Closes the U2/U5 skill-discovery carry-forward so the plugin's interactive ce-*
sessions actually load the stage's bundled skill in a live agent (not just in
scripted-fake tests).
Root cause: createFnAgent built its DefaultResourceLoader without forwarding any
skill-discovery path, and the interactive seam options couldn't carry one. The
loader's skillsOverride only *filters* skills already discovered from cwd's
standard roots, so the plugin-local .fusion-ce-skills/<id>/SKILL.md was never
discoverable.
Fix (end-to-end):
- AgentOptions.additionalSkillPaths forwarded into DefaultResourceLoader
- CreateInteractiveAiSessionOptions gains requestedSkillNames + additionalSkillPaths
- the interactive engine adapter forwards them to createFnAgent (skills +
additionalSkillPaths)
- the orchestrator runs the session with cwd on the real project root and hands
it [stage.skillId] + the install root
Proven: a real DefaultResourceLoader with additionalSkillPaths discovers ce-plan
and filters out ce-work; the orchestrator passes the right id/path/cwd. Plugin 96,
engine 136, core 99 tests green.
- reconciler: a deleted current-stage board task no longer wedges the pipeline
in 'running' forever (terminality computed over existing tasks only; all-deleted
is a no-op, not a wedge)
- session-store: a human-slow awaiting_input session is no longer misclassified
stale (interval rubric applies only to in-flight active/launching turns)
- stage-registry: pipeline progression uses an explicit order ordinal instead of
registry insertion order, so out-of-order registration can't corrupt advancement
- orchestrator.answer(): validate questionId before mutating state, so a stale id
can't destroy the persisted currentQuestion recovery anchor
- orchestrator.resume(): rehydrate a live interactive session by replaying
persisted history (side effects suppressed) so a resumed session is actually
answerable instead of dead-ending; honest interrupted+error fallback when no
factory is available
95 tests (6 new regression tests, each confirmed failing pre-fix).
Quality cleanup across the 9-unit build (behavior-preserving, 89 tests green):
- extract createCeTaskWithLink so the work bridge and reconciler share one
provenance+link contract (prevents drift)
- discovery list scan probes readability via accessSync instead of reading and
discarding full file bytes
- makeError helper replaces ~7 duplicated CeArtifactError literals
- shared asString route helper; drop dead pipelineIds set; collapse a double
pipeline-state write and a redundant link re-query in advance
- resolveStageSkillCwd no longer takes params it ignores
Add settingsSchema (default session provider/model, enabled stages, sync
reconcile-on-hooks toggle, reconcile cadence hint) aligned across manifest.json
and the runtime manifest, with typed getters. Wire the consumed getters:
orchestrator passes defaultProvider/defaultModelId into sessions and rejects
disabled stages; hooks gate their reconcile drain on reconcileOnHooks. Add the
plugin README and docs/plugins/compound-engineering.md documenting the hub,
interactive sessions, work bridge, and the sync ownership model.
reconcileIntervalMinutes is exposed as an operator cadence hint but not yet
consumed (no host scheduler; reconcile is on-demand by design).
Add a ce_pipeline_state machine (currentStage/status) kept distinct from
board-task ownership (task column) — separate tables, no shared column, per
FN-5719. onTaskMoved/onTaskCompleted hooks do only an indexed lookup + enqueue
and return well under the 5s budget; advancement happens in an on-demand
reconciler sweep that re-derives correct pipeline state from board truth, so a
dropped hook event still converges (no tight poll loop). Outbound CE-flow
changes create the next-stage board task. Conflict policy: board authoritative
for task state, CE flow authoritative for artifact/pipeline content (inbound
read and outbound write target different rows, so they cannot contend).
Host-scheduler note: no host timer wired; sweeps run on hook-drain and on the
dashboard/route refresh surface (tracked for the host event-publish follow-up).
When the work stage completes with a derived task list, create Fusion tasks via
ctx.taskStore.createTask tagged CE-originated (sourceType workflow_step +
sourceMetadata marker) and record an authoritative ce_pipeline_links row per
task (back-reference lives in the link table, not task-row JSON, per FN-5719).
Tasks then run the normal lifecycle untouched. Zero derived tasks is a clean
no-op. The link store is intentionally minimal for U8 to extend with the
bidirectional pipeline-state machine.
Extend the stage registry with presentation metadata (icon/label/glob) so adding
a stage stays data-only. Add CeFlow renderer for text/single_select/multi_select/
confirm questions over the U5 polling session routes, with a visibly-marked chat
fallback (AE1) that still completes the stage. Wire the artifact-hub launcher to
list and start registered stages. Add the skill-interaction audit test producing
a measured rich-vs-chat coverage ratio (declared classification: 8/8=100% across
brainstorm/ideate/plan), failing on unclassified interactions.
Add allowlist-based artifact discovery over the conventional CE locations
(STRATEGY.md, docs/ideation, docs/brainstorms, docs/plans, docs/solutions,
CONCEPTS.md), grouped by stage, with error entries for unreadable artifacts and
no reads outside the allowlist. Add list/read/preview routes (self-contained
sandboxed HTML, data-section markers) and the primary CompoundEngineeringView
with explicit empty/partial/error states, viewport-gated fetch, and a short-TTL
cache. Bind the view via registerBundledPluginViews + dashboard workspace dep
(the real binding; manifest componentPath is cosmetic).
Add ce_sessions schema (onSchemaInit), CeSessionStore (via
ctx.taskStore.getDatabase()), and CeOrchestrator driving a stage's skill on the
U4 interactive seam: streams thinking/text, persists questions as
awaiting_input, writes the stage artifact on complete, and auto-saves +
emits an observable event on interrupt/error (never silent loss). Lifecycle:
launching/active/awaiting_input/completed/error/interrupted with resume and an
interval-relative staleness rubric. Start/answer/resume/get-state routes.
Streaming transport is client polling of the get-session-state route for v1:
plugin routes have no native SSE and the loader emitEvent is a logging stub, so
true server push needs a host publish-to-/api/events seam (tracked follow-up).
Skill discovery is cwd+prompt-based; forwarding install paths through the seam
is a tracked follow-up.
Bundle pinned copies of 7 CE pipeline-stage skills (strategy, ideate,
brainstorm, plan, work, code-review, compound) under src/skills/ and declare
them via PluginSkillContribution. Empirical finding: the skills contribution
alone does not make a SKILL.md resolvable in a session -- the engine ingests it
as a name only. So onLoad runs an idempotent, isolation-guarded physical install
into a plugin-local .fusion-ce-skills/ dir (never a global ~/.claude/skills),
which the engine skill-resolver can then discover. Proven against the real
loadSkills + resolveSessionSkills pipeline.
The Binary Release workflow stopped producing any GitHub Release assets
because every release had at least one failing build leg, and the
github-release job (needs: all four builds, no if:) was skipped whenever
any leg failed — suppressing even successfully-built platforms.
Root causes fixed:
- github-release: add `if: !cancelled()` + zero-artifact guard so a single
failing leg yields a partial release instead of none.
- setup-node-pnpm cache key: add runner.arch. runner.os is only
Linux/macOS/Windows, so arm64 runners restored x64 node_modules missing
native deps (@rollup/rollup-linux-arm64-gnu), crashing `pnpm build`.
- macOS CLI sign step: guard on APPLE_CERTIFICATE_BASE64 so unsigned
binaries still publish when certs are absent; add timeout-minutes: 30 to
build-binaries to avoid 24h runner hangs.
- dependency-graph plugin: replace unix cp/mkdir -p (failed on Windows
cmd.exe) with a cross-platform node copy script.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Introduces a Goals REST API (`GET/POST/PUT /api/goals` and `GET/PUT /api/goals/:id`) backed by a new `@fusion/core` goal store and typed goal types, including comprehensive route and store test coverage. Documentation on architecture and storage is updated to reflect the new domain, and a changeset
Fusion-Task-Id: FN-5622
Co-authored-by: Fusion (runfusion.ai) <noreply@runfusion.ai>
Fusion-Task-Id: FN-5622
The api stub from 43be32bbd silenced the fetch-rejection path but the suite
still failed on CI: the actual unhandled error is React's scheduler firing
deferred work via setImmediate after jsdom is torn down — its internal
render then dereferences `window` and throws ReferenceError. @testing-library
only auto-registers cleanup() when vitest `globals: true` is set, and this
package doesn't enable globals, so the React tree from each render() stays
mounted across teardown. Register cleanup manually in test-setup.ts so every
dashboard test unmounts its tree before the environment tears down.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
ReportDetailPanel transitively calls useReportPreview, which fires fetch()
from api.ts. jsdom has no fetch, the promise rejects, and the catch handler's
setError triggers a React update after the test environment is torn down —
React then accesses window and the suite fails with ReferenceError. The
engine failures previously masked this by failing the shard before the
teardown race could surface. Mocking the api module makes the preview
resolve synchronously and keeps the suite clean.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Remove duplicate coreSetup from reports-dashboard project setupFiles.
With extends: true, coreSetup is already inherited from the root config.
Only dashboardSetup needs to be explicitly added at the project level.
Resolves CI failures where Node.js 24 caused ReferenceError: window is not defined
in non-dashboard tests that shared setupFiles with dashboard tests.
Follows the same pattern already used by fusion-plugin-roadmap: split into two
vitest projects with proper environment isolation:
- reports-dashboard: jsdom, includes src/dashboard/** tests + dashboard test-setup.ts
- reports-node: node, includes all other tests, excludes dashboard tests
No test changes — same 104 tests, same pass/fail state.
Merges the Layer 2.5 scope-auto-widen feature (FN-5226) into the merger: a new evaluator module that automatically widens a task's declared file scope based on git attribution prior to the existing scope partition gate, wired into `merger.ts` with full audit taxonomy, persisted task metadata, and re
Fusion-Task-Id: FN-5226
- Add completion handoff limbo recovery state to task storage and audit events for bounded self-healing retries
- Requeue eligible in-review tasks stuck after "Task marked done by agent" with grace-period, active-task, and merge-blocker safeguards
- Isolate executor run mutation context per task to prevent cross-task runId attribution leaks in scope-leak and completion logs
- Add reliability and executor regression tests covering limbo recovery, run-context isolation, and updated persistence behavior
Fusion-Task-Id: FN-4999
The roadmap-dashboard Vitest sub-project overrode setupFiles with only
its jsdom helper. Vitest replaces (does not merge) setupFiles in child
projects, so the central isolation guard at
packages/core/src/__test-utils__/vitest-setup.ts was silently bypassed
for the 94 roadmap-dashboard tests, leaving them able to touch the real
project .fusion / $HOME / child processes without the leak-detection
and AI-CLI block.
Prepend the central guard so it loads before the dashboard jsdom setup.
All 94 roadmap-dashboard tests still pass.
Adds a title-id drift detection and reconciliation system to prevent duplicate tasks when task titles change after initial creation. Core adds the `task-title-id-drift.ts` module with `markTitleIdStable` and `resolveTitleIdDrift` logic, wired through `db.ts`, `store.ts`, and `archive-db.ts`, with co
Fusion-Task-Id: FN-4898
