squally-mcp 0.8.0 → 0.8.1
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/README.md +321 -321
- package/dist/openapi/v1.json +384 -549
- package/package.json +63 -63
package/README.md
CHANGED
|
@@ -1,321 +1,321 @@
|
|
|
1
|
-
# squally-mcp
|
|
2
|
-
|
|
3
|
-
A local [MCP](https://modelcontextprotocol.io) server that lets a coding agent
|
|
4
|
-
read your [Squally](https://app.squally.dev) data: which runs happened, how
|
|
5
|
-
stable each test is, and why a particular test failed.
|
|
6
|
-
|
|
7
|
-
It runs on your machine over stdio and talks to Squally's read API over HTTPS.
|
|
8
|
-
Point Claude Code, Codex, Cursor or Claude Desktop at it and ask "why is
|
|
9
|
-
`checkout.spec.ts` failing on main?".
|
|
10
|
-
|
|
11
|
-
## Read-only, and structurally so
|
|
12
|
-
|
|
13
|
-
The server issues **HTTP GETs and nothing else**. There is no code path that
|
|
14
|
-
writes, and every tool is annotated `readOnlyHint: true`, so a client can run
|
|
15
|
-
them without asking you to confirm each one.
|
|
16
|
-
|
|
17
|
-
It also cannot reach past your own organization: the key is scoped to one
|
|
18
|
-
organization, and scoping, plan checks and secret exclusion are all enforced
|
|
19
|
-
by the API, not by this client.
|
|
20
|
-
|
|
21
|
-
## What an agent can see
|
|
22
|
-
|
|
23
|
-
Worth knowing before you hand a key to an assistant. Through this server an
|
|
24
|
-
agent can read:
|
|
25
|
-
|
|
26
|
-
- your **projects** and their names;
|
|
27
|
-
- your **runs**, CI and local — commit SHA and message, branch, author, pull
|
|
28
|
-
request, job key, a link to the CI run, timing, pass/fail counts;
|
|
29
|
-
- **per-test results** for a run, including which shard ran what;
|
|
30
|
-
- for a failing attempt: the **error message and stack**, the **code snippet**
|
|
31
|
-
— **verbatim source code from your test file** — and the **full-context
|
|
32
|
-
prompt** (the text behind Copy the full context in Squally), which carries the same snippet plus an **ARIA snapshot of your
|
|
33
|
-
application at the moment of failure** (whatever was on screen, truncated to
|
|
34
|
-
3000 characters);
|
|
35
|
-
- **per-test metrics**: how often each test passed on the first try, passed
|
|
36
|
-
only after a retry, or failed, and how much time retries and failures cost;
|
|
37
|
-
- **error signatures**: failures grouped across runs and branches.
|
|
38
|
-
|
|
39
|
-
It cannot see screenshots, videos or traces — artifacts are deliberately out of
|
|
40
|
-
scope — and it cannot see any key, secret or webhook URL.
|
|
41
|
-
|
|
42
|
-
The key reaches **every project in the organization**. Create one per machine so
|
|
43
|
-
a single revocation does not lock out everything, and revoke it under
|
|
44
|
-
Organization › Read keys when the machine is retired.
|
|
45
|
-
|
|
46
|
-
## Setup
|
|
47
|
-
|
|
48
|
-
You need an **organization read key** (`sqly_ro_…`): Squally → Organization ›
|
|
49
|
-
Read keys → *Create read key*. It is shown once. It is not the same thing as the
|
|
50
|
-
project ingest key your Playwright reporter uses — this server refuses that one
|
|
51
|
-
by name.
|
|
52
|
-
|
|
53
|
-
Requires **Node 22 or newer**. Read keys are available from the **Standard**
|
|
54
|
-
plan.
|
|
55
|
-
|
|
56
|
-
### Claude Code
|
|
57
|
-
|
|
58
|
-
```bash
|
|
59
|
-
claude mcp add squally --scope user \
|
|
60
|
-
--env SQUALLY_API_KEY=sqly_ro_your_key_here \
|
|
61
|
-
-- npx -y squally-mcp
|
|
62
|
-
```
|
|
63
|
-
|
|
64
|
-
### Codex
|
|
65
|
-
|
|
66
|
-
`~/.codex/config.toml`:
|
|
67
|
-
|
|
68
|
-
```toml
|
|
69
|
-
[mcp_servers.squally]
|
|
70
|
-
command = "npx"
|
|
71
|
-
args = ["-y", "squally-mcp"]
|
|
72
|
-
env = { SQUALLY_API_KEY = "sqly_ro_your_key_here" }
|
|
73
|
-
```
|
|
74
|
-
|
|
75
|
-
### Cursor
|
|
76
|
-
|
|
77
|
-
`~/.cursor/mcp.json` (or `.cursor/mcp.json` in a project):
|
|
78
|
-
|
|
79
|
-
```json
|
|
80
|
-
{
|
|
81
|
-
"mcpServers": {
|
|
82
|
-
"squally": {
|
|
83
|
-
"command": "npx",
|
|
84
|
-
"args": ["-y", "squally-mcp"],
|
|
85
|
-
"env": { "SQUALLY_API_KEY": "sqly_ro_your_key_here" }
|
|
86
|
-
}
|
|
87
|
-
}
|
|
88
|
-
}
|
|
89
|
-
```
|
|
90
|
-
|
|
91
|
-
### Claude Desktop
|
|
92
|
-
|
|
93
|
-
`claude_desktop_config.json` — macOS:
|
|
94
|
-
`~/Library/Application Support/Claude/`, Windows: `%APPDATA%\Claude\` — then
|
|
95
|
-
restart the app:
|
|
96
|
-
|
|
97
|
-
```json
|
|
98
|
-
{
|
|
99
|
-
"mcpServers": {
|
|
100
|
-
"squally": {
|
|
101
|
-
"command": "npx",
|
|
102
|
-
"args": ["-y", "squally-mcp"],
|
|
103
|
-
"env": { "SQUALLY_API_KEY": "sqly_ro_your_key_here" }
|
|
104
|
-
}
|
|
105
|
-
}
|
|
106
|
-
}
|
|
107
|
-
```
|
|
108
|
-
|
|
109
|
-
## Environment
|
|
110
|
-
|
|
111
|
-
| Variable | Required | Meaning |
|
|
112
|
-
|---|---|---|
|
|
113
|
-
| `SQUALLY_API_KEY` | yes | Organization read key, `sqly_ro_…`. Missing or an ingest key by mistake, and the server exits with one line saying so. |
|
|
114
|
-
| `SQUALLY_API_URL` | no | Base URL, default `https://app.squally.dev`. A trailing `/` or `/api/v1` is tolerated. Useful against a local Squally: `http://localhost:3000`. |
|
|
115
|
-
| `SQUALLY_USE_SYSTEM_CA` | no | `0` keeps Node's bundled certificates only. |
|
|
116
|
-
|
|
117
|
-
squally-mcp trusts your operating system's certificate store, like your browser does; set SQUALLY_USE_SYSTEM_CA=0 to use Node's bundled certificates only.
|
|
118
|
-
|
|
119
|
-
## Troubleshooting
|
|
120
|
-
|
|
121
|
-
### Claude Desktop
|
|
122
|
-
|
|
123
|
-
- **The server does not start in time.** The first start through `npx`
|
|
124
|
-
downloads the package, which can take longer than Claude Desktop's 60-second
|
|
125
|
-
start-up limit. Install it once beforehand with `npm install -g squally-mcp`;
|
|
126
|
-
`npx -y squally-mcp` then runs that copy without a download. Update it with
|
|
127
|
-
the same command.
|
|
128
|
-
- **Where the log is.** The Microsoft Store build of Claude Desktop writes it to
|
|
129
|
-
`%LOCALAPPDATA%\Claude\logs\mcp-server-squally.log`. Its first lines say which
|
|
130
|
-
Node runs the server and whether it trusts the system's certificate store;
|
|
131
|
-
a request that never arrives is logged there with its cause.
|
|
132
|
-
- **Edit the config only while the app is closed.** Quit Claude Desktop
|
|
133
|
-
completely, including from the system tray, before changing
|
|
134
|
-
`claude_desktop_config.json`.
|
|
135
|
-
|
|
136
|
-
## The tools
|
|
137
|
-
|
|
138
|
-
| Tool | What it answers | Cost |
|
|
139
|
-
|---|---|---|
|
|
140
|
-
| `squally-list-projects` | Which projects exist. **Start here** — every other tool needs a `projectId` from it. | cheap |
|
|
141
|
-
| `squally-find-run` | Which runs happened — latest, or by branch, commit SHA, status or source. Counters only. | cheap |
|
|
142
|
-
| `squally-get-run` | One run with its per-test rows, across all shards. Which test is red. | cheap |
|
|
143
|
-
| `squally-debug-failure` | Every attempt of one test in one run: error, stack, full-context prompt. | cheap |
|
|
144
|
-
| `squally-get-test-metrics` | One test's stability, flaky rate and failure rate over a period. | cheap — one test |
|
|
145
|
-
| `squally-list-tests` | Every test with a run in the period, ranked by stability, flaky rate or failures. | cheap — two queries, paged |
|
|
146
|
-
| `squally-list-errors` | Error signatures in a period: what keeps failing, grouped, and how its runs ended. | cheap |
|
|
147
|
-
|
|
148
|
-
For a single test use `squally-get-test-metrics`, not `squally-list-tests` —
|
|
149
|
-
the tool descriptions say so, and the server repeats it in its instructions.
|
|
150
|
-
|
|
151
|
-
### Runs
|
|
152
|
-
|
|
153
|
-
`squally-find-run` lists every run, CI and local alike; `source` (`all` by
|
|
154
|
-
default, `ci` or `local`) narrows it, and each run says whether it was
|
|
155
|
-
`local`. A run's `status` is the dashboard's word: `passed`, `failed`,
|
|
156
|
-
`cancelled`, `incomplete` (a shard timed out, was stopped by the monthly
|
|
157
|
-
result limit or never reported) or `running` - and `?status=` filters by
|
|
158
|
-
any of them. A cancelled run is `cancelled`; `cancellation` says how, when
|
|
159
|
-
and by whom. A sharded run is one row, its counters over all its shards with
|
|
160
|
-
every test once, and the tests a cancel cut off count nowhere.
|
|
161
|
-
|
|
162
|
-
Each run also says whether it is `settled` - final and counted; an
|
|
163
|
-
`incomplete` run that is not settled still waits for a shard within the run
|
|
164
|
-
timeout - and, from squally-reporter 0.11.0, which job of its CI run it is
|
|
165
|
-
(`jobKey`, from `SQUALLY_CI_JOB_KEY`; each job is its own run), which
|
|
166
|
-
`attempt`, which run it re-ran (`rerunOf`) and where the CI run is
|
|
167
|
-
(`ciRunUrl`; for GitHub Actions also for older reporters when the project
|
|
168
|
-
has one linked repository).
|
|
169
|
-
|
|
170
|
-
### What the numbers mean
|
|
171
|
-
|
|
172
|
-
The two test tools count **finished runs, CI and local,** in the period (7, 14, 30 or 90
|
|
173
|
-
days; 30 by default), on every branch unless you name one. A run finishes when
|
|
174
|
-
all its shards have - or, once the run timeout has passed, with the results its
|
|
175
|
-
shards reported, like a cancelled run; a shard stopped by the monthly result
|
|
176
|
-
limit counts like one that timed out. A sharded run counts once, placed in the period
|
|
177
|
-
by its first shard's start; a run that was a mass failure as a whole is left
|
|
178
|
-
out. `source` is `all` (the default since read API 1.0.0-beta.18; `ci`
|
|
179
|
-
before), `ci` or `local` - send `source=ci` for CI runs only - and the
|
|
180
|
-
response's `population` names which runs it counted. Per test:
|
|
181
|
-
|
|
182
|
-
- **stability** — runs that passed on the first try ÷ runs;
|
|
183
|
-
- **flakyRate** — runs that passed only after a retry ÷ runs;
|
|
184
|
-
- **failureRate** — failed runs ÷ runs.
|
|
185
|
-
|
|
186
|
-
Each is a fraction from 0 to 1, and `null` when the test has no run in the
|
|
187
|
-
period. **There is no verdict**: nothing calls a test flaky, broken or
|
|
188
|
-
healthy. The tools return the numbers; the agent — or you — judges them.
|
|
189
|
-
|
|
190
|
-
### Example: `squally-list-tests`
|
|
191
|
-
|
|
192
|
-
```json
|
|
193
|
-
{ "projectId": "3f7c2a91-5b8e-4d0a-9c61-2e4b7f0d8a15", "perPage": 3 }
|
|
194
|
-
```
|
|
195
|
-
|
|
196
|
-
The three least stable tests of the last 30 days (abridged):
|
|
197
|
-
|
|
198
|
-
```json
|
|
199
|
-
{
|
|
200
|
-
"population": "ci_groups_finished_not_mass_failure",
|
|
201
|
-
"days": 30,
|
|
202
|
-
"sort": "stability",
|
|
203
|
-
"total": 409,
|
|
204
|
-
"items": [
|
|
205
|
-
{
|
|
206
|
-
"testName": "chromium > checkout.spec.ts > Checkout > applies a coupon code",
|
|
207
|
-
"filePath": "tests/checkout.spec.ts",
|
|
208
|
-
"runs": 8,
|
|
209
|
-
"stability": 0.5,
|
|
210
|
-
"flakyRate": 0.5,
|
|
211
|
-
"failureRate": 0,
|
|
212
|
-
"timeLostMs": 7361,
|
|
213
|
-
"topBranch": { "branch": "main", "kind": "retries", "count": 3 }
|
|
214
|
-
}
|
|
215
|
-
]
|
|
216
|
-
}
|
|
217
|
-
```
|
|
218
|
-
|
|
219
|
-
### Example: `squally-get-test-metrics`
|
|
220
|
-
|
|
221
|
-
```json
|
|
222
|
-
{
|
|
223
|
-
"projectId": "3f7c2a91-5b8e-4d0a-9c61-2e4b7f0d8a15",
|
|
224
|
-
"testName": "chromium > checkout.spec.ts > Checkout > applies a coupon code"
|
|
225
|
-
}
|
|
226
|
-
```
|
|
227
|
-
|
|
228
|
-
The same numbers as that test's row, for one test (abridged):
|
|
229
|
-
|
|
230
|
-
```json
|
|
231
|
-
{
|
|
232
|
-
"days": 30,
|
|
233
|
-
"branch": null,
|
|
234
|
-
"runs": 8,
|
|
235
|
-
"stableRuns": 4,
|
|
236
|
-
"flakyRuns": 4,
|
|
237
|
-
"failedRuns": 0,
|
|
238
|
-
"stability": 0.5,
|
|
239
|
-
"flakyRate": 0.5,
|
|
240
|
-
"failureRate": 0,
|
|
241
|
-
"branches": ["feature/saved-carts", "main"]
|
|
242
|
-
}
|
|
243
|
-
```
|
|
244
|
-
|
|
245
|
-
Pass the name exactly as Squally shows it. When it exists in several spec
|
|
246
|
-
files, the tool answers with the `filePath` values to retry with; a test with
|
|
247
|
-
results but no counted run in the period answers `runs: 0` and `null`
|
|
248
|
-
rates — an answer, not an error.
|
|
249
|
-
|
|
250
|
-
## Development
|
|
251
|
-
|
|
252
|
-
```bash
|
|
253
|
-
npm install
|
|
254
|
-
npm run build # tsc, then copy the vendored OpenAPI document into dist/
|
|
255
|
-
npm test # builds, then runs the suite
|
|
256
|
-
npm run vendor:openapi # re-fetch the API document; review the diff, then re-pin
|
|
257
|
-
```
|
|
258
|
-
|
|
259
|
-
The tool schemas are **derived from Squally's published OpenAPI document**,
|
|
260
|
-
vendored at `src/openapi/v1.json` — an input schema is the operation's
|
|
261
|
-
parameters, an output schema is its 200 response. Nothing is transcribed by
|
|
262
|
-
hand, so the two cannot drift. `test/drift.test.js` compares the vendored copy
|
|
263
|
-
against the live document and fails when an operation or parameter has moved;
|
|
264
|
-
it skips loudly when offline, and while the live API is still an older version
|
|
265
|
-
than the vendored one — a release that has not been deployed yet.
|
|
266
|
-
|
|
267
|
-
## Releasing
|
|
268
|
-
|
|
269
|
-
Publishing is **staged**: CI uploads the tarball, a maintainer approves it, and
|
|
270
|
-
only then is the version installable. `npm stage publish` never asks for 2FA,
|
|
271
|
-
which is what makes it usable from a workflow; the 2FA prompt moves to the
|
|
272
|
-
approval step, so a compromised workflow can stage a version but cannot put one
|
|
273
|
-
in front of users. Direct `npm publish` is not permitted for this package's
|
|
274
|
-
trusted publisher, so this is the only route.
|
|
275
|
-
|
|
276
|
-
> **0.1.0 is the exception.** It was published by hand with `npm publish`
|
|
277
|
-
> before the trusted publisher existed, so it carries no provenance
|
|
278
|
-
> attestation. Every version from 0.1.1 on goes through the steps below.
|
|
279
|
-
|
|
280
|
-
1. **Bump and tag.** The lockfile belongs to the release commit, so let `npm
|
|
281
|
-
version` write both and commit them together:
|
|
282
|
-
|
|
283
|
-
```bash
|
|
284
|
-
npm install # only if package.json changed by hand
|
|
285
|
-
npm version patch # or minor / major - writes package.json + lockfile, makes the tag
|
|
286
|
-
git push --follow-tags
|
|
287
|
-
```
|
|
288
|
-
|
|
289
|
-
2. **The tag starts the workflow.** `.github/workflows/publish.yml` runs on
|
|
290
|
-
`v*`: it tests, checks that the tag and `package.json` agree, and runs
|
|
291
|
-
|
|
292
|
-
```bash
|
|
293
|
-
npm stage publish --provenance --access public
|
|
294
|
-
```
|
|
295
|
-
|
|
296
|
-
through npm's trusted publishing (OIDC) — no npm token in GitHub secrets.
|
|
297
|
-
`--provenance` attaches a signed statement linking the tarball to that
|
|
298
|
-
workflow run and commit.
|
|
299
|
-
|
|
300
|
-
3. **Approve the staged version.** On
|
|
301
|
-
[npmjs.com/package/squally-mcp](https://www.npmjs.com/package/squally-mcp) →
|
|
302
|
-
**Staged Packages** → **Approve**, which asks for 2FA. Or from a terminal:
|
|
303
|
-
|
|
304
|
-
```bash
|
|
305
|
-
npm stage list squally-mcp
|
|
306
|
-
npm stage view <stage-id> # what is in the tarball
|
|
307
|
-
npm stage approve <stage-id> # also asks for 2FA
|
|
308
|
-
npm stage reject <stage-id> # if something is wrong
|
|
309
|
-
```
|
|
310
|
-
|
|
311
|
-
Until this step the version exists in the registry but installs nothing.
|
|
312
|
-
|
|
313
|
-
**Requirements for step 3 on your machine:** npm **11.15.0 or later** and Node
|
|
314
|
-
**22.14.0 or higher** ([npm docs](https://docs.npmjs.com/staged-publishing/)).
|
|
315
|
-
Node 22 still bundles npm 10.9.9, so check `npm --version` rather than assuming
|
|
316
|
-
your Node version brought a new enough npm. The workflow uses Node 24 for the
|
|
317
|
-
same reason and fails with that requirement if a runner ever ships an older npm.
|
|
318
|
-
|
|
319
|
-
## License
|
|
320
|
-
|
|
321
|
-
MIT
|
|
1
|
+
# squally-mcp
|
|
2
|
+
|
|
3
|
+
A local [MCP](https://modelcontextprotocol.io) server that lets a coding agent
|
|
4
|
+
read your [Squally](https://app.squally.dev) data: which runs happened, how
|
|
5
|
+
stable each test is, and why a particular test failed.
|
|
6
|
+
|
|
7
|
+
It runs on your machine over stdio and talks to Squally's read API over HTTPS.
|
|
8
|
+
Point Claude Code, Codex, Cursor or Claude Desktop at it and ask "why is
|
|
9
|
+
`checkout.spec.ts` failing on main?".
|
|
10
|
+
|
|
11
|
+
## Read-only, and structurally so
|
|
12
|
+
|
|
13
|
+
The server issues **HTTP GETs and nothing else**. There is no code path that
|
|
14
|
+
writes, and every tool is annotated `readOnlyHint: true`, so a client can run
|
|
15
|
+
them without asking you to confirm each one.
|
|
16
|
+
|
|
17
|
+
It also cannot reach past your own organization: the key is scoped to one
|
|
18
|
+
organization, and scoping, plan checks and secret exclusion are all enforced
|
|
19
|
+
by the API, not by this client.
|
|
20
|
+
|
|
21
|
+
## What an agent can see
|
|
22
|
+
|
|
23
|
+
Worth knowing before you hand a key to an assistant. Through this server an
|
|
24
|
+
agent can read:
|
|
25
|
+
|
|
26
|
+
- your **projects** and their names;
|
|
27
|
+
- your **runs**, CI and local — commit SHA and message, branch, author, pull
|
|
28
|
+
request, job key, a link to the CI run, timing, pass/fail counts;
|
|
29
|
+
- **per-test results** for a run, including which shard ran what;
|
|
30
|
+
- for a failing attempt: the **error message and stack**, the **code snippet**
|
|
31
|
+
— **verbatim source code from your test file** — and the **full-context
|
|
32
|
+
prompt** (the text behind Copy the full context in Squally), which carries the same snippet plus an **ARIA snapshot of your
|
|
33
|
+
application at the moment of failure** (whatever was on screen, truncated to
|
|
34
|
+
3000 characters);
|
|
35
|
+
- **per-test metrics**: how often each test passed on the first try, passed
|
|
36
|
+
only after a retry, or failed, and how much time retries and failures cost;
|
|
37
|
+
- **error signatures**: failures grouped across runs and branches.
|
|
38
|
+
|
|
39
|
+
It cannot see screenshots, videos or traces — artifacts are deliberately out of
|
|
40
|
+
scope — and it cannot see any key, secret or webhook URL.
|
|
41
|
+
|
|
42
|
+
The key reaches **every project in the organization**. Create one per machine so
|
|
43
|
+
a single revocation does not lock out everything, and revoke it under
|
|
44
|
+
Organization › Read keys when the machine is retired.
|
|
45
|
+
|
|
46
|
+
## Setup
|
|
47
|
+
|
|
48
|
+
You need an **organization read key** (`sqly_ro_…`): Squally → Organization ›
|
|
49
|
+
Read keys → *Create read key*. It is shown once. It is not the same thing as the
|
|
50
|
+
project ingest key your Playwright reporter uses — this server refuses that one
|
|
51
|
+
by name.
|
|
52
|
+
|
|
53
|
+
Requires **Node 22 or newer**. Read keys are available from the **Standard**
|
|
54
|
+
plan.
|
|
55
|
+
|
|
56
|
+
### Claude Code
|
|
57
|
+
|
|
58
|
+
```bash
|
|
59
|
+
claude mcp add squally --scope user \
|
|
60
|
+
--env SQUALLY_API_KEY=sqly_ro_your_key_here \
|
|
61
|
+
-- npx -y squally-mcp
|
|
62
|
+
```
|
|
63
|
+
|
|
64
|
+
### Codex
|
|
65
|
+
|
|
66
|
+
`~/.codex/config.toml`:
|
|
67
|
+
|
|
68
|
+
```toml
|
|
69
|
+
[mcp_servers.squally]
|
|
70
|
+
command = "npx"
|
|
71
|
+
args = ["-y", "squally-mcp"]
|
|
72
|
+
env = { SQUALLY_API_KEY = "sqly_ro_your_key_here" }
|
|
73
|
+
```
|
|
74
|
+
|
|
75
|
+
### Cursor
|
|
76
|
+
|
|
77
|
+
`~/.cursor/mcp.json` (or `.cursor/mcp.json` in a project):
|
|
78
|
+
|
|
79
|
+
```json
|
|
80
|
+
{
|
|
81
|
+
"mcpServers": {
|
|
82
|
+
"squally": {
|
|
83
|
+
"command": "npx",
|
|
84
|
+
"args": ["-y", "squally-mcp"],
|
|
85
|
+
"env": { "SQUALLY_API_KEY": "sqly_ro_your_key_here" }
|
|
86
|
+
}
|
|
87
|
+
}
|
|
88
|
+
}
|
|
89
|
+
```
|
|
90
|
+
|
|
91
|
+
### Claude Desktop
|
|
92
|
+
|
|
93
|
+
`claude_desktop_config.json` — macOS:
|
|
94
|
+
`~/Library/Application Support/Claude/`, Windows: `%APPDATA%\Claude\` — then
|
|
95
|
+
restart the app:
|
|
96
|
+
|
|
97
|
+
```json
|
|
98
|
+
{
|
|
99
|
+
"mcpServers": {
|
|
100
|
+
"squally": {
|
|
101
|
+
"command": "npx",
|
|
102
|
+
"args": ["-y", "squally-mcp"],
|
|
103
|
+
"env": { "SQUALLY_API_KEY": "sqly_ro_your_key_here" }
|
|
104
|
+
}
|
|
105
|
+
}
|
|
106
|
+
}
|
|
107
|
+
```
|
|
108
|
+
|
|
109
|
+
## Environment
|
|
110
|
+
|
|
111
|
+
| Variable | Required | Meaning |
|
|
112
|
+
|---|---|---|
|
|
113
|
+
| `SQUALLY_API_KEY` | yes | Organization read key, `sqly_ro_…`. Missing or an ingest key by mistake, and the server exits with one line saying so. |
|
|
114
|
+
| `SQUALLY_API_URL` | no | Base URL, default `https://app.squally.dev`. A trailing `/` or `/api/v1` is tolerated. Useful against a local Squally: `http://localhost:3000`. |
|
|
115
|
+
| `SQUALLY_USE_SYSTEM_CA` | no | `0` keeps Node's bundled certificates only. |
|
|
116
|
+
|
|
117
|
+
squally-mcp trusts your operating system's certificate store, like your browser does; set SQUALLY_USE_SYSTEM_CA=0 to use Node's bundled certificates only.
|
|
118
|
+
|
|
119
|
+
## Troubleshooting
|
|
120
|
+
|
|
121
|
+
### Claude Desktop
|
|
122
|
+
|
|
123
|
+
- **The server does not start in time.** The first start through `npx`
|
|
124
|
+
downloads the package, which can take longer than Claude Desktop's 60-second
|
|
125
|
+
start-up limit. Install it once beforehand with `npm install -g squally-mcp`;
|
|
126
|
+
`npx -y squally-mcp` then runs that copy without a download. Update it with
|
|
127
|
+
the same command.
|
|
128
|
+
- **Where the log is.** The Microsoft Store build of Claude Desktop writes it to
|
|
129
|
+
`%LOCALAPPDATA%\Claude\logs\mcp-server-squally.log`. Its first lines say which
|
|
130
|
+
Node runs the server and whether it trusts the system's certificate store;
|
|
131
|
+
a request that never arrives is logged there with its cause.
|
|
132
|
+
- **Edit the config only while the app is closed.** Quit Claude Desktop
|
|
133
|
+
completely, including from the system tray, before changing
|
|
134
|
+
`claude_desktop_config.json`.
|
|
135
|
+
|
|
136
|
+
## The tools
|
|
137
|
+
|
|
138
|
+
| Tool | What it answers | Cost |
|
|
139
|
+
|---|---|---|
|
|
140
|
+
| `squally-list-projects` | Which projects exist. **Start here** — every other tool needs a `projectId` from it. | cheap |
|
|
141
|
+
| `squally-find-run` | Which runs happened — latest, or by branch, commit SHA, status or source. Counters only. | cheap |
|
|
142
|
+
| `squally-get-run` | One run with its per-test rows, across all shards. Which test is red. | cheap |
|
|
143
|
+
| `squally-debug-failure` | Every attempt of one test in one run: error, stack, full-context prompt. | cheap |
|
|
144
|
+
| `squally-get-test-metrics` | One test's stability, flaky rate and failure rate over a period. | cheap — one test |
|
|
145
|
+
| `squally-list-tests` | Every test with a run in the period, ranked by stability, flaky rate or failures. | cheap — two queries, paged |
|
|
146
|
+
| `squally-list-errors` | Error signatures in a period: what keeps failing, grouped, and how its runs ended. | cheap |
|
|
147
|
+
|
|
148
|
+
For a single test use `squally-get-test-metrics`, not `squally-list-tests` —
|
|
149
|
+
the tool descriptions say so, and the server repeats it in its instructions.
|
|
150
|
+
|
|
151
|
+
### Runs
|
|
152
|
+
|
|
153
|
+
`squally-find-run` lists every run, CI and local alike; `source` (`all` by
|
|
154
|
+
default, `ci` or `local`) narrows it, and each run says whether it was
|
|
155
|
+
`local`. A run's `status` is the dashboard's word: `passed`, `failed`,
|
|
156
|
+
`cancelled`, `incomplete` (a shard timed out, was stopped by the monthly
|
|
157
|
+
result limit or never reported) or `running` - and `?status=` filters by
|
|
158
|
+
any of them. A cancelled run is `cancelled`; `cancellation` says how, when
|
|
159
|
+
and by whom. A sharded run is one row, its counters over all its shards with
|
|
160
|
+
every test once, and the tests a cancel cut off count nowhere.
|
|
161
|
+
|
|
162
|
+
Each run also says whether it is `settled` - final and counted; an
|
|
163
|
+
`incomplete` run that is not settled still waits for a shard within the run
|
|
164
|
+
timeout - and, from squally-reporter 0.11.0, which job of its CI run it is
|
|
165
|
+
(`jobKey`, from `SQUALLY_CI_JOB_KEY`; each job is its own run), which
|
|
166
|
+
`attempt`, which run it re-ran (`rerunOf`) and where the CI run is
|
|
167
|
+
(`ciRunUrl`; for GitHub Actions also for older reporters when the project
|
|
168
|
+
has one linked repository).
|
|
169
|
+
|
|
170
|
+
### What the numbers mean
|
|
171
|
+
|
|
172
|
+
The two test tools count **finished runs, CI and local,** in the period (7, 14, 30 or 90
|
|
173
|
+
days; 30 by default), on every branch unless you name one. A run finishes when
|
|
174
|
+
all its shards have - or, once the run timeout has passed, with the results its
|
|
175
|
+
shards reported, like a cancelled run; a shard stopped by the monthly result
|
|
176
|
+
limit counts like one that timed out. A sharded run counts once, placed in the period
|
|
177
|
+
by its first shard's start; a run that was a mass failure as a whole is left
|
|
178
|
+
out. `source` is `all` (the default since read API 1.0.0-beta.18; `ci`
|
|
179
|
+
before), `ci` or `local` - send `source=ci` for CI runs only - and the
|
|
180
|
+
response's `population` names which runs it counted. Per test:
|
|
181
|
+
|
|
182
|
+
- **stability** — runs that passed on the first try ÷ runs;
|
|
183
|
+
- **flakyRate** — runs that passed only after a retry ÷ runs;
|
|
184
|
+
- **failureRate** — failed runs ÷ runs.
|
|
185
|
+
|
|
186
|
+
Each is a fraction from 0 to 1, and `null` when the test has no run in the
|
|
187
|
+
period. **There is no verdict**: nothing calls a test flaky, broken or
|
|
188
|
+
healthy. The tools return the numbers; the agent — or you — judges them.
|
|
189
|
+
|
|
190
|
+
### Example: `squally-list-tests`
|
|
191
|
+
|
|
192
|
+
```json
|
|
193
|
+
{ "projectId": "3f7c2a91-5b8e-4d0a-9c61-2e4b7f0d8a15", "perPage": 3 }
|
|
194
|
+
```
|
|
195
|
+
|
|
196
|
+
The three least stable tests of the last 30 days (abridged):
|
|
197
|
+
|
|
198
|
+
```json
|
|
199
|
+
{
|
|
200
|
+
"population": "ci_groups_finished_not_mass_failure",
|
|
201
|
+
"days": 30,
|
|
202
|
+
"sort": "stability",
|
|
203
|
+
"total": 409,
|
|
204
|
+
"items": [
|
|
205
|
+
{
|
|
206
|
+
"testName": "chromium > checkout.spec.ts > Checkout > applies a coupon code",
|
|
207
|
+
"filePath": "tests/checkout.spec.ts",
|
|
208
|
+
"runs": 8,
|
|
209
|
+
"stability": 0.5,
|
|
210
|
+
"flakyRate": 0.5,
|
|
211
|
+
"failureRate": 0,
|
|
212
|
+
"timeLostMs": 7361,
|
|
213
|
+
"topBranch": { "branch": "main", "kind": "retries", "count": 3 }
|
|
214
|
+
}
|
|
215
|
+
]
|
|
216
|
+
}
|
|
217
|
+
```
|
|
218
|
+
|
|
219
|
+
### Example: `squally-get-test-metrics`
|
|
220
|
+
|
|
221
|
+
```json
|
|
222
|
+
{
|
|
223
|
+
"projectId": "3f7c2a91-5b8e-4d0a-9c61-2e4b7f0d8a15",
|
|
224
|
+
"testName": "chromium > checkout.spec.ts > Checkout > applies a coupon code"
|
|
225
|
+
}
|
|
226
|
+
```
|
|
227
|
+
|
|
228
|
+
The same numbers as that test's row, for one test (abridged):
|
|
229
|
+
|
|
230
|
+
```json
|
|
231
|
+
{
|
|
232
|
+
"days": 30,
|
|
233
|
+
"branch": null,
|
|
234
|
+
"runs": 8,
|
|
235
|
+
"stableRuns": 4,
|
|
236
|
+
"flakyRuns": 4,
|
|
237
|
+
"failedRuns": 0,
|
|
238
|
+
"stability": 0.5,
|
|
239
|
+
"flakyRate": 0.5,
|
|
240
|
+
"failureRate": 0,
|
|
241
|
+
"branches": ["feature/saved-carts", "main"]
|
|
242
|
+
}
|
|
243
|
+
```
|
|
244
|
+
|
|
245
|
+
Pass the name exactly as Squally shows it. When it exists in several spec
|
|
246
|
+
files, the tool answers with the `filePath` values to retry with; a test with
|
|
247
|
+
results but no counted run in the period answers `runs: 0` and `null`
|
|
248
|
+
rates — an answer, not an error.
|
|
249
|
+
|
|
250
|
+
## Development
|
|
251
|
+
|
|
252
|
+
```bash
|
|
253
|
+
npm install
|
|
254
|
+
npm run build # tsc, then copy the vendored OpenAPI document into dist/
|
|
255
|
+
npm test # builds, then runs the suite
|
|
256
|
+
npm run vendor:openapi # re-fetch the API document; review the diff, then re-pin
|
|
257
|
+
```
|
|
258
|
+
|
|
259
|
+
The tool schemas are **derived from Squally's published OpenAPI document**,
|
|
260
|
+
vendored at `src/openapi/v1.json` — an input schema is the operation's
|
|
261
|
+
parameters, an output schema is its 200 response. Nothing is transcribed by
|
|
262
|
+
hand, so the two cannot drift. `test/drift.test.js` compares the vendored copy
|
|
263
|
+
against the live document and fails when an operation or parameter has moved;
|
|
264
|
+
it skips loudly when offline, and while the live API is still an older version
|
|
265
|
+
than the vendored one — a release that has not been deployed yet.
|
|
266
|
+
|
|
267
|
+
## Releasing
|
|
268
|
+
|
|
269
|
+
Publishing is **staged**: CI uploads the tarball, a maintainer approves it, and
|
|
270
|
+
only then is the version installable. `npm stage publish` never asks for 2FA,
|
|
271
|
+
which is what makes it usable from a workflow; the 2FA prompt moves to the
|
|
272
|
+
approval step, so a compromised workflow can stage a version but cannot put one
|
|
273
|
+
in front of users. Direct `npm publish` is not permitted for this package's
|
|
274
|
+
trusted publisher, so this is the only route.
|
|
275
|
+
|
|
276
|
+
> **0.1.0 is the exception.** It was published by hand with `npm publish`
|
|
277
|
+
> before the trusted publisher existed, so it carries no provenance
|
|
278
|
+
> attestation. Every version from 0.1.1 on goes through the steps below.
|
|
279
|
+
|
|
280
|
+
1. **Bump and tag.** The lockfile belongs to the release commit, so let `npm
|
|
281
|
+
version` write both and commit them together:
|
|
282
|
+
|
|
283
|
+
```bash
|
|
284
|
+
npm install # only if package.json changed by hand
|
|
285
|
+
npm version patch # or minor / major - writes package.json + lockfile, makes the tag
|
|
286
|
+
git push --follow-tags
|
|
287
|
+
```
|
|
288
|
+
|
|
289
|
+
2. **The tag starts the workflow.** `.github/workflows/publish.yml` runs on
|
|
290
|
+
`v*`: it tests, checks that the tag and `package.json` agree, and runs
|
|
291
|
+
|
|
292
|
+
```bash
|
|
293
|
+
npm stage publish --provenance --access public
|
|
294
|
+
```
|
|
295
|
+
|
|
296
|
+
through npm's trusted publishing (OIDC) — no npm token in GitHub secrets.
|
|
297
|
+
`--provenance` attaches a signed statement linking the tarball to that
|
|
298
|
+
workflow run and commit.
|
|
299
|
+
|
|
300
|
+
3. **Approve the staged version.** On
|
|
301
|
+
[npmjs.com/package/squally-mcp](https://www.npmjs.com/package/squally-mcp) →
|
|
302
|
+
**Staged Packages** → **Approve**, which asks for 2FA. Or from a terminal:
|
|
303
|
+
|
|
304
|
+
```bash
|
|
305
|
+
npm stage list squally-mcp
|
|
306
|
+
npm stage view <stage-id> # what is in the tarball
|
|
307
|
+
npm stage approve <stage-id> # also asks for 2FA
|
|
308
|
+
npm stage reject <stage-id> # if something is wrong
|
|
309
|
+
```
|
|
310
|
+
|
|
311
|
+
Until this step the version exists in the registry but installs nothing.
|
|
312
|
+
|
|
313
|
+
**Requirements for step 3 on your machine:** npm **11.15.0 or later** and Node
|
|
314
|
+
**22.14.0 or higher** ([npm docs](https://docs.npmjs.com/staged-publishing/)).
|
|
315
|
+
Node 22 still bundles npm 10.9.9, so check `npm --version` rather than assuming
|
|
316
|
+
your Node version brought a new enough npm. The workflow uses Node 24 for the
|
|
317
|
+
same reason and fails with that requirement if a runner ever ships an older npm.
|
|
318
|
+
|
|
319
|
+
## License
|
|
320
|
+
|
|
321
|
+
MIT
|