needs. State what is true.
|
|
206
|
-
- **No open-question lists.** A body that ends in four things somebody should
|
|
207
|
-
think about is a change that was not finished. Decide them, or say plainly in
|
|
208
|
-
one line what is deliberately left and why.
|
|
209
|
-
- **Depth goes in the commit message.** It is the durable record and it is where
|
|
210
|
-
a rejected alternative, a subtle invariant or a caveat belongs. The body is
|
|
211
|
-
the summary a reviewer reads first, not the second copy.
|
|
212
|
-
|
|
213
|
-
## Before opening
|
|
214
|
-
|
|
215
|
-
1. Check the branch is not `main` and the name is neutral.
|
|
216
|
-
2. `git log --oneline main..HEAD` — every commit is a conventional commit, and
|
|
217
|
-
each one is a change you would defend on its own. A trail of `wip`,
|
|
218
|
-
`fixup` or `address review` commits is not history; reshape it before
|
|
219
|
-
opening.
|
|
220
|
-
3. New public API? `examples/` shows it, and the body names the script.
|
|
221
|
-
4. `bun run lint` and ``bun run test` (node, bare and addon lanes)` pass locally. CI runs the same lanes; a
|
|
222
|
-
red PR wastes a reviewer's round trip.
|
|
223
|
-
5. Rebase on `main` rather than merging it in. A merge commit in the branch
|
|
224
|
-
makes the diff unreadable.
|
|
225
|
-
6. Open as a **draft** unless you are asking for review right now.
|
|
226
|
-
|
|
227
|
-
```bash
|
|
228
|
-
gh pr create --repo tetherto/qvac-core --draft --title "type(scope): subject" --body-file <path>
|
|
229
|
-
```
|
|
230
|
-
|
|
231
|
-
Never push to `main`, never force-push a branch someone else is reviewing, and
|
|
232
|
-
never open a PR that carries commits you did not write without saying so.
|
|
233
|
-
|
|
234
|
-
## Version bumps
|
|
235
|
-
|
|
236
|
-
Release PRs in this repo are titled by version alone (`0.2.4 (#32)`,
|
|
237
|
-
`release/0.1.4 (#40)`). They carry the version bump and nothing else — do not
|
|
238
|
-
fold a bump into a feature PR, and do not fold a feature into a bump.
|
|
239
|
-
|
|
240
|
-
## Sibling skills
|
|
241
|
-
|
|
242
|
-
- Reviewing a PR or a local diff → **pr-review**.
|
|
243
|
-
- How the code itself should be shaped → **js-house-style**.
|
|
@@ -1,428 +0,0 @@
|
|
|
1
|
-
---
|
|
2
|
-
name: pr-review
|
|
3
|
-
description: Deep-dive review of a pull request in @qvac/core, or of local uncommitted/branch changes when invoked with `local`. Checks CI and the base branch, the commit type and title, correctness and state/derived-data bugs, contract and API mismatches, security, code quality against the house style, and missing tests. In PR mode it creates PENDING inline comments only after the user confirms them, and never submits the review event. In local mode it prints findings in chat and posts nothing. Use when reviewing a PR, given a PR link, or asked to review the working tree or current branch.
|
|
4
|
-
---
|
|
5
|
-
|
|
6
|
-
# PR review
|
|
7
|
-
|
|
8
|
-
Reviews changes in `@qvac/core`. Two modes.
|
|
9
|
-
|
|
10
|
-
- **PR mode** (a PR number or URL) — a short overview in chat, then a **pending**
|
|
11
|
-
GitHub review of per-file inline comments. The pending review is created only
|
|
12
|
-
after the user confirms the exact set of comments. **Never submit the review
|
|
13
|
-
event** (`COMMENT`, `REQUEST_CHANGES`, `APPROVE`) — that is the user's call,
|
|
14
|
-
from the GitHub UI.
|
|
15
|
-
- **Local mode** (`local`, or "review my changes") — the overview in chat only.
|
|
16
|
-
Posts nothing, anywhere.
|
|
17
|
-
|
|
18
|
-
## Reviewing is read-only
|
|
19
|
-
|
|
20
|
-
This skill never writes to the working tree. Against the user's repo these are
|
|
21
|
-
forbidden, whatever the reason:
|
|
22
|
-
|
|
23
|
-
- `git switch`, `git checkout`, `git reset`, `git restore`, `git clean`
|
|
24
|
-
- `git stash`, in any form
|
|
25
|
-
- `git pull`, `git merge`, `git rebase`, `git cherry-pick`
|
|
26
|
-
- `gh pr checkout`
|
|
27
|
-
- any edit to a file
|
|
28
|
-
|
|
29
|
-
Local mode reads the diff with `git diff`, `git status` and `git ls-files`, and
|
|
30
|
-
nothing else. If the tree is dirty in a way that makes the diff hard to read, say
|
|
31
|
-
so and let the user decide; never tidy it yourself. A review that loses
|
|
32
|
-
uncommitted work is worse than no review.
|
|
33
|
-
|
|
34
|
-
PR mode needs no checkout at all: `gh pr diff` gives the patch, and
|
|
35
|
-
`gh api repos/{owner}/{repo}/contents/{path}?ref={sha}` gives any file at the PR
|
|
36
|
-
head.
|
|
37
|
-
|
|
38
|
-
## Gather
|
|
39
|
-
|
|
40
|
-
PR mode:
|
|
41
|
-
|
|
42
|
-
```bash
|
|
43
|
-
gh pr view <num> --repo tetherto/qvac-core \
|
|
44
|
-
--json title,body,author,headRefName,headRefOid,baseRefName,files,additions,deletions,state,isDraft,mergeable,statusCheckRollup \
|
|
45
|
-
> /tmp/pr-<num>.json
|
|
46
|
-
gh pr diff <num> --repo tetherto/qvac-core --patch > /tmp/pr-<num>.patch
|
|
47
|
-
```
|
|
48
|
-
|
|
49
|
-
Two calls, then read both files with Read/Grep rather than re-running `gh` — it
|
|
50
|
-
is the same bytes every time and the files are greppable. `statusCheckRollup`
|
|
51
|
-
already carries every check, so **do not run `gh pr checks`**; it buys nothing
|
|
52
|
-
the JSON does not have.
|
|
53
|
-
|
|
54
|
-
Local mode. The default scope is everything uncommitted — staged, unstaged and
|
|
55
|
-
untracked:
|
|
56
|
-
|
|
57
|
-
```bash
|
|
58
|
-
git diff --staged > /tmp/local-staged.patch
|
|
59
|
-
git diff > /tmp/local-unstaged.patch
|
|
60
|
-
git status --porcelain=v1
|
|
61
|
-
git ls-files --others --exclude-standard # untracked, easy to miss
|
|
62
|
-
```
|
|
63
|
-
|
|
64
|
-
Scope overrides, when the user asks for one: `--staged` is `git diff --staged`
|
|
65
|
-
alone, `--unstaged` is `git diff` plus untracked, and `--branch [<base>]` is
|
|
66
|
-
`git diff <base>...HEAD` (three dots — the merge base, not the tip), with
|
|
67
|
-
`<base>` defaulting to `main`.
|
|
68
|
-
|
|
69
|
-
Untracked files are reviewed as wholly new content: read them from the working
|
|
70
|
-
tree, they appear in no patch.
|
|
71
|
-
|
|
72
|
-
**If the requested scope is empty, say so and stop.** An empty `--staged` is
|
|
73
|
-
not an invitation to review something else, and it is certainly not an
|
|
74
|
-
invitation to invent findings.
|
|
75
|
-
|
|
76
|
-
## CI
|
|
77
|
-
|
|
78
|
-
One workflow gates this repo: **Integrate** (`.github/workflows/ci.yml`), a
|
|
79
|
-
single `Lint & Tests` job. `publish.yml` runs on release, not on PRs.
|
|
80
|
-
|
|
81
|
-
Its steps are each a separate way to fail, and the last three catch what a
|
|
82
|
-
local `bun run test` does not:
|
|
83
|
-
|
|
84
|
-
| Step | Command | Fails when |
|
|
85
|
-
| ------------------------ | ---------------------- | --------------------------------------------------------------------------------------------- |
|
|
86
|
-
| Lint | `bun run lint` | prettier, lunte or `tsc --noEmit` |
|
|
87
|
-
| Schema is regenerated | `bun run build:schema` | `spec/` is dirty after codegen — someone hand-edited it, or forgot to commit the regeneration |
|
|
88
|
-
| Tests | `bun run test` | the node, bare and addon lanes |
|
|
89
|
-
| Library surface compiles | `bun run build:lib` | the compiled export surface breaks |
|
|
90
|
-
|
|
91
|
-
A PR that touches `schema/` and leaves `spec/` stale is the most common red
|
|
92
|
-
here, and it looks nothing like a test failure.
|
|
93
|
-
|
|
94
|
-
Read the result from `statusCheckRollup` in `/tmp/pr-<num>.json`, and pull logs
|
|
95
|
-
only for a job that actually failed:
|
|
96
|
-
|
|
97
|
-
```bash
|
|
98
|
-
gh run view --repo tetherto/qvac-core --log-failed --job <id> > /tmp/pr-<num>-<id>.log
|
|
99
|
-
```
|
|
100
|
-
|
|
101
|
-
A red `Integrate` is a blocking finding on its own. Read the failing job's log
|
|
102
|
-
before writing a single line of review — half of what looks like a design
|
|
103
|
-
problem is a lint rule.
|
|
104
|
-
|
|
105
|
-
### Base branch
|
|
106
|
-
|
|
107
|
-
`baseRefName` is in the JSON you already fetched. Per CLAUDE.md a PR targets
|
|
108
|
-
`main`; the one exception is a stacked PR, which branches from the PR below it
|
|
109
|
-
and targets that branch.
|
|
110
|
-
|
|
111
|
-
So a base other than `main` is only a finding when the PR is **not** part of a
|
|
112
|
-
stack — check whether the base branch is itself an open PR before raising it.
|
|
113
|
-
If it is, the stack is correct and there is nothing to say. If it is not, that
|
|
114
|
-
is a `[P2]`: the diff will contain the base branch's commits and the PR cannot
|
|
115
|
-
merge as it stands.
|
|
116
|
-
|
|
117
|
-
## What to check, in order
|
|
118
|
-
|
|
119
|
-
Stop at the first tier that produces findings worth the author's time; a review
|
|
120
|
-
that lists twelve nits under a broken contract has buried the real thing.
|
|
121
|
-
|
|
122
|
-
**1. Correctness.** Does it do what the title claims? Trace the changed path
|
|
123
|
-
end to end, including the error path. Look for: a promise not awaited, a
|
|
124
|
-
listener never removed, a resource opened on one path and closed on another, an
|
|
125
|
-
error swallowed into a default.
|
|
126
|
-
|
|
127
|
-
A **regression** belongs here too: a change that reads as additive but alters
|
|
128
|
-
what existing callers get — a signature widened, a default flipped, a branch
|
|
129
|
-
removed, a return value that is semantically different at the same type.
|
|
130
|
-
|
|
131
|
-
**2. State and derived data.** The class of bug a linter cannot see and a diff
|
|
132
|
-
skim misses. Walk any new stateful or arithmetic code against these four.
|
|
133
|
-
|
|
134
|
-
- **Two sources of truth.** The same value in two fields, or the same
|
|
135
|
-
expression computed twice, in one payload. Before writing the comment, read
|
|
136
|
-
the type declaration **and** the consumer: if the fields have distinct
|
|
137
|
-
documented meanings (per-item vs aggregate, raw vs normalised) and a reader
|
|
138
|
-
prefers one, this is a value written to the wrong channel — a contract
|
|
139
|
-
violation — not duplication. Say that, and cite the contract. Only call it
|
|
140
|
-
duplication when nothing distinguishes the fields. The two look identical in
|
|
141
|
-
a diff and have opposite fixes: one keeps both fields and moves a value, the
|
|
142
|
-
other deletes a field. Backwards, and you send the author to delete a channel
|
|
143
|
-
something depends on. `[P2]` either way.
|
|
144
|
-
- **Incremental accumulation.** A running total or offset advanced inside a
|
|
145
|
-
loop, where computing the whole table once would be correct by construction.
|
|
146
|
-
`[P2]` when a later edit could desynchronise it — and the fix shrinks the
|
|
147
|
-
code.
|
|
148
|
-
- **Closure over mutated state.** A callback capturing a variable a later step
|
|
149
|
-
reassigns, especially across an `await`. `[P1]` if a late or reordered
|
|
150
|
-
invocation yields a wrong value; say so plainly, because it will not
|
|
151
|
-
reproduce locally.
|
|
152
|
-
- **Laundered values.** `?? 0`, `|| 0`, `Math.max(x, 0)` turning an impossible
|
|
153
|
-
input into a plausible number instead of letting it fail. `[P2]` unless the
|
|
154
|
-
clamp buys a stated presentational guarantee. Watch an optional `uint` in
|
|
155
|
-
`schema/`: it decodes to `0`, not `null`, so `??` never sees the absent
|
|
156
|
-
value.
|
|
157
|
-
|
|
158
|
-
**3. Contract and API.** This is a published package, so an exported signature
|
|
159
|
-
is a promise to a consumer.
|
|
160
|
-
|
|
161
|
-
- Anything added to or changed in the export map, `index.ts`, `types/`, or
|
|
162
|
-
`package.json#exports` — is it intentional, and is it a breaking change?
|
|
163
|
-
- `schema/` or `spec/` changes: is the wire format still compatible, and was the
|
|
164
|
-
baseline regenerated (`bun run build:schema`) rather than hand-edited?
|
|
165
|
-
- Does the change match its own type declarations, and do callers in this repo
|
|
166
|
-
still typecheck?
|
|
167
|
-
- A route whose behaviour moved: is `docs/api/<module>.md` updated with it — the
|
|
168
|
-
scope line if the mesh handling changed, and an entry under `### [Unreleased]`?
|
|
169
|
-
That file is where a consumer looks before upgrading, so a stale one is a `[P2]`.
|
|
170
|
-
A refactor behind an unchanged route needs nothing.
|
|
171
|
-
|
|
172
|
-
**4. House style.** Load **js-house-style** and run the changed files against
|
|
173
|
-
it — that document, not your own taste or your memory of it. Layering,
|
|
174
|
-
composition on primitives, module size and shape, no defensive programming,
|
|
175
|
-
comment density. A style point the document does not back is not a finding.
|
|
176
|
-
|
|
177
|
-
**5. Tests.** A behaviour change with no test is a finding. Name the missing
|
|
178
|
-
case rather than asking for "more tests": the tests live under `test/`, run on
|
|
179
|
-
both the node and bare lanes, and a fix without a regression guard will come
|
|
180
|
-
back.
|
|
181
|
-
|
|
182
|
-
**6. Examples.** This is a library, and `examples/` is where a consumer learns
|
|
183
|
-
its surface — `mesh-peer.mjs` for standing up a Core and a Client,
|
|
184
|
-
`mesh-lab.mjs` for scenarios on top. New public API with nothing in `examples/`
|
|
185
|
-
is a finding: `[P2]` for a new entry in `package.json#exports` or a new export
|
|
186
|
-
from `index.ts` / `client.ts`, `[P3]` for a capability folded into an area an
|
|
187
|
-
existing script already covers. Name the script that should have changed and
|
|
188
|
-
the one thing it should show; anchor the comment on the new export line, since
|
|
189
|
-
the absence itself has no line.
|
|
190
|
-
|
|
191
|
-
Nothing else triggers this. An internal module, a `tui/` command, a refactor, a
|
|
192
|
-
fix, a test — no example, no finding. Do not ask for an example because a diff
|
|
193
|
-
is large.
|
|
194
|
-
|
|
195
|
-
**7. Commit type and title.** Per **pr-create**: is this really a `fix` (a
|
|
196
|
-
defect a consumer of a released version could hit), or is it `chore` on code
|
|
197
|
-
that never shipped? Is the title imperative and specific? Does the branch name
|
|
198
|
-
avoid `codex` / `cursor` / `claude`?
|
|
199
|
-
|
|
200
|
-
**8. Scope.** Does the PR do one thing? A behaviour change mixed with an
|
|
201
|
-
unrelated rename is two PRs — say so, and say which commit to split out.
|
|
202
|
-
|
|
203
|
-
**9. Security.** Not a sweep of the file — only what this diff introduces. A
|
|
204
|
-
secret or key material in code, a log line or an error message; a peer-supplied
|
|
205
|
-
value trusted without validation at a trust boundary; unsafe deserialization of
|
|
206
|
-
anything off the wire; crypto used in a way the primitive does not promise
|
|
207
|
-
(a reused nonce, a comparison that is not constant-time, a key derived from
|
|
208
|
-
something predictable); a new dependency pulled in for something small.
|
|
209
|
-
|
|
210
|
-
This is a P2P library, so "the input came from another device" is the normal
|
|
211
|
-
case, not the exceptional one. Where it matters, `[P1]`.
|
|
212
|
-
|
|
213
|
-
**10. Decision records.** Only when the diff **contradicts** a record in
|
|
214
|
-
`docs/decisions/` that you have actually read — then the record is updated or
|
|
215
|
-
superseded in the same PR rather than left to rot, and that is a `[P2]`.
|
|
216
|
-
|
|
217
|
-
A missing record is not a finding. Do not open `docs/decisions/INDEX.md`
|
|
218
|
-
looking for something to say, do not ask for an ADR because a change "feels
|
|
219
|
-
architectural", and never make "no decision record" the only thing a review
|
|
220
|
-
found. If you genuinely think a lasting choice went unrecorded, say it once in
|
|
221
|
-
the chat overview as a question, never inline.
|
|
222
|
-
|
|
223
|
-
### Do not over-engineer the review
|
|
224
|
-
|
|
225
|
-
The most damaging thing a review can do is talk a good PR into a worse one. AI
|
|
226
|
-
review skews defensive: a guard here, an abstraction there, a layer to absorb a
|
|
227
|
-
hypothetical, each one permanent debt bolted to a problem that never happened.
|
|
228
|
-
That runs directly against the house style, which does not program defensively.
|
|
229
|
-
|
|
230
|
-
It compounds. Successive reviews each add a layer against a different
|
|
231
|
-
speculative case, one review's fix undoes another's, and the code ends up in a
|
|
232
|
-
corner nobody chose.
|
|
233
|
-
|
|
234
|
-
- **Only raise what is wrong now**, or has a concrete path to being wrong.
|
|
235
|
-
"Someone could later..." is not a finding.
|
|
236
|
-
- **Never propose a guard, wrapper, fallback or abstraction** whose only job is
|
|
237
|
-
to make a hypothetical safe. Let caller bugs crash; that is the contract.
|
|
238
|
-
- **Prefer the fix that deletes code.** If your suggestion makes the diff
|
|
239
|
-
bigger, be sure it earns it.
|
|
240
|
-
- **Simplification is a valid finding; speculative hardening is not.**
|
|
241
|
-
- When you cannot tell whether something is real, ask instead of requesting a
|
|
242
|
-
change, and be willing to land zero findings.
|
|
243
|
-
|
|
244
|
-
## Write the findings
|
|
245
|
-
|
|
246
|
-
Each finding: the file and line, what breaks, and the concrete scenario that
|
|
247
|
-
breaks it. No finding without a failure mode you can state.
|
|
248
|
-
|
|
249
|
-
Lead with the findings, ordered by severity.
|
|
250
|
-
|
|
251
|
-
### Only raise what is worth the author's time
|
|
252
|
-
|
|
253
|
-
A review is not required to produce findings. **"Nothing to flag — the diff is
|
|
254
|
-
clean" is a complete, successful review**, and a good PR reviewed honestly is
|
|
255
|
-
the expected outcome most of the time. Never go hunting for something to say
|
|
256
|
-
because the output looks empty; an invented finding costs the author more than
|
|
257
|
-
a silent review ever could.
|
|
258
|
-
|
|
259
|
-
Every finding passes this bar before it is written down:
|
|
260
|
-
|
|
261
|
-
- You can name **what breaks**, and the concrete path that reaches it.
|
|
262
|
-
- Acting on it **changes the code, the reviewer's decision, or a consumer's
|
|
263
|
-
experience**.
|
|
264
|
-
- It is backed by the house style or a real contract, not your taste.
|
|
265
|
-
|
|
266
|
-
Drop everything else. Not "mention it as a `[P3]`" — **drop it**. A finding you
|
|
267
|
-
cannot defend at `[P2]` is not thereby a `[P3]`; it is noise wearing a smaller
|
|
268
|
-
tag. `[P3]` is for a real, small, correct observation, not a holding pen for
|
|
269
|
-
weak `[P1]`/`[P2]`s.
|
|
270
|
-
|
|
271
|
-
The failure mode this repo has actually seen: a clean PR comes back with four
|
|
272
|
-
`[P3]`s about naming, a speculative race nobody can reach, and a request for a
|
|
273
|
-
decision record. That review is worse than no review — it trains the author to
|
|
274
|
-
skim the next one, and the real finding gets skimmed with it.
|
|
275
|
-
|
|
276
|
-
### Severity tags
|
|
277
|
-
|
|
278
|
-
Tag every finding `[P1]`, `[P2]` or `[P3]` **in the chat overview**, using the
|
|
279
|
-
tag rather than a word, so two reviews of the same repo rank things the same
|
|
280
|
-
way. The tag never appears on the PR — see "Comment style".
|
|
281
|
-
|
|
282
|
-
- `[P1]` — almost-certain bug, data corruption, security issue, CI breaker, or a
|
|
283
|
-
break of behaviour a consumer relies on. In the overview, and **proposed inline
|
|
284
|
-
by default**.
|
|
285
|
-
- `[P2]` — likely bug, non-obvious caveat, contract or API mismatch, subtle
|
|
286
|
-
behaviour change, missing test on a risky path. In the overview, and **proposed
|
|
287
|
-
inline by default**.
|
|
288
|
-
- `[P3]` — minor: a stale comment or doc, an ergonomic improvement, an optional
|
|
289
|
-
cleanup. In the overview when material, and **not proposed inline** unless the
|
|
290
|
-
user asks for them.
|
|
291
|
-
|
|
292
|
-
**Never silently put a `[P3]` in the inline payload.** A review that posts ten
|
|
293
|
-
nits inline teaches the author to skim the next one.
|
|
294
|
-
|
|
295
|
-
**Severity is what the code does today, not what it might do.** A finding whose
|
|
296
|
-
only support is a scenario nobody has hit is not a `[P2]`: either find the path
|
|
297
|
-
that reaches it, or drop it — do not park it at `[P3]` instead. Zero findings is
|
|
298
|
-
a legitimate result.
|
|
299
|
-
|
|
300
|
-
If the user says this is a re-review, say explicitly which earlier concerns are
|
|
301
|
-
resolved and which still stand.
|
|
302
|
-
|
|
303
|
-
### The overview
|
|
304
|
-
|
|
305
|
-
The user's checkout is usually on a different branch or commit than the PR, so
|
|
306
|
-
a bare line number is worth nothing to them. Every `[P1]` and `[P2]` in the
|
|
307
|
-
overview therefore carries two things: a **3–8 line excerpt read at the head
|
|
308
|
-
SHA**, and a **link to that line on the PR**.
|
|
309
|
-
|
|
310
|
-
The excerpt comes from
|
|
311
|
-
`gh api -H "Accept: application/vnd.github.raw" repos/tetherto/qvac-core/contents/<path>?ref=<headRefOid>`,
|
|
312
|
-
never from the local tree. The link is the file-diff anchor:
|
|
313
|
-
|
|
314
|
-
`https://github.com/tetherto/qvac-core/pull/<num>/files#diff-<sha256(path)>R<line>`
|
|
315
|
-
|
|
316
|
-
```bash
|
|
317
|
-
printf '%s' 'lib/chat/compact.js' | shasum -a 256 | awk '{print $1}'
|
|
318
|
-
```
|
|
319
|
-
|
|
320
|
-
If the anchor will not compute, fall back to
|
|
321
|
-
`https://github.com/tetherto/qvac-core/blob/<headRefOid>/<path>#L<line>`.
|
|
322
|
-
|
|
323
|
-
````markdown
|
|
324
|
-
## PR #<num> — review overview
|
|
325
|
-
|
|
326
|
-
<one line: what this PR does>
|
|
327
|
-
|
|
328
|
-
### CI / base
|
|
329
|
-
|
|
330
|
-
<one line; omit entirely when both are clean>
|
|
331
|
-
|
|
332
|
-
### Findings
|
|
333
|
-
|
|
334
|
-
- **[P1]** <short title> — [`<path>:<line>`](link)
|
|
335
|
-
|
|
336
|
-
<one sentence, plus the repro when there is one>
|
|
337
|
-
|
|
338
|
-
```js
|
|
339
|
-
<3-8 line excerpt at the head SHA>
|
|
340
|
-
```
|
|
341
|
-
|
|
342
|
-
- **[P3]** <short title> — [`<path>:<line>`](link) — <one sentence, no excerpt>
|
|
343
|
-
|
|
344
|
-
### Re-review delta
|
|
345
|
-
|
|
346
|
-
<only when the user said this is a re-review>
|
|
347
|
-
|
|
348
|
-
- Resolved: <…>
|
|
349
|
-
- Still open: <…>
|
|
350
|
-
````
|
|
351
|
-
|
|
352
|
-
One sentence per finding here; the inline comment carries the reasoning. A
|
|
353
|
-
clean review is this same block with `Findings` reading "nothing to flag".
|
|
354
|
-
|
|
355
|
-
Local mode uses the same shape, headed `## Local review — <scope>`, with plain
|
|
356
|
-
`path:line` instead of links and the excerpt read from the working tree.
|
|
357
|
-
|
|
358
|
-
## Comment style
|
|
359
|
-
|
|
360
|
-
Write like a colleague leaving a note, not a static analyser filing a report.
|
|
361
|
-
If a human reviewer would not say it out loud, do not write it.
|
|
362
|
-
|
|
363
|
-
- **Brief.** A few lines, one point per comment. If it takes three paragraphs
|
|
364
|
-
to justify, it probably is not a finding.
|
|
365
|
-
- **Ask, don't lecture.** "Can this reuse the existing helper?" beats a
|
|
366
|
-
paragraph on why the new one duplicates knowledge the store already owns.
|
|
367
|
-
- **Link what you refer to.** An existing function, type or rule you want
|
|
368
|
-
reused gets a line link — `[name](<url>#L120)`. A reference the author has to
|
|
369
|
-
hunt for is one they skip.
|
|
370
|
-
- **Show the change in code.** A short before/after, or a `suggestion` block
|
|
371
|
-
whenever the fix is mechanical enough to apply as-is.
|
|
372
|
-
- **Never put a severity marker in a comment body.** No `[P1]`/`[P2]`/`[P3]`,
|
|
373
|
-
no "blocker", no "nit:". Severity is how _you_ decided what to raise; it
|
|
374
|
-
lives in the chat overview and nowhere else. On the PR it reads as grading
|
|
375
|
-
the author and it ages badly once the code moves. If something is minor, say
|
|
376
|
-
it in plain words — "not urgent", "while you're in there".
|
|
377
|
-
- One concrete repro line for a `[P1]`/`[P2]` when one exists. Not a test plan.
|
|
378
|
-
- No praise padding, no "LGTM but…", no restating what the diff does.
|
|
379
|
-
- **Sign off every comment with 🤖 on its own last line**, so it is
|
|
380
|
-
unmistakable that the review came from an AI. Every inline comment, no
|
|
381
|
-
exceptions — the emoji alone, nothing else on the line.
|
|
382
|
-
|
|
383
|
-
## Posting (PR mode only)
|
|
384
|
-
|
|
385
|
-
1. Print every proposed comment in chat first, as it will be posted: file,
|
|
386
|
-
line, then the verbatim body. Carry the severity in the chat line **outside**
|
|
387
|
-
the body. The default set is the `[P1]` and `[P2]` findings; name any `[P3]`
|
|
388
|
-
you are holding back so the user can pull it in. If there is nothing at
|
|
389
|
-
`[P1]`/`[P2]`, say so and stop — never assemble a payload out of `[P3]`s to
|
|
390
|
-
have something to post.
|
|
391
|
-
2. Ask the user to confirm the exact set. Wait for a clear yes.
|
|
392
|
-
3. Write `/tmp/pr-<num>-review.json` from the confirmed set **only**. If the
|
|
393
|
-
confirmed set is empty, stop — never post an empty review.
|
|
394
|
-
4. Pre-flight, before the POST:
|
|
395
|
-
- The comment count matches the confirmed set **exactly**. If it does not,
|
|
396
|
-
stop and reconcile; do not post the larger set and mention it after.
|
|
397
|
-
- Every `line` is a **post-PR line** — the `+` side of the patch, as
|
|
398
|
-
numbered in the file at the head SHA. A line the patch does not touch is a
|
|
399
|
-
`422` that rejects the whole review, so anchor to a touched line or move
|
|
400
|
-
the point to the chat overview.
|
|
401
|
-
- Every body ends with a `🤖` line, and none contains `[P1]`, `[P2]`, `[P3]`,
|
|
402
|
-
"nit" or "blocker". Fix the bodies yourself; do not ask the user.
|
|
403
|
-
- Show the bodies as **rendered Markdown**, one per `file:line`, never raw
|
|
404
|
-
JSON — the user is approving what the author will read.
|
|
405
|
-
5. Create the review as **PENDING**, with no event:
|
|
406
|
-
|
|
407
|
-
```bash
|
|
408
|
-
gh api repos/tetherto/qvac-core/pulls/<num>/reviews \
|
|
409
|
-
--method POST --input /tmp/pr-<num>-review.json
|
|
410
|
-
```
|
|
411
|
-
|
|
412
|
-
where the payload has `commit_id` (the `headRefOid` you fetched) and a
|
|
413
|
-
`comments` array of `{path, line, side, body}`, and **no `event` field**. A
|
|
414
|
-
payload with `event` submits the review; leaving it out is what makes it
|
|
415
|
-
pending. No top-level `body` either, unless the user asks for one — the inline
|
|
416
|
-
comments are the review.
|
|
417
|
-
|
|
418
|
-
If the user edits the selection after the preview, rewrite the file before
|
|
419
|
-
running the command.
|
|
420
|
-
|
|
421
|
-
6. Hand the user the link and stop:
|
|
422
|
-
`https://github.com/tetherto/qvac-core/pull/<num>#pullrequestreview-<id>`
|
|
423
|
-
|
|
424
|
-
## Sibling skills
|
|
425
|
-
|
|
426
|
-
- Writing the PR being reviewed → **pr-create**.
|
|
427
|
-
- The rules tier 4 reviews against → **js-house-style**. Load it; do not review
|
|
428
|
-
from memory of it.
|