b4ba65b9ca
fix(gui): 修复启动报错和对话列表不正确的两个关键Bug
...
Bug1: 'async for' requires __aiter__ method, got coroutine
- EventQueue.subscribe() 在 _closed=True 时直接 return，
Python 将其视为协程而非异步生成器
- 修复: 添加不可达的 yield 语句，确保函数始终为异步生成器
Bug2: 启动时对话列表全显示"对话"，无法识别之前的对话
- list_conversations() 不加载消息，_derive_conversation_title
遍历空 messages 列表导致标题全为"对话"
- 修复: list_conversations 从 SQLite 加载首条用户消息用于标题推导
Bug2b: WebSocket 不响应前端对话切换
- conv 变量只在首条消息时设置，之后忽略 conversation_id
- 修复: 每条消息都检查 conversation_id，切换时更新 conv
2026-06-18 16:26:02 +08:00
5b5291c7e5
fix: WebSocket task persistence three-layer defense with security hardening
...
Fix chat history empty content and task stops on refresh. Implements: result persistence on disconnect, task backgrounding via asyncio + EventQueue, frontend reconnection recovery. Security: fail-closed conversation_id ownership, asyncio.shield on CancelledError cleanup, async TaskStore shim, EventQueue subscriber limit, connection error resilience. 23 tests added.
2026-06-17 22:11:51 +08:00
bbedfff597
feat: hub-and-spoke experts, tiered tool injection, unified event model (U3/U7/U10)
2026-06-17 10:46:16 +08:00
200174c5c7
feat: SQLite persistence, verification loop, spec-driven execution
...
Phase 2 of architecture optimization (U5/U6/U9):
- U5: SqliteConversationStore with WAL mode + LRU cache (1000 convs)
Replaces in-memory ConversationStore in portal.py
Data survives server restarts (ref: Codex Thread persistence)
- U6: VerificationLoop with verify/verify_and_retry
Default commands: pytest + ruff check
ReActEngine integration via verification_enabled flag
New run_tests tool for LLM to invoke verification
- U9: SpecManager for plan-as-contract (ref: Qoder Quest Mode)
Plans persisted to .agentkit/specs/{spec_id}.yaml
API: GET/PUT /api/v1/specs, POST /api/v1/specs/{id}/confirm
PlanExecEngine emits spec_created event after plan generation
Also fixes: portal skill_name routing, app.py SessionManager guard,
test_telemetry CostAwareRouter removal, test_compression_config fixture
2026-06-17 10:45:20 +08:00
5374bc8501
refactor: eliminate routing layer, align with industry best practices
...
Phase 1 of architecture optimization (U1/U2/U4/U8):
- U1: Rename SimpleRouter to RequestPreprocessor, route() to preprocess()
Eliminates misleading routing concept; LLM decides autonomously
in REACT agent loop (matches Codex/Claude Code/Trae pattern)
- U2: Delete CostAwareRouter, HeuristicClassifier, SemanticRouter
(~700 lines removed). skill_routing.py: 1688 to 220 lines
- U4: PlanExecEngine defaults to ReActStepExecutor, delete _LLMStepExecutor
(pure LLM calls without tools = no execution capability)
- U8: ReActEngine defaults to ContextCompressor(keep_recent=10)
Supersedes plans 2026-06-15-002/003/004.
New plan: 2026-06-16-006-refactor-architecture-optimization-evolution-plan.md
2026-06-17 10:44:40 +08:00
b54213b3c6
fix(review): resolve all P0/P1/P2 findings from code review
2026-06-16 09:08:03 +08:00
16ac592855
feat(gateway): empty response auto-retry with fallback model chain
2026-06-16 08:07:21 +08:00
9caf332e9e
fix: ensure agent never returns empty result to user
2026-06-16 08:01:43 +08:00
c4257591d4
refactor(router): replace CostAwareRouter with SimpleRouter and prompt-based tool calling
2026-06-16 03:31:05 +08:00
64d62a2b60
feat: autonomous task execution - connect PlanExecEngine + TeamOrchestrator
...
U1: TeamOrchestrator._execute_phase real execution (Expert.agent.execute)
U2: LLM-based merge strategies (BEST/VOTE/FUSION) with fallback
U3: ReActStepExecutor replacing _LLMStepAgent for tool-enabled steps
U4: SharedWorkspace integration for cross-phase/cross-execution state
U5: GoalPlanner prompt tuning with few-shot and verb pattern matching
U6: Replan-before-fallback in TeamOrchestrator
U7: End-to-end validation tests for multi-step research tasks
U8: WebSocket progress events (step_event_callback + new event types)
Code review fixes: P0 response.strip fix, P1 competitor status check,
milestone real impl, VOTE self-bias fix, confirmation_handler wiring,
ExpertTeam public API, DRY _build_result_summaries, replan tests
Also: geo_server.py refactor (ServerConfig.from_yaml), delete llm_config.yaml
2026-06-15 12:41:32 +08:00
99fe4c99f7
fix: comprehensive code review fixes + WS test stability
2026-06-15 08:17:34 +08:00
7384ecb03e
feat: Expert Team Mode — plan-execute collaboration with conversation UI
...
Implements B+C hybrid Expert Team Mode with ExpertConfig, CollaborationPlan,
TeamOrchestrator, ExpertTeamRouter, HandoffTransport, SharedWorkspace, and
Expert wrapper. Frontend includes ExpertTeamView, ExpertMessage,
PlanVisualization, team store, and WS event handlers.
Code review fixes: sentinel-based close, per-phase retry, name validation,
Vue component integration, teamState dedup, Redis reset, plan reassign,
event_type validation, hmac timing-safe compare, message dedup,
reactive updatePhases, O(1) phase lookup, iterative DFS, bounded Queue.
232 unit tests passing.
2026-06-14 22:20:14 +08:00
0ccef7be5c
feat: P0 production hardening — LLM cache, semantic routing, state persistence
...
U1: LLM Cache Core (exact + semantic match, InMemory + Redis backends)
U2: Cache integration into LLMGateway with CacheConfig
U3: Semantic Router as Layer 1.5 in CostAwareRouter
U4: UsageStore persistence (Redis Hash + InMemory fallback)
U5: CascadeStateStore persistence (Redis INCR + InMemory TTL)
U6: EvolutionStore interface unification (Protocol + PostgreSQL backend)
U7: Configuration integration + E2E tests
Code review fixes:
- P0: date iteration bug (day>=28), semantic router index never built,
Redis connection leak (per-call → persistent pool)
- P1: cache degradation recovery, semantic_search degradation,
double miss counting, asyncio.Lock for PG init, LIMIT on queries,
__import__ anti-pattern → _utcnow()
- P2: InMemory TTL cleanup, embedding preservation on put(),
data TTL = max(exact_ttl, semantic_ttl)
2026-06-14 15:16:00 +08:00
5ef08a3b30
fix(review): comprehensive P0-P2 code review fixes
2026-06-12 22:18:25 +08:00
2e55aae775
fix(review): address code review findings for speed optimization
...
- P0: Rename WAL buffer to pending buffer, add crash-loss warning
- P1: Fix keyword substring false matches with word-boundary regex
- P1: Pass connection pool params in _build_llm_config
- P1: Change parallel_tools default to False (safer default)
- P1: Add classifier value validation in CostAwareRouter
- P2: Replace __import__ with proper datetime import
- P2: Add max_buffer_size enforcement in AsyncWriteQueue
2026-06-12 13:21:44 +08:00
a36bc3d1c1
feat: optimize chat response speed for sub-1s first token latency
...
- Add HeuristicClassifier to replace LLM quick_classify with zero-cost
local heuristic (keyword/length/code-pattern scoring), gated by
router.classifier config (default: heuristic)
- Add parallel tool execution in ReActEngine via asyncio.gather for
multiple independent tool_calls, gated by parallel_tools param
- Add AsyncWriteQueue for non-blocking session persistence with WAL
buffer, gated by async_writes param on SessionManager
- Add httpx.Limits connection pool config to all LLM providers
- Add router config section to ServerConfig and agentkit.yaml
- All optimizations have config switches for safe rollback
2026-06-12 13:15:06 +08:00
55421dd126
fix: get_tools() and get_system_prompt() now read from tool_registry too
...
Root cause: app.py registers tools via agent._tool_registry.register()
which adds to the ToolRegistry but NOT to agent._tools (which is only
populated by use_tool() from config). Both get_tools() and
get_system_prompt() were reading only _tools, missing all post-init
registered tools. Now both methods merge _tools with
_tool_registry.list_tools().
2026-06-11 22:17:14 +08:00
f7225bc91a
fix: include available tools in system prompt so LLM knows what it can call
...
Previously get_system_prompt() only returned identity/instructions but
did not tell the LLM what tools are available. The LLM would therefore
refuse to call tools even when they were registered, saying it had no
tools. Now the system prompt includes a '## 可用工具' section listing
all registered tools with their descriptions and parameters.
2026-06-11 22:00:30 +08:00
32c800d1e4
fix: portal routing + response speed + IME input
...
1. Portal unified routing: ws_chat now uses CostAwareRouter uniformly
(handles Layer 0/1/2), replacing direct IntentRouter calls.
Greeting/chat_mode requests skip IntentRouter LLM call entirely.
2. Response speed: greeting & simple chat now use direct LLM call
(no ReAct loop), zero-cost Layer 0 detection.
3. IME input fix: use e.isComposing (native browser property)
instead of compositionstart/end for Enter key detection.
4. Test: fix InMemoryMessageBus.request() parameter name
timeout -> timeout_seconds.
2026-06-11 21:30:25 +08:00
d47f279887
fix: resolve code review issues from deferred improvements
...
1. InMemoryMessageBus.request(): fix param name (timeout→timeout_seconds) to match ABC
2. InMemoryMessageBus: track consumer tasks, cancel on unsubscribe
3. InMemoryMessageBus: _try_resolve_pending() in queue consumer path
4. evolve_soul(): use "default" category when patterns is empty
5. quick_classify(): use delimiter-based prompt to mitigate injection risk
6. Use asyncio.get_running_loop() instead of deprecated get_event_loop()
2026-06-11 13:49:02 +08:00
ec51dbb259
feat: optimize劣势项 — 拍卖开关/审计采样/线程安全/评分锚定
...
1. 拍卖机制: 已有配置开关(marketplace.auction_enabled), 默认关闭
2. LLM审计采样: 新增 audit_sample_rate (0.0-1.0), 默认1.0, 可降低审计频率
3. AlignmentConfig.from_dict: 忽略未知键, 防止YAML额外字段崩溃
4. 配置热重载线程安全: 用 threading.Event 替代布尔标志, 消除数据竞态
5. Reflexion评分锚定: 添加评分维度(Completeness/Correctness/Clarity)和锚定点
2026-06-11 13:04:36 +08:00
cc2cd414c9
fix: resolve all code review issues from cross-validation
...
1. Critical: Add missing TaskResult import in plan_exec_engine.py
2. Critical: Fix ReWOOEngine param name (max_steps → max_plan_steps)
3. Major: Remove duplicate token counting in reflexion.py
4. Major: LLM audit failure now passes (trusts rule check) instead of failing
5. Major: Fix dict iteration with del using list() copy in lifecycle.py
6. Major: Fix Chinese content tokenization using regex split instead of space split
7. Minor: _is_positive_mention now checks all occurrences, not just the first
2026-06-11 06:22:35 +08:00
79eb8469f9
fix: address remaining code review issues
...
- AlignmentGuard: direction-aware constraint checking (negation/affirmation detection)
instead of simple substring matching to reduce false positives
- Reflexion: extract actual token usage from LLM response instead of hardcoded 1
- MemoryTool: protect version/history sections from update_soul modification
- Fix AsyncMock warnings for sync find_best_agent method
2026-06-11 00:14:11 +08:00
5b42487d8a
feat(core): add ReWOO, Plan-and-Execute, Reflexion execution engines
...
Phase A of Multi-Agent Marketplace architecture:
- ReWOOEngine: plan-all-then-execute pattern for parallel data fetch
- PlanExecEngine: adapter wrapping GoalPlanner+PlanExecutor+PipelineReplanner
- ReflexionEngine: ReAct + Evaluate + Reflect + Retry for high-precision tasks
- SkillConfig: extend VALID_EXECUTION_MODES with rewoo/plan_exec/reflexion
- ConfigDrivenAgent: add _handle_rewoo/_handle_plan_exec/_handle_reflexion routes
- 5 professional agent YAML configs with layered model defaults
- 107 unit tests passing
2026-06-10 17:08:48 +08:00
6852dfe892
fix(security,reliability): resolve all P2 findings from code review
2026-06-10 15:05:40 +08:00
658e188939
fix(review): resolve P0/P1 findings from final code review
2026-06-10 09:57:29 +08:00
1d1805753c
fix: resolve key P2 findings from code review
...
- Shell whitelist: use exact binary match instead of startswith
- Shell audit log: use deque(maxlen=10000) to cap memory
- Terminal history: use deque(maxlen) for O(1) eviction
- Path optimizer: cap _pending_paths at 50 entries per task_type
- Pitfall detector: only add tips to matching steps, not all
- Experience store: handle non-numeric _parse_time_window input
- Extract shared is_safe_url() to utils/security.py (DRY)
- Workflow condition evaluator: handle float() ValueError
2026-06-10 09:01:23 +08:00
7874e875af
merge: integrate feat/agentkit-phase8-chat-adaptive (chat/gui commands + GUI mode)
...
Restores agentkit chat, agentkit gui CLI commands, onboarding wizard,
and GUI mode (AGENTKIT_GUI_MODE) with static file serving.
Resolves merge conflicts in orchestrator.py, app.py, tools/__init__.py, shell.py.
2026-06-10 07:44:06 +08:00
9e9f1314f6
fix(security): resolve all P0/P1 findings from code review
2026-06-10 07:12:41 +08:00
fd4a811929
feat(phase1): implement core kernel and experience foundation (U1-U5)
...
- U1: GoalPlanner - structured goal decomposition wrapping _decompose_task()
- U2: PlanExecutor - parallel execution with retry/skip/replace strategies
- U3: PlanChecker - quality gate + review + experience writing
- U4: Skill spec upgrade - dependencies, capabilities, version management
- U5: ExperienceStore - PostgreSQL+pgvector task experience storage
208 new tests passing, fully backward compatible.
2026-06-09 23:57:03 +08:00
45283d31e8
feat(core): integrate MessageBus into Orchestrator and AgentPool (U7)
...
- Orchestrator accepts optional message_bus parameter; workers publish
task.progress messages via MessageBus after each subtask execution
- AgentPool accepts optional message_bus; auto-registers agents on
create and auto-unregisters on remove
- app.py initializes MessageBus from config and injects into AgentPool
- ServerConfig adds bus configuration field
- 5 new tests, all passing
2026-06-08 00:03:40 +08:00
88d8298871
feat(core): add Orchestrator adaptive task decomposition (U5)
...
- execute_adaptive(): iterative execute→evaluate→re-decompose loop
- OrchestratorConfig: adaptive, max_iterations, quality_threshold
- _evaluate_quality(): LLM-based or rule-based quality scoring (0-1)
- _reexecute_failed(): preserves completed subtask results, retries
failed ones with improvement feedback injected into input_data
- OrchestrationResult.metadata field for tracking iteration history
- 10 new tests, all passing
2026-06-07 23:50:54 +08:00
7054ac02b6
feat(tools): add AskHumanTool + token streaming in ReAct execute_stream
...
- AskHumanTool: Human-in-the-Loop tool for Chat mode, pushes questions
via WebSocket callback and waits for user reply via asyncio.Future
- Token streaming: execute_stream() now uses chat_stream() instead of
chat(), yielding token-type ReActEvents for each StreamChunk
- _build_response_from_stream() static method constructs LLMResponse
from accumulated stream data
- Export AskHumanTool from tools/__init__.py
- 12 new tests (7 AskHumanTool + 5 token streaming), all passing
2026-06-07 23:40:43 +08:00
b34b06724d
fix(agentkit): resolve all P0/P1/P2/P3 issues from code review
2026-06-07 22:05:18 +08:00
286804792d
feat(compression): U4 ServerConfig compression field and Agent injection
...
Add compression config to ServerConfig (following telemetry pattern),
create compressor in create_app, pass through AgentPool to
ConfigDrivenAgent, and inject into ReActEngine.execute() calls.
2026-06-07 18:20:05 +08:00
fcb4fb33f3
feat(compression): U3 ReAct engine tool result compression and incremental compress
...
Extend _build_tool_result_message to accept compressor parameter for
tool output compression. Add _should_compress helper for token budget
checking. Add incremental compression within ReAct loop when
conversation exceeds threshold.
2026-06-07 18:19:53 +08:00
ea705b979b
feat(compression): U2 HeadroomCompressor with SmartCrusher and CCR cache
...
Add HeadroomCompressor implementing CompressionStrategy Protocol with
content-type routing (JSON→SmartCrusher, code→CodeCompressor), CCR
reversible compression cache, and graceful degradation when headroom-ai
is not installed.
2026-06-07 18:19:41 +08:00
5d3a5f2bf3
feat(compression): U1 CompressionStrategy Protocol and create_compressor factory
...
Add runtime-checkable CompressionStrategy Protocol with compress(),
compress_tool_result(), and is_available() methods. Add compress_tool_result
and is_available to existing ContextCompressor. Add create_compressor()
factory function with headroom/summary provider routing and ImportError
fallback.
2026-06-07 18:19:27 +08:00
239009357a
feat(telemetry): U7 OpenTelemetry integration with zero-dependency no-op pattern
...
Add telemetry module with tracing (agent/tool/llm/pipeline_step spans),
metrics (5 histograms/counters), and setup with optional OTLP exporters.
Uses no-op pattern when opentelemetry not installed. GenAI Semantic
Conventions for LLM spans. Integrated into ReactEngine, LLMGateway,
ToolBase, and FastAPI app.
2026-06-07 17:26:21 +08:00
24e501f745
fix(core): U10 Agent status lock timeout and config hot-reload audit
...
- Added _acquire_status_lock with timeout (30s) to prevent deadlocks
- Added _release_status_lock for safe lock release
- Added config_version tracking on BaseAgent
- Config hot-reload now increments version and propagates to agents
- Audit logging with config version in _on_config_change
2026-06-06 22:52:51 +08:00
23934602c0
feat(core): U4 multi-agent Orchestrator with SharedWorkspace
...
- Orchestrator: Orchestrator-Worker pattern with LLM-driven task decomposition
- SharedWorkspace: Redis-backed shared state with versioning and distributed locks
- SubTask dependency graph, parallel group building, result aggregation
- 16 tests passing
2026-06-06 22:25:12 +08:00
6e362a8ae7
feat(agentkit): Phase 4 enterprise production upgrade — 12 Implementation Units
...
Phase A (P0): EpisodicMemory pgvector search+EmbeddingCache, ReAct timeout+CancellationToken, evolution system fix (A/B test+LLMPromptOptimizer+StrategyTuner), AnthropicProvider native Messages API
Phase B (P1): RetryPolicy+CircuitBreaker, chat_stream fallback chain, WebSocket endpoint, SSE stream fix, Evolution+Memory API routes (7 endpoints), embedding cache+Enhanced Search per-KB degradation fix
Phase C (P2): GeminiProvider native generateContent API, Agent state lock+config hot-reload
Tests: 1301 passed, 18 skipped, 0 failed
2026-06-06 21:51:04 +08:00
e33dc25ad3
feat(memory): RAG pipeline optimization — 5 Implementation Units
...
U1: QueryTransformer — LLM/rule-based query rewriting + sub-query decomposition
U2: HttpRAGService enhanced_search() — rerank + compression via /bases/{kb_id}/retrieve
U3: Structured context injection — source attribution headers in RAG results
U4: RetrieveKnowledgeTool — built-in tool for mid-reasoning knowledge retrieval
U5: Configurable retrieval params + per-KB weights + CJK token estimation
Config example:
memory:
retrieval:
top_k: 5
token_budget: 2000
context_template: structured
query_transform:
enabled: true
strategy: llm
semantic:
search_mode: enhanced
use_rerank: true
kb_weights:
industry-kb-id: 1.2
enterprise-kb-id: 0.8
Tests: 1037 passed, 18 skipped, 0 failed
2026-06-06 19:27:09 +08:00
cd5b39087e
feat(memory): add HttpRAGService for config-driven knowledge base integration
2026-06-06 18:36:05 +08:00
0456429beb
fix(review): address all 14 P2 advisory findings
2026-06-06 18:20:46 +08:00
8620751864
fix(review): address P0+P1 findings from Tier 2 code review
...
P0: MemoryRetriever.retrieve score mutation fix
P1: Redis atomic Lua script, deprecated API fix, SQLite WAL mode,
Redis URL masking, UniqueConstraint, TraceRecorder completed flag,
EpisodicMemory recall improvement, LLMReflector sanitization,
A/B test safety, generator cleanup, ContextCompressor guards,
OpenAIEmbedder reuse, Pipeline failure handling, Metrics O(1),
Health check Redis PING, CLI skill loading, CORS config,
API key direct pass-through
Tests: 924 passed, 18 skipped, 0 failed
2026-06-06 17:57:47 +08:00
f858d279f3
feat(agentkit): Phase 3 upgrade - persistence, memory, evolution, observability
...
10 Implementation Units across 3 phases:
Phase A - Infrastructure:
- U1: RedisTaskStore with Redis/memory backend + factory function
- U2: TraceRecorder for execution trace recording
- U3: PersistentEvolutionStore with SQLite backend
Phase B - Core Capabilities:
- U4: MemoryRetriever integration into ReAct engine
- U5: Embedder abstraction + EpisodicMemory vector search
- U6: LLMReflector for LLM-in-the-loop reflection
- U7: SkillPipeline for multi-skill orchestration
Phase C - Enhancement:
- U8: SKILL.md format + progressive disclosure levels
- U9: ContextCompressor + prompt cache rendering
- U10: Structured logging + metrics endpoint + enhanced health check
Tests: 924 passed, 18 skipped, 0 failed
2026-06-06 17:17:45 +08:00
acec8ff743
feat(evolution): Phase A - lifecycle hooks + EvolutionConfig
...
U11: EvolutionMixin integrated into ConfigDrivenAgent lifecycle
- on_task_complete triggers evolve_after_task
- on_task_failed records failure patterns
- Evolution errors never break main task flow
U12: EvolutionConfig added to SkillConfig
- enabled, reflect_on_failure, auto_apply, min_quality_threshold
- Backward compatible: defaults to enabled=False
21 new tests passing, no regression.
2026-06-06 12:05:56 +08:00
2844eeb548
feat(streaming): Phase C - LLM streaming + ReAct events + SSE endpoint
...
U8: StreamChunk protocol + OpenAI chat_stream + Gateway streaming with usage tracking
U9: ReActEvent dataclass + execute_stream() yielding thinking/tool_call/tool_result/final_answer
U10: POST /tasks/stream SSE endpoint + Client SDK stream_task()
15 new tests passing, no regression.
2026-06-06 11:54:17 +08:00
5f1c51cf9a
feat(server): Phase B - auth, rate limiting, SSRF protection, handler whitelist
...
U1: API Key authentication middleware (dev mode skip, health whitelist)
U2: Rate limiting middleware (fixed-window, 60 req/min default)
U3: Callback URL SSRF protection (private IP blocking)
U4: custom_handler module prefix whitelist
65 tests passing. CORS conflict fixed.
2026-06-05 23:37:36 +08:00
chiguyong