@antelopejs/tooling-configs 0.3.7 → 0.4.0
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 +1 -1
- package/dist/bin/sync.mjs +6 -7
- package/dist/bin/verify.mjs +424 -565
- package/dist/{map-COx26Bu6.mjs → map-B23_WMGE.mjs} +52 -51
- package/package.json +1 -1
package/README.md
CHANGED
|
@@ -66,7 +66,7 @@ Both profiles of `antelopejs-sync` keep the shared CI caller on the job id `ci`,
|
|
|
66
66
|
|
|
67
67
|
`antelopejs-sync` also installs the `technical-writing` and `unslop` skills in every full-profile repository and in a template source. Each is `.agents/skills/<name>/SKILL.md` with a `.claude/skills/<name>` link. The text is vendored unchanged from Upd4ting/agent-config `claude/skills/<name>/SKILL.md` at commit 4e0a848, and `--check` reports an edited file as drift. Write documentation with them: `technical-writing` picks the page's mode and sets the sentence rules, and `unslop` lists the patterns to cut.
|
|
68
68
|
|
|
69
|
-
Every deterministic step of a verification is an `antelopejs-verify` command: `map`, `plan`, `new`, `doctor`, `journey`, `formal`, `doors`, `clean
|
|
69
|
+
Every deterministic step of a verification is an `antelopejs-verify` command: `map`, `plan`, `new`, `doctor`, `journey`, `formal`, `doors`, `clean` and `verdict`. The synced `SKILL.md` documents each one for the agent that verifies.
|
|
70
70
|
|
|
71
71
|
The verify skills drive a browser with `playwright-core`. [Browser driver for the verify skills](./docs/browser-driver.md) records why, with the measured speed, CPU and memory.
|
|
72
72
|
|
package/dist/bin/sync.mjs
CHANGED
|
@@ -8,7 +8,7 @@ import { a as gitFiles, c as repositoryRoot } from "../files-DJ0TeaTU.mjs";
|
|
|
8
8
|
import { i as recordedRelease, t as measureBaseline } from "../ratchet-Q-JXuZIX.mjs";
|
|
9
9
|
import { a as ciWorkflow, c as ciNodeCallers, l as ciRenovateSyncCallers, o as placeholderTestScript, p as renovateArtifacts, r as CI_PATH, s as MISSING_CHECKS_WORKFLOW, u as hasChecksWorkflow } from "../ci-workflow-37qJa9LX.mjs";
|
|
10
10
|
import { n as SUPPRESSIONS_FILE, r as adoptedBudget } from "../suppression-budget-C0SkPO5R.mjs";
|
|
11
|
-
import { t as DEFAULT_SHARED } from "../map-
|
|
11
|
+
import { t as DEFAULT_SHARED } from "../map-B23_WMGE.mjs";
|
|
12
12
|
import { chmodSync, existsSync, lstatSync, mkdirSync, readFileSync, readdirSync, readlinkSync, rmdirSync, statSync, symlinkSync, unlinkSync, writeFileSync } from "node:fs";
|
|
13
13
|
import { dirname, join, posix, resolve } from "node:path";
|
|
14
14
|
import { execFileSync } from "node:child_process";
|
|
@@ -1208,7 +1208,7 @@ const KIND_TEXT = {
|
|
|
1208
1208
|
- \`saas: { env, sandbox }\`: a SaaS module runs against the provider's sandbox with keys from these environment variables. Without every key, each journey is NOT VERIFIED and so is the verdict, never PASS or PASS+NOTES.
|
|
1209
1209
|
- \`journeys\`: the module's own journeys. Each writes through the module with \`context.call(...)\` and reads the result back with \`context.native(...)\` (\`redis-cli\`, \`mongosh\`, \`psql\`, the Mailpit API) inside the service container.`,
|
|
1210
1210
|
smoke: "`module-dev` and one write read back through the native client",
|
|
1211
|
-
gotchas: ["Every container is the run's own. Never point a journey at a shared instance such as a developer's local database.", "A sandbox key missing from the environment is a NOT VERIFIED verdict
|
|
1211
|
+
gotchas: ["Every container is the run's own. Never point a journey at a shared instance such as a developer's local database.", "A sandbox key missing from the environment is a NOT VERIFIED verdict."]
|
|
1212
1212
|
},
|
|
1213
1213
|
template: {
|
|
1214
1214
|
description: "Drive this AntelopeJS template the way a new user does: scaffold a project with `ajs project init` from an offline interfaces manifest, install it, run its own `pnpm run checks`, run `ajs project dev --watch` and hit its welcome page, endpoint or ready line. Use to prove a change to the template works before merge, to verify a PR, or to run the release canary.",
|
|
@@ -1266,12 +1266,12 @@ Doctor checks that HEAD has not moved since \`new\`, \`ajs\` comes from the repo
|
|
|
1266
1266
|
|
|
1267
1267
|
\`\`\`bash
|
|
1268
1268
|
${BIN} plan --base origin/main
|
|
1269
|
-
${BIN} journey <run> --plan
|
|
1269
|
+
${BIN} journey <run> --plan
|
|
1270
1270
|
\`\`\`
|
|
1271
1271
|
|
|
1272
1272
|
\`plan\` maps the diff to features and lists unmapped surfaces. \`journey <run> --plan\` runs every feature \`plan\` lists as \`run\` for the run's base, \`smoke\` first, in one call: it prints each feature's lines in turn, keeps going after a feature fails, and exits 1 when any journey did not pass. Run it once instead of one call per feature.
|
|
1273
1273
|
|
|
1274
|
-
When verify.config.ts declares no \`services\`,
|
|
1274
|
+
Features run one after another by default. When verify.config.ts declares no \`services\`, use \`--jobs 4\` when the machine is otherwise idle (load below the core count); otherwise run sequentially. \`--jobs\` runs up to that many features at once, each in a process of its own, capped at half the machine's cores; it finishes sooner but costs about twice the CPU time. The output, the evidence and the exit code are those of the sequential call, and a feature that lists a journey an earlier feature lists waits for it, so it still reuses that PASS. A feature whose process dies or leaves an uncaught error is named on stderr, and the others still run. On an interrupt, a feature's process that has not stopped 5 s later, or at a second interrupt, is killed with the apps it started; the call prints every feature that ended and names the rest in \`not run: <features>\`. \`--jobs\` above 1 is refused when verify.config.ts declares \`services\`, because each journey resets them as it starts, which would wipe the data of a journey running beside it. Without \`--jobs\`, a journey that leaves an uncaught error behind ends the call and the line \`not run: <features>\` names what to rerun.
|
|
1275
1275
|
|
|
1276
1276
|
To rerun after a fix, run one feature, or only the journeys you name, or several features by name, with \`--jobs\` too:
|
|
1277
1277
|
|
|
@@ -1345,14 +1345,13 @@ Write your judgement to \`<run>/evidence/judgement.json\`:
|
|
|
1345
1345
|
}
|
|
1346
1346
|
\`\`\`
|
|
1347
1347
|
|
|
1348
|
-
A door row is \`{ "kind", "location", "irreversible", "containment" }\`, where \`containment\` says what limits the damage. Then render
|
|
1348
|
+
A door row is \`{ "kind", "location", "irreversible", "containment" }\`, where \`containment\` says what limits the damage. Then render the verdict:
|
|
1349
1349
|
|
|
1350
1350
|
\`\`\`bash
|
|
1351
1351
|
${BIN} verdict <run> --pr <n>
|
|
1352
|
-
${BIN} publish <run> --pr <n>
|
|
1353
1352
|
\`\`\`
|
|
1354
1353
|
|
|
1355
|
-
\`verdict\` writes \`evidence/verdict.md\` and exits 1 unless the result is PASS or PASS+NOTES.
|
|
1354
|
+
\`verdict\` writes \`evidence/verdict.md\` and exits 1 unless the result is PASS or PASS+NOTES. Report the verdict and what you saw in your own reply, from a thread that did not write the change.
|
|
1356
1355
|
|
|
1357
1356
|
## Trust
|
|
1358
1357
|
|