deep-research-agent

Name: deep-research-agent
Author: shobcoder

578stars0forks

How to install deep-research-agent?

shobcoder/shob/deep-research-agent

$npx -y skills add shobcoder/shob --skill deep-research-agent

Installs into the current project.

Summary

Comprehensive research agent for in-depth investigation. Use when users ask for deep research, comprehensive analysis, market research, academic surveys, competitive analysis, technology trends, or any topic requiring 100+ source verification. Triggers on requests like "investiga

Files · 1

View on GitHub

SKILL.md

1	# DeepResearch Agent
2
3	Autonomous multi-phase research agent that decomposes queries, gathers information from diverse sources, verifies facts, and synthesizes structured reports with 100+ source citations.
4
5	## Core Workflow
6
7	### Phase 1: Query Decomposition & Planning
8
9	Input: User's research query (natural language)
10
11	Process:
12	1. Analyze the query intent
13	- Identify the primary research objective
14	- Determine required expertise domains (History/Technology/Market/Challenges/Regulations etc.)
15	- Assess depth requirements (surface-level vs comprehensive)
16
17	2. Generate multi-dimensional search queries
18	- Historical context queries (when applicable)
19	- Technical specification queries
20	- Market/industry trend queries
21	- Challenge/pain point queries
22	- Regulatory/compliance queries (if applicable)
23	- Future outlook/prediction queries
24
25	3. Build investigation roadmap
26	- Define search priority order
27	- Identify cross-cutting themes
28	- Plan for iterative deep-diving
29	- Set minimum source targets per topic area
30
31	Output: `research_plan` object containing:
32	```json
33	{
34	"primary_topic": "string",
35	"sub_topics": ["string"],
36	"search_queries": [{"query": "string", "domain": "string", "priority": 1}],
37	"target_sources": 100,
38	"timeline_phases": ["phase1", "phase2", "phase3"]
39	}
40	```
41
42	### Phase 2: Autonomous Information Gathering
43
44	Tools Used: `batch_web_search`, `extract_content_from_websites`
45
46	Process:
47	1. Initial breadth search
48	- Execute parallel searches across all primary query dimensions
49	- Gather minimum 20-30 URLs per major topic area
50	- Prioritize authoritative sources (official docs, academic, established media)
51
52	2. Source classification
53	- Categorize by source type: News, Academic Papers, Whitepapers, Technical Documentation, Forums, Blogs
54	- Assess domain authority and reliability
55	- Flag sources requiring deeper analysis
56
57	3. Iterative deep-diving
58	- Extract key terms and concepts from initial results
59	- Generate follow-up queries using discovered terminology
60	- Expand search to related topics and subtopics
61	- Loop until saturation (no new significant information)
62
63	4. Diverse source coverage
64	- Ensure geographic diversity (US/EU/Asia when relevant)
65	- Cover multiple stakeholder perspectives
66	- Include both primary and secondary sources
67
68	Target: Minimum 100 unique, verified sources
69
70	### Phase 3: Content Reading & Reasoning
71
72	Tools Used: `extract_content_from_websites`, `extract_pdfs_key_info`
73
74	Process:
75	1. Content extraction
76	- Access each promising URL
77	- Extract structured information: facts, statistics, quotes, dates, claims
78	- Parse PDF documents for detailed data
79
80	2. Relevance assessment
81	- Score content against research objectives (1-5 scale)
82	- Filter out low-relevance or duplicate content
83	- Prioritize high-value sources for deep analysis
84
85	3. Information extraction matrix
86	```
87	For each source:
88	- Source metadata (title, author, date, URL)
89	- Key findings (bullet points)
90	- Supporting evidence (quotes, statistics)
91	- Contradicting information (if any)
92	- Confidence level (high/medium/low)
93	```
94
95	4. Pattern recognition
96	- Identify consensus areas (multiple sources agree)
97	- Detect controversy or debate points
98	- Find knowledge gaps or underreported aspects
99
100	### Phase 4: Verification & Gap Filling
101
102	Process:
103	1. Cross-verification protocol
104	- Check consistency across independent sources
105	- Verify statistics with multiple citations
106	- Confirm quotes with original context
107
108	2. Contradiction resolution
109	- Document conflicting information
110	- Assess source credibility differences
111	- Note the nature of disagreement (factual vs interpretive)
112	- Present multiple perspectives when resolution impossible
113
114	3. Gap identification
115	- Compare gathered information against research plan
116	- Identify missing perspectives or outdated information
117	- Flag areas needing additional primary source verification
118
119	4. Iteration loop (if gaps identified)
120	- Return to Phase 2 with targeted queries
121	- Focus on specific missing elements
122	- Repeat until research objectives are satisfied
123
124	### Phase 5: Structured Report Synthesis
125
126	Output Format: Comprehensive research report
127
128	Structure:
129	```
130	# [Research Title]
131
132	## Executive Summary
133	[2-3 paragraph overview of key findings]
134
135	## 1. Background and Purpose
136	[Context and research motivation]
137
138	## 2. Key Findings
139
140	### 2.1 [Topic Area 1]
141	#### Facts and Data
142	#### Analysis and In

People who liked this also viewed

6 picks

Type

deep-research-agent

578stars0forks

View on GitHub

How to install deep-research-agent?

shobcoder/shob/deep-research-agent

$npx -y skills add shobcoder/shob --skill deep-research-agent

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# DeepResearch Agent
2
3	Autonomous multi-phase research agent that decomposes queries, gathers information from diverse sources, verifies facts, and synthesizes structured reports with 100+ source citations.
4
5	## Core Workflow
6
7	### Phase 1: Query Decomposition & Planning
8
9	Input: User's research query (natural language)
10
11	Process:
12	1. Analyze the query intent
13	- Identify the primary research objective
14	- Determine required expertise domains (History/Technology/Market/Challenges/Regulations etc.)
15	- Assess depth requirements (surface-level vs comprehensive)
16
17	2. Generate multi-dimensional search queries
18	- Historical context queries (when applicable)
19	- Technical specification queries
20	- Market/industry trend queries
21	- Challenge/pain point queries
22	- Regulatory/compliance queries (if applicable)
23	- Future outlook/prediction queries
24
25	3. Build investigation roadmap
26	- Define search priority order
27	- Identify cross-cutting themes
28	- Plan for iterative deep-diving
29	- Set minimum source targets per topic area
30
31	Output: `research_plan` object containing:
32	```json
33	{
34	"primary_topic": "string",
35	"sub_topics": ["string"],
36	"search_queries": [{"query": "string", "domain": "string", "priority": 1}],
37	"target_sources": 100,
38	"timeline_phases": ["phase1", "phase2", "phase3"]
39	}
40	```
41
42	### Phase 2: Autonomous Information Gathering
43
44	Tools Used: `batch_web_search`, `extract_content_from_websites`
45
46	Process:
47	1. Initial breadth search
48	- Execute parallel searches across all primary query dimensions
49	- Gather minimum 20-30 URLs per major topic area
50	- Prioritize authoritative sources (official docs, academic, established media)
51
52	2. Source classification
53	- Categorize by source type: News, Academic Papers, Whitepapers, Technical Documentation, Forums, Blogs
54	- Assess domain authority and reliability
55	- Flag sources requiring deeper analysis
56
57	3. Iterative deep-diving
58	- Extract key terms and concepts from initial results
59	- Generate follow-up queries using discovered terminology
60	- Expand search to related topics and subtopics
61	- Loop until saturation (no new significant information)
62
63	4. Diverse source coverage
64	- Ensure geographic diversity (US/EU/Asia when relevant)
65	- Cover multiple stakeholder perspectives
66	- Include both primary and secondary sources
67
68	Target: Minimum 100 unique, verified sources
69
70	### Phase 3: Content Reading & Reasoning
71
72	Tools Used: `extract_content_from_websites`, `extract_pdfs_key_info`
73
74	Process:
75	1. Content extraction
76	- Access each promising URL
77	- Extract structured information: facts, statistics, quotes, dates, claims
78	- Parse PDF documents for detailed data
79
80	2. Relevance assessment
81	- Score content against research objectives (1-5 scale)
82	- Filter out low-relevance or duplicate content
83	- Prioritize high-value sources for deep analysis
84
85	3. Information extraction matrix
86	```
87	For each source:
88	- Source metadata (title, author, date, URL)
89	- Key findings (bullet points)
90	- Supporting evidence (quotes, statistics)
91	- Contradicting information (if any)
92	- Confidence level (high/medium/low)
93	```
94
95	4. Pattern recognition
96	- Identify consensus areas (multiple sources agree)
97	- Detect controversy or debate points
98	- Find knowledge gaps or underreported aspects
99
100	### Phase 4: Verification & Gap Filling
101
102	Process:
103	1. Cross-verification protocol
104	- Check consistency across independent sources
105	- Verify statistics with multiple citations
106	- Confirm quotes with original context
107
108	2. Contradiction resolution
109	- Document conflicting information
110	- Assess source credibility differences
111	- Note the nature of disagreement (factual vs interpretive)
112	- Present multiple perspectives when resolution impossible
113
114	3. Gap identification
115	- Compare gathered information against research plan
116	- Identify missing perspectives or outdated information
117	- Flag areas needing additional primary source verification
118
119	4. Iteration loop (if gaps identified)
120	- Return to Phase 2 with targeted queries
121	- Focus on specific missing elements
122	- Repeat until research objectives are satisfied
123
124	### Phase 5: Structured Report Synthesis
125
126	Output Format: Comprehensive research report
127
128	Structure:
129	```
130	# [Research Title]
131
132	## Executive Summary
133	[2-3 paragraph overview of key findings]
134
135	## 1. Background and Purpose
136	[Context and research motivation]
137
138	## 2. Key Findings
139
140	### 2.1 [Topic Area 1]
141	#### Facts and Data
142	#### Analysis and In

People who liked this also viewed

6 picks

Type