680 Commits

Author	SHA1	Message	Date
xiaoju	adc3982a4a	fix: bootstrap agent discovery + adapter version independence (#120) ... - Step 1: detect hermes/claude before choosing adapter - Adapter versions independent from CLI — install @latest - ACP verification: hermes acp --help - Remove uwf-builtin (not ready) Refs #120	2026-06-05 22:29:35 +00:00
xiaoju	caba82fe36	fix: bootstrap PATH fix guidance — find binary location + update shell config (#118 #1)	2026-06-05 16:45:33 +00:00
xiaoju	6aee2ed5ef	fix: bootstrap docs — pnpm/npm parity, adapter order, preset table (#118) ... - Show pnpm and npm install commands side-by-side - Clarify adapter must be installed before uwf setup --agent - Add version verification steps with PATH troubleshooting - --agent takes adapter command name (uwf-hermes), not npm package - Preset providers shown as table with default base URLs - Non-preset providers must specify --base-url manually Fixes #118 (#2, #3, #4, #5)	2026-06-05 16:41:35 +00:00
xiaoju	7a788a9d90	fix: suppress ExperimentalWarning, PEP 668 guidance, setup help ... - All 5 CLI bins: shebang --disable-warning=ExperimentalWarning - Remove NODE_OPTIONS injection from thread.ts spawn (redundant now) - Bootstrap pip install: venv (recommended) / pipx / source options - setup --help mentions interactive wizard mode - Update shebang test to accept -S flag Fixes #116	2026-06-05 16:12:06 +00:00
xiaoju	fde87b6274	fix: setup UX improvements — adapter check, ENOENT, SQLite warning, VERSION, PATH docs ... - setup validates adapter binary availability, prints install command if missing - setup prints 'Config saved to <path> ✓' on success - spawn ENOENT gives actionable error with which command - SQLite ExperimentalWarning suppressed via NODE_OPTIONS - bootstrap VERSION reads cli package.json (was reading util) - bootstrap PATH guidance is shell-agnostic Fixes #114	2026-06-05 15:42:22 +00:00
xiaoju	3be92bfac2	fix: bootstrap adds Step 0 environment pre-flight check ... - Node.js, pnpm/npm, global bin PATH, hermes CLI checks with FIX instructions - Agent must pass all checks before proceeding to install - Install commands changed from npm to pnpm (with npm fallback) - hermes PATH guidance moved from Step 1 to Step 0 Fixes #112	2026-06-05 14:09:33 +00:00
xiaoju	5450bc1230	fix: workflow-authoring flat schema, bootstrap PATH guidance ... - #110.3: flat schema example uses enum: [done] instead of bare const (bare const fails validate-semantic hasStatusEnum check) - #110.4: bootstrap adds 'which hermes' PATH check and venv guidance - #110.1: already fixed in rc.1 (inline hello.yaml) - #110.2: already fixed in rc.1 (capabilities: [] present) Fixes #110	2026-06-05 11:44:20 +00:00
xiaoju	57ae6d1755	fix: preset base-url auto-fill, bootstrap ACP docs, friendlier errors ... - #106: uwf setup --provider <preset> now auto-fills --base-url - #107: bootstrap documents hermes ACP dependency (pip install hermes-agent[acp]) - #107: verify step uses inline hello.yaml instead of missing examples/eval-simple.yaml - #108: workflow name mismatch error suggests how to fix (rename file or change YAML name) Fixes #106, Fixes #107, Fixes #108	2026-06-05 11:06:35 +00:00
xiaoju	c5eb8b79d1	fix: expand bootstrap prompt with full onboarding and upgrade guide ... - Fresh install: CLI + adapter install, uwf setup, skills, e2e verify - Upgrade: update packages, regenerate skills, migrate workflows - Explicitly tells agent to ask user for provider/api-key/model - Lists all available adapters with install commands - Documents v0.2.0 $START migration Fixes #104	2026-06-05 10:35:01 +00:00
xiaoju	36a3ca6a08	chore: bump cli@0.2.0, util@0.1.2	2026-06-05 10:11:19 +00:00
xiaoju	a47871ec4e	chore: remove unused moderator-reference and yaml-reference ... These generate* functions were exported from util but never consumed by any code. Dead exports are maintenance burden. Refs #101	2026-06-05 09:44:50 +00:00
xiaoju	fbfd31a042	feat: replace $START _ status with new/resume semantics ... BREAKING: All workflow YAML files must update $START._ to $START.new + $START.resume. The resume edge prompt replaces the previously hardcoded resume message. - evaluate.ts: remove START_ROLE/START_STATUS special case, use $status like all nodes - thread.ts: resolveEvaluateArgs passes 'new', cmdThreadResume passes 'resume' - validate.ts: reject '_' everywhere (no longer valid) - validate-semantic.ts: require 'new' and 'resume' edges on $START - All workflow YAMLs and test fixtures updated Fixes #101	2026-06-05 09:30:09 +00:00
xiaoju	a536efee00	fix: simplify prompt subcommands, framework-agnostic bootstrap ... - `uwf prompt usage` now outputs only the usage skill (was three combined) - `uwf prompt bootstrap` replaces `setup` with framework-agnostic instructions - Remove `usage-reference` and `setup` subcommands - Remove `generateBootstrapReference` from util (moved to cli) Fixes #99 小橘 🍊（NEKO Team）	2026-06-05 08:52:35 +00:00
xiaoju	9260d81084	chore: version bump for --version fix ... agent-hermes@0.1.2 agent-claude-code@0.1.1 agent-builtin@0.1.1 agent-mock@0.1.1 eval@0.1.3 util@0.1.1 小橘 🍊（NEKO Team）	2026-06-05 08:12:50 +00:00
xiaoju	abeb465f46	fix: acp-client reports own package version, not util VERSION ... Address review nit from PR #97: clientInfo.version should be agent-hermes's own version for correct identification under independent versioning. 小橘 🍊（NEKO Team）	2026-06-05 07:50:03 +00:00
xiaoju	794f9db568	fix: add --version to adapter CLIs, read VERSION from package.json ... - All uwf-* adapter CLIs now support --version / -V - util VERSION constant reads from package.json at runtime - agent-hermes ACP clientInfo uses dynamic VERSION 小橘 🍊（NEKO Team）	2026-06-05 07:29:54 +00:00
xiaoju	1cf8f350d0	fix: read eval CLI version from package.json ... Fixes #95 小橘 🍊（NEKO Team）	2026-06-05 06:43:27 +00:00
xiaoju	427568a21d	chore: version bump agent-hermes@0.1.1 cli@0.1.1 eval@0.1.2 ... 小橘 🍊（NEKO Team）	2026-06-05 06:29:25 +00:00
xiaoju	8085d1d6e0	fix: read token usage from ACP response instead of DB ... Tokens (inputTokens, outputTokens) now come from ACP PromptResponse.usage which is populated synchronously from run_conversation() — no WAL race. Turns still come from DB before/after snapshot. Previously both were read from hermes state.db after ACP prompt returned, but WAL write lag caused incomplete token data (e.g. 235 vs actual 26,080). Refs #91	2026-06-05 06:08:11 +00:00
xiaoju	825f0c641a	fix: resolve --agent override via config alias before raw command ... When --agent is passed to uwf thread exec, try config.agents[alias] first (e.g. 'hermes' → config.agents.hermes = {command: 'uwf-hermes'}), then fall back to parseAgentOverride for raw command names. Also change eval CLI default --agent from 'hermes' to 'uwf-hermes' so it works without config alias lookup. Refs #91	2026-06-05 04:20:09 +00:00
xiaoju	81bbe1178f	chore: release @united-workforce/eval@0.1.1	2026-06-05 03:02:05 +00:00
xiaoju	a08775896f	fix: frontmatter judge handles parsed object output ... The extract pipeline stores step output as a JSON object in CAS, but the frontmatter judge only checked for raw markdown strings. Now accepts both formats: parsed objects check $status directly, raw strings go through YAML frontmatter extraction. Fixes eval frontmatter-compliance scoring 0 on valid outputs.	2026-06-05 02:55:58 +00:00
xiaoju	c892b9125b	chore: remove prepublishOnly guards (proman handles release)	2026-06-05 02:29:53 +00:00
xiaoju	5edb67b79d	chore: prepare 0.1.0 release ... - Remove legacy .changeset/ directory (no longer used) - Add eval package to proman.yaml - Set eval package to public for npm publishing	2026-06-05 02:21:24 +00:00
xiaoju	63cb4d3645	fix: remove _ single-exit for user roles ... $START keeps _ (special entry node). All user-defined roles now require explicit $status enum in frontmatter + matching graph keys. - moderator: remove UNIT_STATUS fallback, error on missing $status - validate: reject _ graph keys for non-$START roles - validate-semantic: remove checkSingleExitRole(), require $status enum - update all test fixtures to use explicit status values - fix examples/analyze-topic.yaml Fixes #86	2026-06-05 02:00:45 +00:00
xiaoju	ae81e4b5ac	feat: eval report, diff, list commands ... Implement the 3 read commands for eval framework: - report: read eval-run from CAS, render formatted text (task, overall, config, judges table, thread ID) - diff: side-by-side comparison with ▲/▼ delta indicators and config change markers - list: scan @uwf/eval/*/latest variables, sort by timestamp desc, --task filter, --limit pagination Architecture: pure formatting functions (format.ts) + data access (read.ts) + thin CLI handlers. Types in types.ts. 11 new tests (formatReport, formatDiff, formatList, selectEntries) Refs #72	2026-06-05 00:19:25 +00:00
xiaoju	8c26f16716	feat: builtin judges — frontmatter + token-stats (deterministic) + upstream/hallucination (stubs) ... Implement 4 builtin judges for eval framework: - frontmatter-compliance: validates YAML frontmatter with $status field, score = stepsValid / stepsTotal - token-stats: aggregates Usage from step nodes, always score 1.0 (informational only) - upstream-consumption: LLM-as-judge stub (score 0, TODO) - hallucination: LLM-as-judge stub (score 0, TODO) Infrastructure: - judge/builtin/read-steps.ts — shell out to uwf step list - judge/builtin/types.ts — BuiltinJudge, BuiltinJudgeOutput - runner/collect.ts — dispatch builtin judges by name 9 new tests (frontmatter validation + token aggregation) Refs #71	2026-06-05 00:09:06 +00:00
xiaoju	fae9e9ed3a	feat: eval run command — prepare, execute, collect pipeline ... Implement the uwf-eval run <task-dir> command with 3-phase pipeline: - prepare: read task.yaml, copy fixture/ to temp workdir - execute: shell out to uwf thread start + exec - collect: run judges, compute weighted score, store CAS node, set @uwf/eval/<task>/latest variable Changes: - src/runner/ — types, prepare, execute, collect, index - src/storage/store.ts — createEvalStore(), setEvalLatest() - src/commands/run.ts — full pipeline wiring with --agent/--model/--count - 9 new tests (prepare + collect + weighted scoring) Builtin judges return placeholder score 0 (Phase 1c). Refs #70	2026-06-04 23:59:21 +00:00
xiaoju	99619d85db	feat: eval package scaffold with CLI, schemas, types, task loader ... New package @united-workforce/eval (uwf-eval CLI): - CLI skeleton: run/report/diff/list subcommands (stubs) - 5 OCAS schemas: eval-run, judge-frontmatter, judge-upstream, judge-hallucination, judge-token-stats - TaskManifest type + parser/validator for task.yaml - JudgeOutput/JudgeInput types for judge contract - EvalRunPayload/EvalRunConfig/EvalJudgeRecord storage types - 19 unit tests: task loader validation + schema definitions Refs #69	2026-06-04 23:42:16 +00:00
xiaoju	1593dbb521	fix: compute usage as delta for session re-entry ... On session resume, turns/inputTokens/outputTokens were cumulative (entire session history) instead of per-step increments. Now we snapshot metrics before prompt, compare after, and report the delta. Changes: - acp-client: add getSessionId() accessor - hermes: extract snapshotUsage() + computeUsageDelta() pure functions - hermes: runPrompt/runHermes/continueHermes use before/after snapshots - 9 new unit tests for usage delta computation Refs #68	2026-06-04 23:22:16 +00:00
xiaoju	d1c523c442	feat: agent-hermes reads real token counts from session DB ... - Add inputTokens/outputTokens to HermesSessionJson type - Query input_tokens, output_tokens from sessions table in loadHermesSessionFromDb - Update test fixture schema with token columns - runPrompt now reports real token counts from Hermes state.db Refs #76, #68	2026-06-04 23:06:52 +00:00
xiaoju	be92cb2dd2	feat: agent-claude-code reports real $usage from stream-json output ... - Map parsed numTurns, inputTokens, outputTokens, durationMs to Usage type - Add @united-workforce/protocol dependency + tsconfig reference - 747 tests pass Fixes #77 Refs #68	2026-06-04 22:36:44 +00:00
xiaoju	7681e8b8e2	feat: agent-hermes reports $usage (turns + duration) ... - Count assistant turns from session messages - Measure wall-clock duration per prompt call - inputTokens/outputTokens remain 0 (ACP protocol doesn't expose token data yet) - Both runPrompt and continueHermes report usage Fixes #76 Refs #68	2026-06-04 22:30:14 +00:00
xiaoju	248ac710fd	feat: agent-mock emits fixed $usage stats ... - Mock agent returns {turns:1, inputTokens:0, outputTokens:0, duration:0} - E2E test 1 (linear workflow) asserts usage in CAS step nodes - 747 tests pass Fixes #75 Refs #68	2026-06-04 22:19:29 +00:00
xiaomo	172c232e61	Merge pull request 'feat: add $usage field to adapter protocol' (#80) from feat/74-usage-in-protocol into main ... feat: add $usage field to adapter protocol (#80)	2026-06-04 22:14:12 +00:00
xiaoju	99f40c2488	feat: add $usage field to adapter protocol ... - Add Usage type to protocol (turns, inputTokens, outputTokens, duration) - Add usage to StepRecord, StepNodePayload, StepEntry, STEP_NODE_SCHEMA - Thread usage through util-agent extract pipeline (writeStepNode → persistStep → createAgent) - All adapters return usage: null as placeholder (mock, hermes, claude-code, builtin) - 746 tests pass, no breaking changes (usage not in schema required array) Fixes #74 Refs #68	2026-06-04 15:41:07 +00:00
xingyue	bf489c59a5	fix: agent bin fields point to dist/cli.js instead of src/cli.ts ... All three agent packages had bin pointing to ./src/cli.ts (bun-era leftover). Node cannot execute .ts files directly, causing ERR_MODULE_NOT_FOUND when spawning agents. Closes #78	2026-06-04 23:25:39 +08:00
xingyue	83bcda60ff	refactor(prompt): rename subcommands and add frontmatter output ... - Rename: user→usage-reference, author→workflow-authoring, adapter→adapter-developing - Remove: developer (content lives in CLAUDE.md) - All prompts output complete SKILL.md with YAML frontmatter - Setup instructions simplified: uwf prompt bootstrap > SKILL.md - Remove all bun references, use pnpm/npm - Fix CLAUDE.md: fixed→independent versioning - Delete old reference files (user/author/developer/adapter) Closes #66	2026-06-04 22:46:11 +08:00
xiaoju	3401873051	chore: rebranding cleanup — reset versions to 0.1.0, bun→pnpm in docs ... - All 9 packages reset to version 0.1.0 - CLAUDE.md: bun→pnpm, fixed→independent versioning, proman commands - docs/architecture.md: bun→pnpm in toolchain table - docs/sync-readme.md: bun→pnpm in conventions	2026-06-04 13:05:26 +00:00
xiaoju	18170a4313	refactor: extract validateCount, replace CLI spawn with direct import ... - Extract validateCount() from cmdThreadExec (throw instead of process.exit) - 5 validation tests now import validateCount directly (no subprocess) - Only --help tests still spawn CLI (need Commander output) - Test time: 1.7s → 475ms Fixes #61	2026-06-04 12:31:17 +00:00
xiaoju	8bf5b88172	chore: remove integration tests, clean up CI exclusion ... Deleted: - acp-client.integration.test.ts (3 cases) - resume-e2e.integration.test.ts (1 case, already skipped) These tests spawn a real hermes CLI and hit live LLM, belonging to the eval layer (#34), not CI. ACP protocol parsing is already covered by unit test acp-client.test.ts. Also removed the --exclude integration/ hack from test:ci. Fixes #60	2026-06-04 12:19:24 +00:00
xiaoju	66c2e2a79b	fix: use node dist/cli.js instead of npx tsx in thread-step-count tests ... npx tsx hangs in CI Docker (30s+ timeout). node dist/cli.js runs in <2s.	2026-06-04 11:57:32 +00:00
xiaoju	58b58d511e	fix: add timeout to cmdThreadExec count logic tests	2026-06-04 11:48:46 +00:00
xiaoju	596c05bfcc	fix: use node dist/cli.js instead of npx tsx in prompt help test ... npx tsx fails in CI (tsx not found, npm tries to install it)	2026-06-04 11:32:09 +00:00
xiaoju	d26f54e8ea	fix: biome format + remove unused noConsole suppressions	2026-06-04 11:22:46 +00:00
xiaoju	883bd79bcb	fix: add timeout to CI-slow tests + check stderr for help output	2026-06-04 11:18:49 +00:00
xiaoju	63454a4cfd	fix: OCAS_DIR → OCAS_HOME in test helpers + exclude integration tests from CI ... - Remaining OCAS_DIR references caused test isolation failures - agent-hermes integration tests need 'hermes' CLI, skip in CI Fixes #58	2026-06-04 11:06:42 +00:00
xiaoju	9f5891169e	fix: add missing workflow destructure in current-role test ... The createMarker call used shorthand 'workflow' but the variable was not destructured from cmdThreadStart. Fixes #56	2026-06-04 10:56:44 +00:00
xiaomo	f56e24cf82	Merge pull request 'test: expand E2E coverage — suspend, count, mustache, completed resume' (#51) from test/33-more-e2e into main ... test: expand E2E coverage — suspend, count, mustache, completed resume (#51)	2026-06-04 09:04:09 +00:00
xiaoju	974c2b8f1b	test: add E2E tests for suspend/resume, --count, mustache, and completed resume (#33) ... 4 new E2E scenarios: 4. $SUSPEND → resume lifecycle (suspendedRole/suspendMessage metadata) 5. --count 3 runs entire pipeline in one invocation 6. mustache template variables rendered into edgePrompt 7. completed thread resume (衔尾蛇: end → start, CAS chain preserved) Total: 7 E2E scenarios, all passing.	2026-06-04 09:03:01 +00:00