feat(roundtable): 会商席位「技能识别强化」改为按需开关(智能体级 + 链条每席位级)

弱模型补偿指令从全域生效改为默认关闭、两个 opt-in 开关取 OR:
- 智能体级: AgentConfig.seat_skill_directive(config.yaml),在智能体编辑弹窗
  (agent-card / agent-detail-panel)开关;agents 路由 create/update/读写全程穿。
- 业务链条每席位级: 存进链条 seats JSON(ChainSeatSchema.seat_skill_directive,
  无需 DB 列/迁移),在链条编辑器每个席位卡片右侧 ✨ 图标→下拉框(开关+说明)配置;
  按 agentId 查表穿到前台 useStep2Orchestration 各席位调用 + 后台作业
  seat_skill_directives:{agentId:bool} → InProcessRoundtableGateway.run_seat。
- ROUNDTABLE_SEAT_SKILL_DIRECTIVE 降级为总熔断开关(默认放行,不再 force-on)。

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This commit is contained in:
董宏凯andClaude Opus 4.8 committed 2026-06-20 13:37:29 +08:00
1 parent a8c06ca7dc
commit 9a1b674950
24 files changed
+298 -29

No files matched your search

+1 -1
View File
@@ -247,7 +247,7 @@ FastAPI application on port 8001 with health check at `GET /health`. Set `GATEWA
| **Public Shares** (`/api/public/shares`) | **Unauthenticated** (the `/api/public/` prefix is whitelisted in `auth_middleware._PUBLIC_PATH_PREFIXES`). `GET /{code}` - read-only conversation snapshot (title + serialized messages + artifacts) for a `view`-mode share; messages are read straight from the latest checkpoint. `GET /{code}/files/{path}` - serves a file referenced by that conversation (images, Markdown, text, PDF, ?) so the shared page's links are openable; path resolution is confined to the owner's thread directory. Active content (`text/html`, `application/xhtml+xml`, `image/svg+xml`) is forced to download (`_public_share_disposition`); everything else is served inline. Missing/revoked/non-`view` codes all return 404. |
| **Roundtable Drafts** (`/api/roundtable-drafts`) | **Strictly per-user** persistence for the roundtable-planning page (replaces the old browser `localStorage`). `GET /` - list drafts (lightweight meta); `POST /` - create; `GET/PUT/DELETE /{id}` - full read / partial update / delete (all owner-scoped). A draft = one full session (Step 1 intent + Step 2 roundtable), step snapshots stored as JSON in `PortableLongText`. Recommendation history bound to a draft: `GET /{id}/recommendations` - list newest-first; `POST /{id}/recommendations` - append one run (objective/status/rationale/picks/candidates). taskId 深链会商记录 are **not** here — they live in the separate `/api/roundtable-task-drafts` store (see below) so personal history never mixes with task records. Backed by `deerflow.persistence.roundtable_drafts` (tables `roundtable_drafts`, `roundtable_recommend_history`); store wired in `deps.py` as `app.state.roundtable_draft_store`. Tests: `tests/test_roundtable_task_drafts.py`. |
| **Roundtable Task Drafts** (`/api/roundtable-task-drafts`) | **Task-scoped, unpartitioned** store for taskId-deeplink roundtable 聊天记录 (无界嵌入抽屉). Keyed **only by `task_id`** (one task → many records), **never partitioned by user** — any authenticated caller may read/write/delete a task's records (`created_by` is audit-only). Physically separate from `roundtable_drafts` so the two never mix. `GET /by-task/{task_id}` - all drafts for a task, newest-first (history dropdown), enriched with the latest background-job status per draft via `RoundtableJobRepository.status_map_by_drafts`; `GET /by-task/{task_id}/latest` - newest draft (回显最新一条); `POST /` - create (`task_id` required); `GET/PUT/DELETE /{id}` - full read / partial update / delete (all user-agnostic). Backed by `deerflow.persistence.roundtable_task_drafts` (table `roundtable_task_drafts`, auto-created by `create_all`; migration `20260618_01`); store wired in `deps.py` as `app.state.roundtable_task_draft_store`. Existing `task_id`-bearing `roundtable_drafts` rows are migrated over by `scripts/migrate_roundtable_task_drafts.py`. Tests: `tests/test_roundtable_task_drafts.py`, `tests/test_roundtable_drafts_by_task.py`. |
| **Roundtable Diagnostics** (`/api/roundtable-diagnostics`) | 会商全流程诊断日志 (报错 + 关键事件). One row per event in `roundtable_diagnostics`: `{scope (background = 后台挂起 / foreground = 前台服务端观测 / client = 前端上报), stage (step1_intent/step2_init/step2_leader/step2_seat/step2_recommend/step2_orchestration/step3_summary/step3_report/step3_dashboard/step3_structure/upstream/job), level (info/warning/error), event, message, detail (JSON), job_id/draft_id/task_id/user_id/agent_id/agent_name/cycle, created_at}`. **Reads are admin-only**: `GET /` - filter (scope/stage/level/event/job_id/draft_id/task_id/user_id/q/since/until; `scope` accepts a comma-separated set so 前端「前台调用」= `foreground,client`) + paginate (newest-first), each item enriched with `userName` (user_id → 邮箱 `@`-prefix via `get_local_provider().get_user`, best-effort batched cache); `GET /facets` - distinct values for filter dropdowns; `POST /admin/cleanup` - retention purge. **`POST /report` is NOT admin-gated** — any authenticated user's frontend posts the errors *the user actually hit* (HTTP 502 before any SSE frame / network drop / SSE error frame) so they're collected too; stored as `scope="client"`. Backed by `deerflow.persistence.roundtable_diagnostics` (table `roundtable_diagnostics`, auto-created by `create_all`; migration `20260619_01`); store wired in `deps.py` as `app.state.roundtable_diagnostic_store` + a process-level default store (`register_default_store`) so the harness-side engine/gateway can write across the app/harness boundary. **Writers** (all best-effort — diagnostics never break a run): background path uses `DiagnosticsRecorder` (bound to job ctx) threaded through `RoundtableJobExecutor` → `JobProgressWriter.diag(...)` (engine transitions/step3 failures/empty-delivery) + `InProcessRoundtableGateway` (per-turn timeout/report·summary·dashboard truncation+missing); foreground path uses `app/gateway/roundtable_diag.py::record_foreground_diag` in `multi_agent.py`/`intent.py`/`recommend.py` error frames. **Bug-fix shipped alongside**: `InProcessRoundtableGateway._run_agent_turn` guards each agent run with an **inactivity (idle) timeout** (`_await_run_with_idle_guard`, `ROUNDTABLE_SEAT_IDLE_TIMEOUT_SECONDS` default 1200s = 20min, tuned high for unstable intranet models) — **not** a total-duration cap, so a slow-but-alive run that keeps streaming tokens runs as long as it needs; only a run that emits **zero StreamBridge events for `idle_timeout` seconds straight** (a true freeze: 内网模型连接池挂死 / sqlite 锁在挂载卷上挂死) is cancelled + raises (顶层把 job 标 error,而非永久冻结) + records a `seat_turn_timeout` diagnostic. Activity is observed via `app.state.stream_bridge` (the worker publishes every chunk there). Optional absolute cap `ROUNDTABLE_SEAT_MAX_TOTAL_SECONDS` (default 0 = off). Frontend: admin page `pages/RoundtableDiagnosticsPage.tsx` (route `/page/strategy/admin/roundtable-diagnostics`), opened from the **会商页右上角设置按钮** (admin only — `handleSettingsClick` in `roundtable-planning/pages/RoundtablePlanningPage.tsx`); admin read client `strategy-components/api/roundtable-diagnostics.ts`. **Client error reporter** `roundtable-planning/api/diagnostics-report.ts` (`reportRoundtableError` fire-and-forget + `keepalive`, `setRoundtableErrorContext` for draft/task attribution set by the page, `stageForRoundtableAgent`) is called at every client-observable failure: `multi-agent.ts` (`streamMultiAgent` — covers Step2 leader/seat AND Step3 report/summary/dashboard/structure — reports HTTP error + SSE `{status:"error"}` frame + worker `event:error` frame [模型调用没成功] + fetch-reject [网络不可达] + mid-stream drop [连接重置/超时], deduped via a `reported` flag; `initMultiAgent` HTTP/error-frame/incomplete/mid-stream), `intent.ts` (Step1 HTTP+error-frame), `recommend.ts` (Step2 recommend HTTP+error-frame), `roundtable-jobs.ts` (job api/stream). The admin page's first filter is a **后台挂起 / 前台调用** toggle (frontend = `foreground,client`); the stage dropdown options adapt to it (background → step2+step3+job; frontend → step1+step2+step3) and each row shows the reporting user's name + 👤. Tests: `tests/test_roundtable_diagnostics.py`, `tests/test_roundtable_inprocess_timeout.py`. **Seat skill-recognition (弱模型补偿)**: a seat = one custom agent run whose configured skills (`agent config.yaml → skills`) are injected **only** via the system prompt's `<skill_system>` block (model must `read_file` the SKILL.md itself) — weak models (deepseek-chat 等) ignore it and never use their skills. `app/gateway/roundtable_seat_skills.py::append_seat_skill_directive(agent_id, task)` appends an explicit Chinese directive (each configured-and-enabled skill's name + description + SKILL.md container path + "先 read_file 再按其流程取信息/执行") to the **seat's per-turn tLine truncated
| **Roundtable Diagnostics** (`/api/roundtable-diagnostics`) | 会商全流程诊断日志 (报错 + 关键事件). One row per event in `roundtable_diagnostics`: `{scope (background = 后台挂起 / foreground = 前台服务端观测 / client = 前端上报), stage (step1_intent/step2_init/step2_leader/step2_seat/step2_recommend/step2_orchestration/step3_summary/step3_report/step3_dashboard/step3_structure/upstream/job), level (info/warning/error), event, message, detail (JSON), job_id/draft_id/task_id/user_id/agent_id/agent_name/cycle, created_at}`. **Reads are admin-only**: `GET /` - filter (scope/stage/level/event/job_id/draft_id/task_id/user_id/q/since/until; `scope` accepts a comma-separated set so 前端「前台调用」= `foreground,client`) + paginate (newest-first), each item enriched with `userName` (user_id → 邮箱 `@`-prefix via `get_local_provider().get_user`, best-effort batched cache); `GET /facets` - distinct values for filter dropdowns; `POST /admin/cleanup` - retention purge. **`POST /report` is NOT admin-gated** — any authenticated user's frontend posts the errors *the user actually hit* (HTTP 502 before any SSE frame / network drop / SSE error frame) so they're collected too; stored as `scope="client"`. Backed by `deerflow.persistence.roundtable_diagnostics` (table `roundtable_diagnostics`, auto-created by `create_all`; migration `20260619_01`); store wired in `deps.py` as `app.state.roundtable_diagnostic_store` + a process-level default store (`register_default_store`) so the harness-side engine/gateway can write across the app/harness boundary. **Writers** (all best-effort — diagnostics never break a run): background path uses `DiagnosticsRecorder` (bound to job ctx) threaded through `RoundtableJobExecutor` → `JobProgressWriter.diag(...)` (engine transitions/step3 failures/empty-delivery) + `InProcessRoundtableGateway` (per-turn timeout/report·summary·dashboard truncation+missing); foreground path uses `app/gateway/roundtable_diag.py::record_foreground_diag` in `multi_agent.py`/`intent.py`/`recommend.py` error frames. **Bug-fix shipped alongside**: `InProcessRoundtableGateway._run_agent_turn` guards each agent run with an **inactivity (idle) timeout** (`_await_run_with_idle_guard`, `ROUNDTABLE_SEAT_IDLE_TIMEOUT_SECONDS` default 1200s = 20min, tuned high for unstable intranet models) — **not** a total-duration cap, so a slow-but-alive run that keeps streaming tokens runs as long as it needs; only a run that emits **zero StreamBridge events for `idle_timeout` seconds straight** (a true freeze: 内网模型连接池挂死 / sqlite 锁在挂载卷上挂死) is cancelled + raises (顶层把 job 标 error,而非永久冻结) + records a `seat_turn_timeout` diagnostic. Activity is observed via `app.state.stream_bridge` (the worker publishes every chunk there). Optional absolute cap `ROUNDTABLE_SEAT_MAX_TOTAL_SECONDS` (default 0 = off). Frontend: admin page `pages/RoundtableDiagnosticsPage.tsx` (route `/page/strategy/admin/roundtable-diagnostics`), opened from the **会商页右上角设置按钮** (admin only — `handleSettingsClick` in `roundtable-planning/pages/RoundtablePlanningPage.tsx`); admin read client `strategy-components/api/roundtable-diagnostics.ts`. **Client error reporter** `roundtable-planning/api/diagnostics-report.ts` (`reportRoundtableError` fire-and-forget + `keepalive`, `setRoundtableErrorContext` for draft/task attribution set by the page, `stageForRoundtableAgent`) is called at every client-observable failure: `multi-agent.ts` (`streamMultiAgent` — covers Step2 leader/seat AND Step3 report/summary/dashboard/structure — reports HTTP error + SSE `{status:"error"}` frame + worker `event:error` frame [模型调用没成功] + fetch-reject [网络不可达] + mid-stream drop [连接重置/超时], deduped via a `reported` flag; `initMultiAgent` HTTP/error-frame/incomplete/mid-stream), `intent.ts` (Step1 HTTP+error-frame), `recommend.ts` (Step2 recommend HTTP+error-frame), `roundtable-jobs.ts` (job api/stream). The admin page's first filter is a **后台挂起 / 前台调用** toggle (frontend = `foreground,client`); the stage dropdown options adapt to it (background → step2+step3+job; frontend → step1+step2+step3) and each row shows the reporting user's name + 👤. Tests: `tests/test_roundtable_diagnostics.py`, `tests/test_roundtable_inprocess_timeout.py`. **Seat skill-recognition (弱模型补偿)**: a seat = one custom agent run whose configured skills (`agent config.yaml → skills`) are injected **only** via the system prompt's `<skill_system>` block (model must `read_file` the SKILL.md itself) — weak models (deepseek-chat 等) ignore it and never use their skills. `app/gateway/roundtable_seat_skills.py::append_seat_skill_directive(agent_id, task, *, chain_enabled=False)` appends an explicit Chinese directive (each configured-and-enabled skill's name + description + SKILL.md container path + "先 read_file 再按其流程取信息/执行") toLine truncated
| **Roundtable Draft Shares** (`/api/roundtable-draft-shares`) | Per-draft sharing for the roundtable-planning page, mirroring Thread Shares but the shared resource is a `roundtable_drafts` row. Two modes (`roundtable_draft_shares.mode`). **`import`**: `POST /` `{draft_id, mode}` - create (or reuse) a revocable code; `GET /{code}/preview` + `POST /{code}/import` - copy that draft (step1/2/3) into the caller's account under a deterministic per-(code,recipient) id (`imp_<sha1>` ? re-import is a no-op). **`view`**: a public, read-only result page ? see Public Roundtable Shares. Both: `GET /` - list my codes; `DELETE /{code}` - revoke. Backed by `deerflow.persistence.roundtable_draft_shares` (table `roundtable_draft_shares`, migration `20260606_01`); store wired in `deps.py` as `app.state.roundtable_draft_share_store`. Tests: `tests/test_roundtable_draft_shares.py`. |
| **Public Roundtable Shares** (`/api/public/roundtable-shares`) | **Unauthenticated** (`/api/public/` whitelisted). `GET /{code}` - read-only snapshot for a `view`-mode draft share: returns the Step 3??????report HTML (`step3.html`, a self-contained document) + title/summary/`has_report`. The frontend renders it in a sandboxed `iframe` (no `allow-scripts`) at `#/roundtable-share/<code>`. Missing/revoked/non-`view` codes all return 404. |
| **Scheduled Tasks** (`/api/scheduled-tasks`) | Per-user cron-like agent tasks. `GET/POST /` - list/create owned tasks; `PATCH/DELETE /{id}` - update/delete (owner); `POST /{id}/{pause,resume,trigger}`; `GET /{id}/runs`, `GET /runs/{rid}`, run-file preview/edit/download; `GET /runs/{rid}/messages` - agent messages for the run's underlying agent run (used to inspect a stuck/running run; authorized via `_load_run`); `GET /runs/{rid}/html` - generated HTML document for an `html_page` run; `DELETE /runs/{rid}` - owner-scoped, with an admin fallback so admins can delete any user's run; publish/subscribe under `/{id}/{publish,unpublish,subscribe}` and `GET /published`. **Admin (cross-user):** `GET /admin/all` - every task; `PATCH/DELETE /admin/{id}` - edit/delete any task; `POST /admin/{id}/trigger`; `GET /admin/{id}/runs`. Admin routes require `system_role == "admin"`. |
@@ -339,6 +339,7 @@ async def langgraph_runtime(app: FastAPI) -> AsyncGenerator[None, None]:
seat_thinking_enabled=p.seat_thinking_enabled,
seat_reasoning_effort=p.seat_reasoning_effort,
seat_subagent_enabled=p.seat_subagent_enabled,
seat_skill_directives=p.seat_skill_directives,
# 诊断日志上下文:网关每轮 leader/seat/report/summary/dashboard 报错均带这些归属。
job_id=p.job_id,
draft_id=p.draft_id,
@@ -525,6 +525,7 @@ class InProcessRoundtableGateway:
seat_thinking_enabled: bool | None = None,
seat_reasoning_effort: str | None = None,
seat_subagent_enabled: bool = False,
seat_skill_directives: dict[str, bool] | None = None,
job_id: str | None = None,
draft_id: str | None = None,
task_id: str | None = None,
@@ -547,6 +548,10 @@ class InProcessRoundtableGateway:
self._seat_thinking_enabled = seat_thinking_enabled
self._seat_reasoning_effort = seat_reasoning_effort
self._seat_subagent_enabled = seat_subagent_enabled
# 每席位「技能识别强化」开关映射 {agent_id: bool}(沿用 Step2 所选业务链条里每个席位
# 的配置)。run_seat 按 agent_id 查表,与各席位 agent 自身的 config.yaml 开关取 OR,
# 传给 append_seat_skill_directive。默认空 dict(仅各 agent 自身开关生效)。
self._seat_skill_directives = dict(seat_skill_directives or {})
self._seat_intro = ""
self._first_leader = True
@@ -615,8 +620,12 @@ class InProcessRoundtableGateway:
policy = roundtable_run_policy("seat")
# 弱模型补偿:把该席位配置的技能(名称 + SKILL.md 路径 + 必须先 read_file 再按其
# 流程取信息的硬指令)显式拼进本轮任务,避免弱模型无视系统提示里的技能块(见
# roundtable_seat_skills 模块说明)。无配置技能时原样返回 task。
seat_task = append_seat_skill_directive(agent_id, task)
# roundtable_seat_skills 模块说明)。默认关:是否注入 = 该席位 agent 的 config.yaml
# 开关 OR 该席位在所选业务链条里的开关(按 agent_id 查表)。无配置技能 / 两开关都关时
# 原样返回 task。
seat_task = append_seat_skill_directive(
agent_id, task, chain_enabled=bool(self._seat_skill_directives.get(agent_id, False))
)
# 席位模式覆盖 policy 的 thinking/reasoning(None → policy 默认),与前端 _special_run 一致。
messages = await _run_agent_turn(
self._app,
@@ -17,7 +17,7 @@ from __future__ import annotations
import asyncio
import logging
from collections.abc import Callable
from dataclasses import dataclass
from dataclasses import dataclass, field
from datetime import UTC, datetime
from typing import Any
@@ -189,6 +189,10 @@ class StartParams:
seat_thinking_enabled: bool | None = None
seat_reasoning_effort: str | None = None
seat_subagent_enabled: bool = False
# 每席位「技能识别强化」开关映射 {agent_id: bool}(沿用 Step2 所选业务链条里每个席位的
# 配置)。透传给进程内网关,run_seat 按 agent_id 查表,与各席位 agent 自身 config.yaml 的
# 同名开关取 OR。默认空 dict。
seat_skill_directives: dict[str, bool] = field(default_factory=dict)
# 网关工厂:给定一次启动参数,返回一个 RoundtableGateway 实例。
@@ -11,12 +11,20 @@
强得多——这与本仓已有的弱模型补偿手法(``_SEAT_NO_ASK``、总控派活强化、AI 写作 SKILL.md
注入,见 ``project-ai-writing-skill-agent``)一脉相承。
**默认关闭、按需开启(2026-06-20 改造)**:不再全域生效。是否对某席位注入由两个开关
取 **OR** 决定——
1. **智能体级**:该席位 agent 的 ``config.yaml → seat_skill_directive``(智能体编辑页
可开关);
2. **业务链条级**:本次会商所选业务链条的同名开关,由调用方读出后以 ``chain_enabled``
传入(前端手动与后台挂起两条路径都按席位透传)。
任一为真即注入;两者都假 → 返回空串、原样使用任务文本(零副作用)。
两条席位执行路径共用本 helper:
- 前端手动:``multi_agent._special_run``(seat / seat_parallel 角色);
- 后台挂起:``roundtable_inprocess_gateway.InProcessRoundtableGateway.run_seat``。
无配置技能的席位 → 返回空串(完全无副作用)。可用 ``ROUNDTABLE_SEAT_SKILL_DIRECTIVE=0``
全局关闭(默认开)。
``ROUNDTABLE_SEAT_SKILL_DIRECTIVE`` 仅作**总开关(kill-switch)**:默认放行,显式设为
0/false 时彻底关闭本功能(不论两个开关如何),用于线上一键熔断。它**不再** force-on。
"""
from __future__ import annotations
@@ -27,25 +35,36 @@ import os
logger = logging.getLogger(__name__)
_TRUTHY = frozenset({"1", "true", "yes", "on", "y"})
_FALSY = frozenset({"0", "false", "no", "off", "n"})
def _directive_enabled() -> bool:
"""读 ``ROUNDTABLE_SEAT_SKILL_DIRECTIVE``(默认开)。空 / 未设 = 开。"""
return (os.getenv("ROUNDTABLE_SEAT_SKILL_DIRECTIVE") or "1").strip().lower() in _TRUTHY
def _master_enabled() -> bool:
"""``ROUNDTABLE_SEAT_SKILL_DIRECTIVE`` 总开关:默认放行(kill-switch)。
空 / 未设 / 非法 = 放行;仅显式 0/false/no/off/n 时整体关闭。注意它**只熔断不强开**,
具体是否注入仍由 agent / chain 两个开关取 OR 决定。
"""
return (os.getenv("ROUNDTABLE_SEAT_SKILL_DIRECTIVE") or "").strip().lower() not in _FALSY
def build_seat_skill_directive(agent_id: str | None) -> str:
def build_seat_skill_directive(agent_id: str | None, *, chain_enabled: bool = False) -> str:
"""为某席位 agent 构建「必须先读 SKILL.md 再用技能」的本轮任务追加指令。
返回 ``""`` 表示无需追加(开关关闭 / 无 agent_id / 该席位没有配置技能 / 解析失败),
调用方原样使用任务文本,零副作用。
是否注入由 **agent 开关(config.yaml)OR chain 开关(``chain_enabled``)** 决定,
默认两者皆关 → 不注入。返回 ``""`` 表示无需追加(总开关熔断 / 两开关都关 / 无 agent_id /
该席位没有配置技能 / 解析失败),调用方原样使用任务文本,零副作用。
"""
if not _directive_enabled() or not agent_id:
if not _master_enabled() or not agent_id:
return ""
try:
from deerflow.config.agents_config import load_agent_config
cfg = load_agent_config(agent_id)
# agent 级开关 OR chain 级开关(默认关)。任一打开才继续,否则零副作用返回。
agent_enabled = bool(cfg and cfg.seat_skill_directive)
if not (agent_enabled or chain_enabled):
return ""
configured = [name for name in (cfg.skills or []) if name] if cfg else []
if not configured:
return ""
@@ -87,9 +106,12 @@ def build_seat_skill_directive(agent_id: str | None) -> str:
return ""
def append_seat_skill_directive(agent_id: str | None, task: str) -> str:
"""把技能识别指令追加到席位本轮任务文本后(无指令时原样返回)。"""
directive = build_seat_skill_directive(agent_id)
def append_seat_skill_directive(agent_id: str | None, task: str, *, chain_enabled: bool = False) -> str:
"""把技能识别指令追加到席位本轮任务文本后(无指令时原样返回)。
``chain_enabled``:本次会商所选业务链条是否打开「技能识别强化」。与 agent 级开关取 OR。
"""
directive = build_seat_skill_directive(agent_id, chain_enabled=chain_enabled)
if not directive:
return task
return f"{task}\n\n---\n\n{directive}"
@@ -64,6 +64,11 @@ class AgentResponse(BaseModel):
description="入库模板名 (external knowledge-ingest templateName). When set, markdown artifacts produced in this agent's chats can be published to the external knowledge-ingest API.",
)
seat_skill_directive: bool | None = Field(
default=None,
description="会商席位「技能识别强化」开关 (config.yaml). True = 本智能体作为会商席位时强化技能识别 (与业务链条同名开关取 OR). None/False = 关 (默认).",
)
tags: list[dict] = Field(default_factory=list, description="Tags assigned to this agent")
featured_order: int | None = Field(
@@ -115,6 +120,7 @@ class AgentCreateRequest(BaseModel):
tool_groups: list[str] | None = Field(default=None, description="Optional tool group whitelist persisted to config.yaml")
common_questions: list[CommonQuestion] | None = Field(default=None, description="Recommended questions (常用问题) persisted to config.yaml. Each item is {title, prompt}; legacy strings are coerced.")
template_name: str | None = Field(default=None, description="入库模板名 persisted to config.yaml. Enables publishing markdown artifacts to the external knowledge-ingest API.")
seat_skill_directive: bool | None = Field(default=None, description="会商席位「技能识别强化」开关 persisted to config.yaml. None/False = off (默认).")
class AgentUpdateRequest(BaseModel):
@@ -131,6 +137,7 @@ class AgentUpdateRequest(BaseModel):
tool_groups: list[str] | None = Field(default=None, description="Updated tool group whitelist (config.yaml)")
common_questions: list[CommonQuestion] | None = Field(default=None, description="Updated recommended questions (常用问题, config.yaml). Each item is {title, prompt}; legacy strings are coerced.")
template_name: str | None = Field(default=None, description="Updated 入库模板名 (config.yaml). Empty string clears it.")
seat_skill_directive: bool | None = Field(default=None, description="Updated 会商席位「技能识别强化」开关 (config.yaml).")
class AgentSelectorResponse(BaseModel):
@@ -236,6 +243,7 @@ def _write_agent_config(
tool_groups: list[str] | None = None,
common_questions: list[CommonQuestion] | None = None,
template_name: str | None = None,
seat_skill_directive: bool | None = None,
) -> None:
agent_dir = get_paths().agent_dir(agent_id)
agent_dir.mkdir(parents=True, exist_ok=True)
@@ -264,13 +272,16 @@ def _write_agent_config(
]
if template_name is not None and template_name.strip():
config_data["template_name"] = template_name.strip()
# 会商席位技能识别强化开关:仅当显式设置时写入(None = 不写, 由 AgentConfig 默认为关)。
if seat_skill_directive is not None:
config_data["seat_skill_directive"] = bool(seat_skill_directive)
with open(config_file, "w", encoding="utf-8") as f:
yaml.dump(config_data, f, default_flow_style=False, allow_unicode=True)
def _load_config_extras(agent_id: str) -> dict[str, Any]:
"""Read skills/model/tool_groups/common_questions from the agent's config.yaml, tolerant of missing files."""
empty = {"skills": None, "model": None, "tool_groups": None, "common_questions": None, "template_name": None}
empty = {"skills": None, "model": None, "tool_groups": None, "common_questions": None, "template_name": None, "seat_skill_directive": None}
try:
cfg = load_agent_config(agent_id)
except (FileNotFoundError, ValueError):
@@ -283,6 +294,7 @@ def _load_config_extras(agent_id: str) -> dict[str, Any]:
"tool_groups": cfg.tool_groups,
"common_questions": cfg.common_questions,
"template_name": cfg.template_name,
"seat_skill_directive": cfg.seat_skill_directive,
}
@@ -443,7 +455,7 @@ def _record_to_response(
can happen lazily when the agent is actually opened.
"""
soul: str | None = None
extras: dict[str, Any] = {"skills": None, "model": None, "tool_groups": None, "common_questions": None, "template_name": None}
extras: dict[str, Any] = {"skills": None, "model": None, "tool_groups": None, "common_questions": None, "template_name": None, "seat_skill_directive": None}
if include_extras:
_ensure_agent_dir_from_record(record)
extras = _load_config_extras(record["id"])
@@ -467,6 +479,7 @@ def _record_to_response(
skills=extras["skills"],
common_questions=extras.get("common_questions"),
template_name=extras.get("template_name"),
seat_skill_directive=extras.get("seat_skill_directive"),
featured_order=record.get("featured_order"),
)
@@ -793,6 +806,7 @@ async def create_agent_endpoint(request: AgentCreateRequest, http_request: Reque
tool_groups=request.tool_groups,
common_questions=request.common_questions,
template_name=request.template_name,
seat_skill_directive=request.seat_skill_directive,
)
_write_agent_soul(agent_id, request.soul)
@@ -873,8 +887,9 @@ async def update_agent(agent_id: str, request: AgentUpdateRequest, http_request:
next_tool_groups = request.tool_groups if "tool_groups" in fields_set else existing_extras["tool_groups"]
next_common_questions = request.common_questions if "common_questions" in fields_set else existing_extras["common_questions"]
next_template_name = request.template_name if "template_name" in fields_set else existing_extras["template_name"]
next_seat_skill_directive = request.seat_skill_directive if "seat_skill_directive" in fields_set else existing_extras["seat_skill_directive"]
config_dirty = bool(update_data) or any(key in fields_set for key in ("skills", "model", "tool_groups", "common_questions", "template_name"))
config_dirty = bool(update_data) or any(key in fields_set for key in ("skills", "model", "tool_groups", "common_questions", "template_name", "seat_skill_directive"))
if config_dirty:
_write_agent_config(
agent_id,
@@ -885,6 +900,7 @@ async def update_agent(agent_id: str, request: AgentUpdateRequest, http_request:
tool_groups=next_tool_groups,
common_questions=next_common_questions,
template_name=next_template_name,
seat_skill_directive=next_seat_skill_directive,
)
# Keep the listing cache in lock-step with config.yaml, but only
# when the cached fields were actually touched — a partial update
@@ -1737,6 +1737,7 @@ async def _special_run(
thinking_enabled: bool | None = None,
reasoning_effort: str | None = None,
subagent_enabled: bool = False,
seat_skill_directive: bool = False,
) -> AsyncIterator[str | dict[str, Any]]:
"""子智能体(special)单轮:执行总控下发的 task 文本,交付 content。
@@ -1789,10 +1790,12 @@ async def _special_run(
# 弱模型补偿:仅对真正的研讨席位(seat / seat_parallel;report/summary/dashboard/ingest
# 等 Step3 单例不在此列)把其配置技能(名称 + SKILL.md 路径 + 必须先 read_file 再按其
# 流程取信息的硬指令)显式拼进本轮任务,避免弱模型无视系统提示里的技能块(见
# roundtable_seat_skills 模块说明)。无配置技能时原样返回。
# roundtable_seat_skills 模块说明)。默认关闭:是否注入 = 该席位 agent 的开关
# (config.yaml)OR 本次会商所选业务链条的开关(seat_skill_directive 透传)。无配置
# 技能 / 两开关都关时原样返回。
seat_message = new_message
if role in ("seat", "seat_parallel"):
seat_message = append_seat_skill_directive(agent_name, new_message)
seat_message = append_seat_skill_directive(agent_name, new_message, chain_enabled=seat_skill_directive)
body = _run_payload(
agent_name=agent_name,
model_name=model_name,
@@ -1957,6 +1960,9 @@ async def run_multi_agent_stream(request: Request, payload: dict = Body(...)) ->
_re_raw = payload.get("reasoning_effort")
seat_reasoning_effort: str | None = _re_raw if isinstance(_re_raw, str) and _re_raw else None
seat_subagent_enabled: bool = bool(payload.get("subagent_enabled") or False)
# 业务链条级「技能识别强化」开关(仅 special 席位消费):前端读所选业务链条的同名配置
# 后逐席位透传。与该席位 agent 自身的 config.yaml 开关取 OR;缺省 False(默认关)。
seat_skill_directive: bool = bool(payload.get("seat_skill_directive") or False)
if not agent_name or agent_name not in agent_threads:
raise HTTPException(
@@ -2007,6 +2013,7 @@ async def run_multi_agent_stream(request: Request, payload: dict = Body(...)) ->
thinking_enabled=seat_thinking_enabled,
reasoning_effort=seat_reasoning_effort,
subagent_enabled=seat_subagent_enabled,
seat_skill_directive=seat_skill_directive,
)
async for chunk in gen:
if isinstance(chunk, dict):
@@ -48,6 +48,11 @@ class ChainSeatSchema(BaseModel):
# 数组以 JSON 透传持久化(见 persistence/roundtable_chains),故新增此字段无需迁移;
# 仅需在此 schema 显式声明,否则 model_dump() 会把未知字段剥掉。
model: str | None = Field(default=None, max_length=128)
# 该席位「会商技能识别强化」开关(弱模型补偿)。True = 该席位作为会商子智能体执行时,
# 把其配置的技能(名称 + SKILL.md 路径 + 必须先 read_file 的硬指令)拼进本轮任务。
# 与该席位 agent 自身 config.yaml 的同名开关取 OR。默认 False(关)。整条 seats 数组以
# JSON 透传持久化,故新增此字段无需迁移;仅需在此 schema 显式声明,否则 model_dump() 会剥掉它。
seat_skill_directive: bool = False
class ChainResponse(BaseModel):
@@ -77,6 +77,10 @@ class StartJobRequest(BaseModel):
thinking_enabled: bool | None = None
reasoning_effort: str | None = None
subagent_enabled: bool = False
# 每席位「技能识别强化」开关映射 {agent_id: bool}(前端读所选业务链条里**每个席位**的
# 同名配置后下发)。后台席位 run 据此(与各席位 agent 自身 config.yaml 的开关取 OR)决定
# 是否注入。缺省空 dict(全部按 False/关,仅各 agent 自身开关生效)。
seat_skill_directives: dict[str, bool] = Field(default_factory=dict)
class ResumeJobRequest(BaseModel):
@@ -235,6 +239,7 @@ async def start_job(request: Request, body: StartJobRequest) -> RoundtableJobRes
seat_thinking_enabled=body.thinking_enabled,
seat_reasoning_effort=body.reasoning_effort,
seat_subagent_enabled=body.subagent_enabled,
seat_skill_directives=dict(body.seat_skill_directives or {}),
)
)
return _to_response(row)
@@ -103,6 +103,12 @@ class AgentConfig(BaseModel):
# 入库模板名(外部知识入库接口的 templateName)。配置后该智能体会话中的
# Markdown 产物可一键发布到外部入库接口;None/空 = 不支持入库。
template_name: str | None = None
# 会商席位「技能识别强化」开关(弱模型补偿)。作为会商席位运行时,是否把本智能体
# 配置的技能(名称 + SKILL.md 路径 + 必须先 read_file 再按其流程取信息的硬指令)显式
# 拼进本轮任务,避免弱模型无视系统提示里的技能块。None/False = 关(默认);True = 开。
# 与「业务链条配置」里的同名开关取 OR:任一打开即对该席位生效。见
# ``app/gateway/roundtable_seat_skills.py``。
seat_skill_directive: bool | None = None
def load_agent_config(agent_id: str | None) -> AgentConfig | None:
@@ -177,12 +177,12 @@ def test_load_config_extras_returns_values_written_to_yaml(isolated_paths) -> No
)
extras = _load_config_extras(agent_id)
assert extras == {"skills": ["s1"], "model": None, "tool_groups": None, "common_questions": None}
assert extras == {"skills": ["s1"], "model": None, "tool_groups": None, "common_questions": None, "template_name": None, "seat_skill_directive": None}
def test_load_config_extras_missing_directory_returns_none_triple(isolated_paths) -> None:
extras = _load_config_extras("does-not-exist")
assert extras == {"skills": None, "model": None, "tool_groups": None, "common_questions": None}
assert extras == {"skills": None, "model": None, "tool_groups": None, "common_questions": None, "template_name": None, "seat_skill_directive": None}
def test_write_agent_config_persists_common_questions(isolated_paths) -> None: