Back to Details

foundry-research

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Research Skill

研究Skill

Overview

概述

  • Purpose: AI-powered research with multiple reasoning strategies
  • Scope: Five workflows, persistent thread and session management
  • Entry: 
    Skill(foundry:foundry-research)
    or user invocation
  • 用途：采用多种推理策略的AI驱动型研究工具
  • 范围：五种工作流，持久化线程与会话管理
  • 入口：
    Skill(foundry:foundry-research)
    或用户调用

Flow

流程

[x?]
=decision 
(GATE)
=user approval 
→
=sequence
- **Entry** → [route?]
  - [explicit?] → Dispatch → Execute → Persist thread → Response + thread_id
  - [thread-id?] → Resume → Dispatch → Execute → Persist thread → Response + thread_id
  - [research-id?] → SessionMgmt
  - [sessions?] → ListSessions
  - [no args?] → (GATE: choose workflow) → AutoRoute → Dispatch → Execute → Persist thread → Response + thread_id
  - [deep?] → Start → Poll → Report (background execution)
CRITICAL for 
deep
workflow: Read references/deep-research-workflow.md before execution. Contains required polling strategy and MCP parameters.
[x?]
=决策 
(GATE)
=用户审批 
→
=顺序执行
- **Entry** → [route?]
  - [explicit?] → Dispatch → Execute → Persist thread → Response + thread_id
  - [thread-id?] → Resume → Dispatch → Execute → Persist thread → Response + thread_id
  - [research-id?] → SessionMgmt
  - [sessions?] → ListSessions
  - [no args?] → (GATE: choose workflow) → AutoRoute → Dispatch → Execute → Persist thread → Response + thread_id
  - [deep?] → Start → Poll → Report (background execution)
deep
工作流注意事项：执行前请阅读references/deep-research-workflow.md，其中包含必需的轮询策略和MCP参数。

Deep Research Polling Protocol (MANDATORY)

Deep Research轮询协议（强制要求）

BEFORE EVERY status check, you MUST:
  1. Generate user-facing text about current progress
  2. Do NOT call deep-research-status immediately after another MCP tool
  3. Track your check count (max 5)
Check #What to Say
1"Research underway. Currently in {phase} phase..."
2-3"Progress: {queries_completed}/{total} queries done..."
4"Still working ({elapsed} minutes)..."
5Stop polling, use AskUserQuestion with options
Stall Detection: Only after 
elapsed_ms > 300000
AND no progress change.
NO INDEPENDENT RESEARCH: While deep research is running, do NOT use WebSearch, WebFetch, or other research tools. The deep research workflow handles all web gathering - doing your own searches is redundant and wastes resources.
每次状态检查前，必须执行以下操作：
  1. 生成面向用户的当前进度文本
  2. 请勿在调用其他MCP工具后立即调用deep-research-status
  3. 跟踪检查次数（最多5次）
检查次数提示话术
1"研究进行中，当前处于{phase}阶段..."
2-3"进度：已完成{queries_completed}/{total}个查询..."
4"仍在处理中（已耗时{elapsed}分钟）..."
5停止轮询，使用AskUserQuestion并提供选项
停滞检测：仅当
elapsed_ms > 300000
且无进度变化时触发。
禁止独立研究：当deep research运行时，请勿使用WebSearch、WebFetch或其他研究工具。deep research工作流会处理所有网络信息收集操作——自行搜索属于冗余操作，会浪费资源。

MCP Tooling

MCP工具集

RouterActions
research
chat
, 
consensus
, 
thinkdeep
, 
ideate
, 
deep-research
, 
deep-research-status
, 
deep-research-report
, 
deep-research-list
, 
deep-research-delete
, 
thread-list
, 
thread-get
, 
thread-delete
, 
node-execute
, 
node-record
, 
node-status
, 
node-findings
路由操作
research
chat
, 
consensus
, 
thinkdeep
, 
ideate
, 
deep-research
, 
deep-research-status
, 
deep-research-report
, 
deep-research-list
, 
deep-research-delete
, 
thread-list
, 
thread-get
, 
thread-delete
, 
node-execute
, 
node-record
, 
node-status
, 
node-findings

MCP Contract

MCP协议

ActionRequiredOptionalErrors
chat
prompt
thread_id
, 
provider_id
THREAD_NOT_FOUND
consensus
prompt
providers
, 
strategy
NO_MODELS_AVAILABLE
thinkdeep
prompt
thread_id
, 
depth
MAX_DEPTH_EXCEEDED
ideate
prompt
thread_id
, 
phase
INVALID_PHASE
deep-research
query
max_iterations
, 
max_sub_queries
, 
follow_links
RESEARCH_TIMEOUT
deep-research-status
research_id
-
RESEARCH_NOT_FOUND
deep-research-report
research_id
-
RESEARCH_NOT_FOUND
deep-research-list
-
limit
, 
completed_only
-
deep-research-delete
research_id
-
RESEARCH_NOT_FOUND
thread-*
thread_id
limit
THREAD_NOT_FOUND
node-status
spec_id
, 
research_node_id
-
NODE_NOT_FOUND
node-execute
spec_id
, 
research_node_id
prompt
NODE_NOT_FOUND
, 
INVALID_TYPE
node-record
spec_id
, 
research_node_id
, 
result
summary
, 
key_insights
, 
recommendations
, 
confidence
NODE_NOT_FOUND
node-findings
spec_id
, 
research_node_id
-
NODE_NOT_FOUND
, 
NO_FINDINGS
操作必填项可选项错误类型
chat
prompt
thread_id
, 
provider_id
THREAD_NOT_FOUND
consensus
prompt
providers
, 
strategy
NO_MODELS_AVAILABLE
thinkdeep
prompt
thread_id
, 
depth
MAX_DEPTH_EXCEEDED
ideate
prompt
thread_id
, 
phase
INVALID_PHASE
deep-research
query
max_iterations
, 
max_sub_queries
, 
follow_links
RESEARCH_TIMEOUT
deep-research-status
research_id
-
RESEARCH_NOT_FOUND
deep-research-report
research_id
-
RESEARCH_NOT_FOUND
deep-research-list
-
limit
, 
completed_only
-
deep-research-delete
research_id
-
RESEARCH_NOT_FOUND
thread-*
thread_id
limit
THREAD_NOT_FOUND
node-status
spec_id
, 
research_node_id
-
NODE_NOT_FOUND
node-execute
spec_id
, 
research_node_id
prompt
NODE_NOT_FOUND
, 
INVALID_TYPE
node-record
spec_id
, 
research_node_id
, 
result
summary
, 
key_insights
, 
recommendations
, 
confidence
NODE_NOT_FOUND
node-findings
spec_id
, 
research_node_id
-
NODE_NOT_FOUND
, 
NO_FINDINGS

Workflow Selection

工作流选择

SignalWorkflow
Follow-up, iteration
chat
Multiple perspectives
consensus
Complex problem
thinkdeep
Brainstorming
ideate
Comprehensive research, multiple sources
deep
触发信号工作流
跟进、迭代
chat
多视角分析
consensus
复杂问题
thinkdeep
头脑风暴
ideate
全面研究、多数据源
deep

User Gates

用户交互节点

  • No args: workflow selection
  • Ambiguous: clarify before auto-route
  • Consensus: strategy selection
  • Ideate: phase transition
  • Deep: progress updates during background execution
  • 无参数：选择工作流
  • 指令模糊：自动路由前先澄清
  • Consensus：策略选择
  • Ideate：阶段转换
  • Deep：后台执行时的进度更新

Output Formats

输出格式

WorkflowResponse
chat
{response, thread_id, model}
consensus
{responses[], synthesis, strategy}
thinkdeep
{findings[], confidence, thread_id}
ideate
{ideas[], phase, selected[]}
deep
{research_id, status, report{summary, findings[], sources[]}}
工作流响应内容
chat
{response, thread_id, model}
consensus
{responses[], synthesis, strategy}
thinkdeep
{findings[], confidence, thread_id}
ideate
{ideas[], phase, selected[]}
deep
{research_id, status, report{summary, findings[], sources[]}}

References

参考文档

  • Chat | Consensus | Deep
  • ThinkDeep | Ideate
  • Sessions | Auto-Route
  • Chat | Consensus | Deep
  • ThinkDeep | Ideate
  • Sessions | Auto-Route