Commit Graph

101 Commits

Author SHA1 Message Date
xingyue bbb79f821e feat(init): generate bunfig.toml with Gitea scoped registry + fix versions
- init workspace now generates bunfig.toml pointing @uncaged scope to Gitea
- Fix template versions: "*" → "^0.3.1" (packages are now published)
2026-05-12 11:31:41 +08:00
xingyue 0fbbf37548 chore(cli): add bun run link scripts + fix init template versions
- Add link/link:consume/link:unlink scripts to root package.json
- Document cross-repo bun link workflow in CLAUDE.md
- Fix hardcoded @uncaged/workflow-runtime "^0.1.0" → "*" in
  init workspace and init template scaffolds (actual version is 0.3.x)
2026-05-12 11:03:12 +08:00
xiaomo 1c68ce6217 Merge pull request 'feat: Phase 3 — agent observability for Merkle call stack' (#203) from feat/197-agent-observability into main 2026-05-12 02:34:47 +00:00
xingyue 9cb7d68abe feat: Dashboard workflow graph visualization with React Flow (#198)
Phase 1: API + static graph rendering

Backend:
- GET /workflows/:name now returns descriptor (with graph) from bundle YAML
- Graceful fallback to null if YAML missing/invalid

Frontend:
- New workflow-graph/ component module (7 files)
- React Flow + dagre auto-layout (TB direction)
- Custom nodes: RoleNode (rounded rect) + TerminalNode (circle for START/END)
- Custom edges: dashed for FALLBACK, solid with label for conditions
- Self-loop edges supported (e.g. coder → coder)
- Node states: default/completed/active with color-coded borders
- Active node pulse animation
- Collapsible graph panel (300px) above thread records
- Dark theme using existing CSS variables

Integration:
- ThreadDetail extracts workflow name → fetches descriptor → computes node states → renders graph
- Node states derived from ThreadRecord[] (completed/active/default)
2026-05-12 10:27:07 +08:00
xiaoju 98122b446d feat: Phase 3 — agent observability for Merkle call stack
- StartStep gains parentState: string | null (from StartNodePayload)
- buildAgentPrompt injects Parent Context section when parentState is set
- CLI thread show outputs parentState (top-level) and childThread (per step)
- 2 new prompt tests + thread show assertion updates

Refs #197, #194

小橘 🍊(NEKO Team)
2026-05-12 02:23:15 +00:00
xiaomo aebff8b906 Merge pull request 'refactor: replace Moderator function with ModeratorTable in WorkflowDefinition' (#201) from refactor/200-moderator-table into main 2026-05-12 02:06:03 +00:00
xingyue db45089922 refactor: replace Moderator function with ModeratorTable in WorkflowDefinition (#200)
- WorkflowDefinition.moderator → WorkflowDefinition.table (ModeratorTable)
- Moderator type + tableToModerator no longer exported from protocol/runtime
- tableToModerator internalized in workflow-execute engine layer
- WorkflowDescriptor gains graph: WorkflowGraph (auto-extracted from table)
- buildDescriptor extracts serializable graph edges from ModeratorTable
- validateWorkflowDescriptor validates graph structure
- All templates (develop, solve-issue) export table directly
- CLI init scaffold updated to use ModeratorTable
- 99 tests pass, 0 failures
2026-05-12 10:01:30 +08:00
xiaoju a98431a12a feat(#194): Phase 1 — parentState / childThread in CAS nodes
- Protocol: StartNodePayload.parentState, StateNodePayload.childThread
- CAS: putStartNode refs include parentState, collectRefs includes childThread
- Parsing: legacy nodes without new fields default to null
- Engine + fork: all callers pass parentState: null / childThread: null
- Tests: 8 new cases for refs, parsing, collect-refs (+208 lines)

Phase 1 of #194 (Merkle Call Stack). Closes #195.

小橘 <xiaoju@shazhou.work>
2026-05-12 01:42:10 +00:00
xiaoju cf0540d7fa fix: stabilize kill-thread test by polling instead of fixed delay
Replace 50ms setTimeout with waitUntilPredicate polling for first role
completion before issuing kill. Same pattern used by pause/resume test.

小橘 <xiaoju@shazhou.work>
2026-05-11 13:56:14 +00:00
xiaoju 376dd87b6b chore: bump to v0.3.1, fix workspace:* in published packages
v0.3.0 was published with workspace:* deps (npm doesn't resolve them).
Re-published as v0.3.1 with concrete version refs.
Local monorepo deps restored to workspace:*.

小橘 <xiaoju@shazhou.work>
2026-05-11 11:00:49 +00:00
xiaoju 4d8469a649 chore: bump all packages to v0.3.0
Published to Gitea npm registry (git.shazhou.work).

小橘 <xiaoju@shazhou.work>
2026-05-11 10:52:35 +00:00
xiaoju ec13c19505 Merge pull request 'refactor: replace maxRounds with supervisor check interval' (#186) from refactor/185-remove-max-rounds into main 2026-05-11 09:01:24 +00:00
xiaoju 2b587612d5 refactor: replace maxRounds with supervisor check interval
Removes maxRounds as a hard stop limit from the entire stack. The supervisor
(already configured via workflow.yaml supervisorInterval) is now the sole
termination authority.

Changes across 27 files in 11 packages:
- workflow-protocol: StartStep.meta is now empty, StartNodePayload drops maxRounds
- workflow-cas: isStartPayload no longer checks maxRounds
- workflow-execute: engine, worker, fork-thread all stripped of maxRounds plumbing
- cli-workflow: --max-rounds flag removed from CLI and HTTP API
- workflow-runtime: build-context and create-workflow no longer reference maxRounds
- workflow-dashboard: UI no longer sends maxRounds
- workflow-template-develop/solve-issue: moderator no longer checks rounds remaining
- All tests updated

Fixes #185
2026-05-11 08:51:35 +00:00
xingyue 2342a6e3bd fix: login.tsx use new gateway endpoint path 2026-05-11 16:46:38 +08:00
xingyue 0021596ff0 Merge pull request 'refactor: simplify ExtractFn to (schema, contentHash)' (#184) from refactor/180-simplify-extract-fn into main 2026-05-11 08:03:48 +00:00
xiaoju 1742ced6df refactor: simplify ExtractFn to (schema, contentHash)
- Remove extractPrompt from RoleDefinition
- Remove ExtractContext type
- ExtractFn now takes (schema, contentHash) instead of (schema, prompt, ExtractContext)
- createExtract reads CAS content by hash, keeps ReAct loop with cas_get
- Coder schema uses .describe() for phase hash hint
- All role definitions, CLI templates, and skill output updated

Refs #180, closes #174, closes #181
2026-05-11 07:54:09 +00:00
xingyue 93145cf08c refactor: reorganize gateway routes under /api/ prefix (closes #178, closes #179)
- Gateway management: /api/gateway/register, /api/gateway/endpoints
- Agent proxy: /api/agents/:agent/*
- /healthz stays at root (CF/k8s convention)
- Skip dashboard auth for gateway register routes
- Update CLI serve tunnel registration paths
- Update dashboard API client paths

Ref: #177
2026-05-11 15:48:13 +08:00
xiaoju da6bcb10d6 feat(workflow): add declarative ModeratorTable type and migrate templates
Migrate workflow-template-develop and workflow-template-solve-issue
moderators to use the declarative ModeratorTable/tableToModerator
pattern. Update workflow-runtime re-exports and workflow-execute
engine to use renamed types.

Fixes #172
2026-05-11 06:25:39 +00:00
xiaoju 93d9821f64 docs: update CLI skill with serve command, thread status, defaults, env vars
小橘 <xiaoju@shazhou.work>
2026-05-10 01:57:42 +00:00
xiaoju 2e9d939f8e fix: thread detail API returns correct status instead of source
小橘 <xiaoju@shazhou.work>
2026-05-10 01:39:09 +00:00
xiaoju 064a24f093 fix: no-ctl threads should be failed, not active
小橘 <xiaoju@shazhou.work>
2026-05-10 01:36:14 +00:00
xiaoju 2a52b930b9 chore: raise default maxRounds from 5 to 10 (CLI, matches API default)
小橘 <xiaoju@shazhou.work>
2026-05-09 13:17:57 +00:00
xiaoju bf2f790e6e fix: detect crashed threads even when .running marker is already gone
Check worker PID liveness as final fallback — if worker is dead
and thread has no __end__ node, it crashed.

小橘 <xiaoju@shazhou.work>
2026-05-09 12:52:39 +00:00
xiaoju 08a79b77db fix: SSE sends 'done' event for non-running threads, frontend stops reconnecting
- routes-live: emit 'done' event before closing SSE for non-running threads
- use-sse: handle 'done' event — set completed, disconnect, stop reconnect
- Prevents 'Live' badge flash on failed/completed threads

小橘 <xiaoju@shazhou.work>
2026-05-09 12:49:20 +00:00
xiaoju 22a6200b69 fix: close SSE stream for non-running threads, fix Live badge
- routes-live: check .running marker before keeping SSE open;
  if thread is not running, emit existing records and close
- thread-detail: only show Live badge when connected AND not completed

小橘 <xiaoju@shazhou.work>
2026-05-09 12:45:58 +00:00
xiaoju 7e7f6aa6d6 fix: detect crashed threads by checking worker PID liveness
When .running marker exists but no __end__ in CAS chain,
check if the worker process is actually alive. Dead PID
means the worker crashed without cleanup → status 'failed'.

Fixes #170

小橘 <xiaoju@shazhou.work>
2026-05-09 12:38:18 +00:00
xiaoju d6fe3f844c fix: detect crashed threads as failed instead of stuck running
- resolveThreadListStatus() checks CAS chain for __end__ node
- Stale .running markers no longer cause false 'running' status
- Distinguish 'failed' (returnCode != 0) from 'completed'
- Worker signal handlers (SIGINT/SIGTERM) clean up .running files
- listRunningThreads filters out terminated threads with stale markers

Fixes #170

小橘 <xiaoju@shazhou.work>
2026-05-09 12:28:33 +00:00
xiaoju d0803019b5 feat: ephemeral agent token for serve ↔ gateway auth
- serve generates random UUID on startup
- registration sends agentToken to gateway, stored in KV
- gateway injects X-Agent-Token header when proxying to agent
- serve rejects /api/* requests without valid token
- healthz remains unauthenticated
- tunnel URL is now protected — direct access returns 401

小橘 <xiaoju@shazhou.work>
2026-05-09 12:05:10 +00:00
xiaoju d96e10b0fc feat(dashboard): structured record rendering with markdown support (#169)
- API returns structured fields for thread-start (workflow, prompt, status)
  and workflow-result (returnCode, content, timestamp)
- New RecordCard component renders by type:
  - StartCard: workflow name badge + prompt blockquote
  - RoleMessage: role-colored badges (preparer/agent/extractor) + markdown
  - ResultCard: success/fail status badge + summary
- Added react-markdown + shiki for markdown rendering with syntax highlighting
- Replaces generic <pre> blocks with proper structured rendering

Refs #169
小橘 <xiaoju@shazhou.work>
2026-05-09 10:41:13 +00:00
xiaoju 8e36d3e1f5 fix: use getContentMerklePayload to extract prompt text
Was showing raw YAML of the CAS node instead of the payload string.

小橘 <xiaoju@shazhou.work>
2026-05-09 10:34:43 +00:00
xiaoju bbe4fe0ed1 fix: include prompt text in thread-start record
Read prompt from StartNode refs[0] CAS blob and display it.

小橘 <xiaoju@shazhou.work>
2026-05-09 10:32:59 +00:00
xiaoju e105c5cac1 fix: show workflow name instead of bundle hash in thread-start record
小橘 <xiaoju@shazhou.work>
2026-05-09 10:31:08 +00:00
xiaoju 578776fccf fix: add standard fields to thread-start record
小橘 <xiaoju@shazhou.work>
2026-05-09 10:27:03 +00:00
xiaoju cb756a999a fix: normalize workflow-result records to match ThreadRecord shape
Both REST and SSE endpoints now return workflow-result with standard
fields (role, content, timestamp) instead of non-standard (summary).
Fixes 'Invalid Date' and empty content in dashboard.

小橘 <xiaoju@shazhou.work>
2026-05-09 10:24:48 +00:00
xiaoju e0577ceefe fix: add /api/healthz alias for gateway proxy health check
Gateway proxies /api/neko/healthz → /api/healthz on the agent,
but healthz was only on /healthz. Dashboard status bar showed
permanent Offline.

小橘 🍊(NEKO Team)
2026-05-09 10:05:46 +00:00
xiaoju fd8943f131 feat: serve auto-tunnel + gateway registration
Phase B of #164:
- serve --name <agent> starts cloudflared quick tunnel automatically
- Registers with CF Worker gateway, heartbeat every 60s
- Graceful unregister on SIGINT/SIGTERM
- --no-tunnel flag for local dev
- Default name from hostname

Ref: #164, closes #166

小橘 🍊(NEKO Team)
2026-05-09 09:53:08 +00:00
xiaoju e29d1bf345 feat: Phase 5 — CLI + Dashboard CAS adaptation, cleanup .data.jsonl
- Align REST API contracts for Dashboard (threads list, detail, SSE)
- Add content resolution from CAS in thread show + API responses
- Rename dataWatcher → threadsJsonWatcher in SSE routes
- Update docs (CLAUDE.md, architecture.md, skill.ts) to reflect CAS storage
- Zero .data.jsonl code paths in production code
- All 166 tests pass, bun run check clean

Refs #155, closes #160

小橘 <xiaoju@shazhou.work>
2026-05-09 08:16:04 +00:00
xiaoju f3aedf8d6c feat: Phase 4 — CAS-based fork + mark-and-sweep GC
- Rewrite fork to create StateNode pointing to fork point (zero duplication)
- Rewrite GC as mark-and-sweep: roots from threads.json + history, findReachableHashes via refs[]
- Remove .data.jsonl code paths
- Fix all 7 previously failing CLI tests
- New: gc-mark-sweep.test.ts verifying shared nodes survive GC
- All 166 tests pass

Refs #155, closes #159

小橘 <xiaoju@shazhou.work>
2026-05-09 08:12:49 +00:00
xiaoju 26cf51366f feat: Phase 3 — engine read path + runtime context builder
- Add buildThreadContext(headHash, cas) to workflow-runtime
- Expand extract phase to return { meta, contentPayload, refs[] }
- Add parseCasThreadNode() to workflow-cas for node type parsing
- Update createWorkflow to write ContentMerkleNode with artifact refs
- Tests: 4 pass (build-context + extract-refs)
- Biome format pass on all files

Refs #155, closes #158

小橘 <xiaoju@shazhou.work>
2026-05-09 08:00:24 +00:00
xiaoju 0f28e9b61a refactor: remove non-index re-exports
Remove re-exports from non-index files to enforce folder module discipline:
- workflow-agent-hermes: remove buildAgentPrompt re-export
- workflow-agent-cursor: remove buildAgentPrompt re-export
- cli-workflow/cli-dispatch: remove CommandEntry/CommandGroup/DispatchFn/getCommandRegistry re-exports

小橘 <xiaoju@shazhou.work>
2026-05-09 04:19:49 +00:00
xingyue 727b4bb3ed refactor(workflow): fix tsconfig references, template imports, delete old packages/workflow
- Update root tsconfig.json references: replace packages/workflow with 6 new packages
- Update cli-workflow tsconfig references to new packages
- Add tsconfig references to workflow-util, workflow-runtime, workflow-execute
- Fix workflow-agent-llm, workflow-agent-cursor, workflow-agent-hermes, workflow-util-agent
  tsconfig references (../workflow -> ../workflow-runtime)
- Remove stale @uncaged/workflow deps from agent package.json files
- Change template packages to import buildDescriptor from @uncaged/workflow-register
- Normalize package.json exports field across all new packages
- Delete old packages/workflow/ directory
2026-05-09 11:46:57 +08:00
xingyue 9bbdfc41bd feat(execute): create @uncaged/workflow-execute + CLI migration
Phase 7: Engine + extract + workflow-as-agent merged into execute package.
All CLI imports migrated from @uncaged/workflow to specific packages.
105 CLI tests pass, 0 failures.

Changes:
- New @uncaged/workflow-execute package (engine/, extract/, workflow-as-agent)
- CLI src/ and __tests__/ rewritten to import from split packages
- bundle-validator updated to allow @uncaged/workflow-cas imports
- ensure-uncaged-workflow-symlink creates symlinks for all new packages

Ref: #143, closes #150
2026-05-09 11:35:03 +08:00
xiaomo 2c0e744ebf Merge pull request 'perf(serve): SSE live pump reads incrementally' (#137) from fix/130-sse-incremental into main 2026-05-08 15:00:44 +00:00
xingyue 61fc1cfe1b perf(serve): SSE live pump reads incrementally instead of full file
Use Bun.file().slice() to read only new bytes from the last known
offset instead of re-reading the entire JSONL file on every fs watch.

- readNewBytes() helper with byte-offset tracking
- Handles file truncation (resets offset)
- Early return when no new data

Closes #130
2026-05-08 18:14:04 +08:00
xingyue 6b1e728700 fix(serve): error handling, CORS, body limit, CAS store reuse
- Global error handler (app.onError → 500 JSON)
- JSON parse validation on POST routes (400)
- CORS restricted to localhost origins
- 1MB body size limit on POST (413)
- CAS store created once per route group, not per-request
- 6 new tests covering all changes

Closes #120
2026-05-08 18:11:59 +08:00
Scott Wei cc3f2b576c refactor(workflow): decouple agent context from CAS and fix monorepo checks
Move CAS access into extract dependencies so AgentContext stays state-only, and clean up type/lint/check regressions across CLI/dashboard to keep full check green.

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-05-08 17:30:07 +08:00
Scott Wei a11cc62a81 refactor(workflow-runtime): use full ThreadContext in WorkflowFn
Redefine WorkflowFn to accept a complete ThreadContext plus WorkflowRuntime dependencies, removing ThreadInput and WorkflowFnOptions.

Move thread context construction into engine executeThread, update runtime loop/agent paths, and align templates/docs/tests with template-only definition exports.

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-05-08 17:08:01 +08:00
Scott Wei 34f5e655d1 refactor(workflow): unify extraction behind ExtractFn
Route createExtract through reactExtract with plain-JSON correction retry.

Remove WorkflowFnOptions.llmProvider, ExtractMode, RoleDefinition.extractMode, ResolveRoleMetaFn.

Runtime createWorkflow calls options.extract directly; engine passes extract only.

Update templates, CLI skill docs, and tests.

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-05-08 17:08:01 +08:00
xingyue cdcaff15ab feat(serve+dashboard): write endpoints, SSE live, run dialog
Serve API:
- POST /api/threads — run a new thread
- POST /api/threads/:id/kill — kill thread
- POST /api/threads/:id/pause — pause thread
- POST /api/threads/:id/resume — resume thread
- GET /api/threads/:id/live — SSE stream of thread records

Dashboard:
- Run Thread dialog (select workflow, enter prompt, set maxRounds)
- Thread detail controls (pause/resume/kill buttons)
- postJson API helper for write operations

262 tests pass. Refs: #118
2026-05-08 16:07:02 +08:00
xiaoju 587518ac09 refactor(workflow): cleanup engine re-exports, final verification (Phase 4)
- Remove all re-exports from @uncaged/workflow -> @uncaged/workflow-runtime
- Fix cli-workflow imports to use @uncaged/workflow-runtime for types
- Update bundle-validator to allow @uncaged/workflow-runtime imports
- Update init templates to reference @uncaged/workflow-runtime
- 378 tests passing, build + check clean

Refs #121, relates #125
2026-05-08 06:37:56 +00:00