- boot-smoke: shutdown verdict now requires SIGTERM actually delivered and a
clean exit (code 0 or SIGTERM); EADDRINUSE port race retries with a fresh
port (3 attempts)
- test-changed: rename shouldForceFullSuite -> isSharedInfraChange (it routes
to gate mode, not full); run the changed-mode gate under the isolation guard
- workflows: least-privilege permissions (contents: read) on pr-checks and
full-suite
- decideExecutionPlan: implicit wide-blast reasons (missing base, diff failed,
no changes, shared infra, unmapped package) route to new gate mode instead of full
- CI===true force-full branch removed (CI no longer calls test-changed.mjs)
- changed mode runs pnpm test:gate before the affected set
- full suite reachable only via explicit --full / FUSION_TEST_FULL=1
- characterization tests updated + new only-explicit-full invariant test
- pr-checks.yml: gate job = boot smoke + pnpm test:gate; shards/slow/inventory removed
- full-suite.yml (new): 4-way shards, engine slow tier, inventory guard on push to main
- ci.yml deleted (trigger-disabled since FN-1541; was dead config)
- ci-workflow.test.ts rewritten to pin the new gate shape; release/signing blocks preserved
- docs/contributing.md: pnpm test:gate is the merge gate; verify:workspace repositioned as deep opt-in
BRANCH-PROTECTION CUTOVER: set required checks to exactly [Lint, Typecheck, Build, Gate] after merging; open PRs must rebase.
U5: action-gating contexts, the heartbeat deferral gate, a two-pass
resumeTaskForAgent, and the reverse-direction agent.taskId guards (via a
new isAgentEffectivelyExecuting callback wired at the in-process runtime)
all consult the column-effective agent; the restart watcher re-resolves
column bindings per tick for bound graph sessions, hot-swapping on
agent-changed and falling back without restart on agent-deleted.
- Add heal block to createPluginRouter's enable handler so it matches
routes.ts (directory-path registrations re-pointed at entry files)
- Test the 400 "no loadable entry file" install branch
- Add a real-fs drift-guard test asserting the CLI and @fusion/core
copies of resolvePluginEntryPath resolve identically
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Plugins installed from Settings → Built-in Plugins registered the
manifest directory as the plugin path, but since FN-4128 the loader
requires a loadable entry FILE (Node ESM cannot import directories),
so enabling failed with "Plugin entry must be a file, got directory".
Only the CLI startup path had been migrated to entry-file resolution,
which is why CLI-auto-installed plugins worked and Settings installs
never did.
- Add resolvePluginEntryPath (bundled.js → dist/index.js →
src/index.ts) to @fusion/core; the CLI keeps its local copy (its
test fs mocks don't reach externalized core) with sync comments
both ways.
- Register the resolved entry file in both dashboard install routes;
400 with a clear message when a package has no loadable entry.
- Heal legacy directory-path registrations on enable, mirroring the
CLI's startup heal, so existing broken rows recover from the UI
without a restart.
- Route tests: assert installs register entry files, cover the
enable-route heal, and update existing install tests to the
entry-file contract.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- Update test assertions for bare model IDs
- Add deduplication guard for models with same bare ID
- Add validation for empty model IDs after prefix stripping
- Add test for API key forwarded as env var to spawn
- Add test for deduplication and empty model ID guard
- Extract shared handleOpencodeGoApiKeySaved helper
- Add changeset for the published package
normalizeOpencodeGoModel was prefixing model IDs with 'opencode-go/'
(e.g. 'opencode-go/deepseek-v4-flash'), but the Pi SDK sends the model
id field as the model name in API requests. The OpenCode API expects
bare model names (e.g. 'deepseek-v4-flash'), not prefixed ones.
Now strips the 'opencode/' or 'opencode-go/' prefix entirely so the
registered model ID matches what the API expects.
discoverOpencodeGoModels() spawns 'opencode models opencode --refresh' but
never passed the saved API key as OPENCODE_API_KEY. The opencode CLI's
internal OpencodePlugin checks this env var to decide whether to show
paid models; without it, only free (cost.input === 0) models appear.
Now threads the apiKey from auth storage through to the spawned process
environment, so the CLI sees the user's Go subscription and returns the
full model catalog including paid models like Claude, GPT-5.x, Gemini, etc.
Callers in serve.ts, daemon.ts, and dashboard.ts all updated to read
the key from dashboardAuthStorage and pass it to refreshOpencodeGoModels.
syncStartupModels reads from authStorage in StartupSyncOptions.
Installing the Compound Engineering plugin from Settings → Built-in
Plugins failed with "Plugin manifest not found": the dashboard's
bundled-plugin fallback set (BUNDLED_PLUGIN_IDS in routes.ts) was
missing fusion-plugin-compound-engineering, so when the relative
./plugins/... path missed the server cwd, the bundled lookup was never
attempted. fusion-plugin-cli-printing-press had the same gap.
- Add both ids to the dashboard's BUNDLED_PLUGIN_IDS fallback set.
- Stage fusion-plugin-compound-engineering into dist/plugins via
bundlePluginEntry in the CLI tsup config so packaged installs can
resolve the bundled copy (every other bundled plugin already had a
staging block).
- Add route tests that actually exercise the bundled fallback (the
existing ones let cwd resolution succeed, so the fallback path was
untested).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
vi.waitFor polls, so the bound adds zero time to green runs; ink frame
scheduling has flaked past 3s under shard CPU contention while passing
instantly in isolation.
The curated-lane contract still holds — default lane never runs broad
dashboard-app/dashboard-api — but test:quality:api now chains
curated + backfill sub-lanes, so the --project arg lives one level down.
Contract strengthened to also pin the backfill completeness net.
Migration (multi-agent sweep over 216 files, 60 batches):
- Every user-visible dashboard + TUI string moved to t() with the exact
English inline default (en rendering byte-identical)
- Catalogs merged from per-batch fragments: en/zh-CN/zh-TW/fr/es now
carry ~5,930 keys each across common/app/errors/cli namespaces;
CLI bundles regenerated (6 locales incl. ko)
Integration fixes:
- 18 type errors: reserved {{count}} interpolations renamed, malformed
plural call, hand-rolled t-param types replaced with TFunction<"app">
- 23 lint errors: superseded label constants/helpers removed
- ExecutorStatusBar hook-order violation (keyboard-open early return
moved below hooks)
- TUI tests wrapped in I18nextProvider (uninitialized fallback renders
literal {{placeholders}}); dashboard vitest.setup boots a minimal en
i18next instance for the same reason
Known WIP (next commits): ~457 residual strings across 50 batches,
Korean drafts for swept keys, and a dashboard test-suite pass that is
still being stabilized (~283 failures under investigation — fake-timer
waitFor interaction, likely stale node_modules vs merged lockfile).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
skill-sync.test.ts ("engine-tools.md documents all engine session-scoped
tools") scans the engine sources for `name: "fn_*"` and requires each to appear
in packages/cli/skill/fusion/references/engine-tools.md. The two new workflow
tools were added to agent-tools.ts/executor without a doc entry, so the sync
test failed (masked earlier by the dashboard build break). Add both to the
shared agent-tools table.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- resolve execFile lazily via namespace import in coordinator/merger/
task-lifecycle so the repo's exec-only child_process test mocks load again
(10+ engine suites failed at import); dashboard.test.ts mock gains execFile
so the argv-based git probes hit the mock instead of spawning real git
- TaskCard: capture optional branchContext.groupId into a const (narrowing
doesn't survive into the onClick closure; app tsconfig caught it in CI)
- planning e2e: bounded poll past the 2.5s listTasks startup memo that served
a pre-landing snapshot on fast CI runs
- ko added to SUPPORTED_LOCALES and every enumeration site (config,
settings enum, help text, tests); Korean catalogs authored for all
current keys; CLI bundles regenerated for 6 locales; 한국어 endonym
- README translated into zh-CN, zh-TW, fr, es, ko; every README carries
a language-switcher line and the localized ones note that English is
canonical
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- abandon (route + CLI) preserves prState 'none' for groups that never had a
PR instead of falsely persisting 'closed'; regression tests both sides
- stale-snapshot write guard extracted to syncGroupPrOnLanding and covered by
a fast in-memory unit test (FN-5048); the slow real-git duplicate removed
Both sides bootstrapped CONCEPTS.md — kept main's broader structure (Missions,
Merge lifecycle, CE sessions) and folded the branch-group vocabulary into its
Merge-lifecycle cluster (enriched Shared branch group + Self-healing sweep;
added Branch assignment mode, Landed, Group promotion). Union-merged core
index/store imports; took main's AGENTS.md discoverability wording. Also adds
execFile to the two child_process test mocks that only stubbed exec (the
coordinator's argv-based git calls now require it).
- syncGroupPrCallback forwards owner/repo to updatePr (multi-project daemons
could 404 or edit an unrelated same-numbered PR via process-cwd fallback)
- merger background reconcile re-reads the group before persisting and skips
the write when the PR snapshot changed (stale-write race vs newer open PR)
- branchContext.groupId trimmed on metadata emit/parse round-trip
- triageSlice non-shared invariant assertions (no groupId, no group row)
