deep-research

Deep Research

Create high-fidelity research reports with strict format control, evidence mapping, source governance, and multi-pass synthesis.

Architecture: Lead Agent + Subagents

Lead Agent (coordinator — minimizes raw search context)
  |
  P0: Environment + source policy setup
  |
  P1: Question and claim map (decision questions, evidence routes, stop rules)
  |
  Dispatch ──→ Subagent A ──→ writes task-a.md ──┐
           ──→ Subagent B ──→ writes task-b.md ──┤ (parallel)
           ──→ Subagent C ──→ writes task-c.md ──┘
  |                                               |

Context discipline: Keep raw search-result noise in task workspaces. Pass evidence packets to the lead agent, including locators and short source excerpts. Notes are routing aids, not authorities: the lead agent must open the original source for every load-bearing claim, conflicting claim, and exact figure/date/quotation used in the report.

Mode Selection

Determine the research mode before starting:

• Enterprise Research Mode: Question-led company research with optional analysis frameworks selected only when they help answer the decision
• General Research Mode: Standard P0-P7 research pipeline with source governance
• Depth Selection: Choose from the number and consequence of unresolved questions, not prompt length, task count, or a target word count

Source Governance (V6)

Source Accessibility Classification

Classify every source by accessibility:

First-party boundary: User-authorized records may establish what the organization did, agreed, paid, delivered, or observed. They do not count as independent external validation of market standing, customer sentiment, regulatory compliance, or third-party claims. Never relabel an internal record as an external finding.

✅ EXCLUSIVE INFORMATION ADVANTAGE: You SHOULD:

• Use user's Crunchbase Pro to research competitors
• Use user's proprietary databases for market research
• Use user's private APIs for investment analysis
• Leverage any exclusive source user provides for third-party research

Source Type Labels

Every source MUST also be tagged with:

Coverage diagnostics: Track source counts, domains, source-type mix, and concentration to reveal thin coverage. Never pass or fail research from these totals alone. Gate on whether each decision question and load-bearing claim has fit-for-purpose evidence, whether counter-evidence was sought, and whether remaining unknowns are explicit.

AS_OF Date Policy

Set AS_OF date explicitly at P0. For all time-sensitive claims:

• Include source publication date with every citation
• Downgrade confidence if source is older than relevant horizon
• Define a freshness horizon per claim class and flag material outside it; a universal age cutoff is only a diagnostic

P0: Environment & Policy Setup

Check capabilities before starting:

Set policy variables:

• AS_OF: Today's date (YYYY-MM-DD) - mandatory for timed topics
• MODE: Standard (default) or Lightweight, justified by the question map
• SOURCE_TYPE_POLICY: Enforce official/academic/secondary/journalism/community/other labels
• COUNTER_REVIEW_PLAN: What evidence would overturn each provisional conclusion

Report: [P0 complete] Subagent: {yes/no}. Mode: {standard/lightweight}. AS_OF: {YYYY-MM-DD}.

When researching a specific company, use the specialized workflow to route evidence by question. Treat the six dimensions as a coverage map, not a mandatory report outline.

Enterprise Workflow Overview

P1: Research Task Board

Decompose the assignment into decision questions. Create tasks only where separate evidence routes or expertise make the work clearer.

Each task assignment includes:

• Expert Role: Specialist persona (e.g., "Policy Historian", "Ecosystem Mapper")
• Objective: One-sentence investigation goal
• Queries: 2-3 pre-planned search queries
• Depth: DEEP (fetch 2-3 full articles) or SCAN (snippets sufficient)
• Output: Path to research notes file
• Parallel Group: Group A (independent) or Group B (depends on Group A)
• Decision Question: The exact question this task helps answer
• Load-Bearing Claims: Provisional claims that would change the conclusion
• Disconfirming Evidence: What would weaken or overturn each claim
• Evidence Route: Which source owners or record systems can actually observe the fact
• Stop Rule: What counts as answered, contradicted, or still unknown

Task Decomposition Rules

1. Each task covers one coherent sub-topic a specialist would own
2. Group A tasks must be logically independent; source independence is assessed by underlying evidence, ownership, and incentive, not domain count
3. Max 3 tasks per parallel group (concurrency limit)
4. Every task must flag time-sensitive claims, counter-evidence sought, and expected citation aging risk

Enterprise Research Integration

When in Enterprise Research Mode, map questions to the relevant dimensions rather than creating all six tasks automatically:

• Task A: Company fundamentals (entity, founding, funding, ownership)
• Task B: Business & products (segments, products, revenue structure)
• Task C: Competitive position (industry rank, competitors, barriers)
• Task D: Financial & operations (3-year financials, efficiency metrics)
• Task E: Recent developments (6-month events, strategic signals)
• Task F: Authorized first-party records (when they can answer a business fact; never counted as external corroboration)

Report: [P1 complete] {N} tasks in {M} groups. Dispatching Group A.

Enterprise Research Mode (Specialized Pipeline)

When researching a specific company, route each decision question through the relevant enterprise dimensions. Use the dimensions to find missing evidence paths; do not run all six or add quantified frameworks by default.

E1: Intake

Same as P0/P1 above, plus:

• Confirm the exact legal entity being researched (parent vs subsidiary)
• Select research depth from the decision questions, evidence difficulty, and requested output; page counts are planning diagnostics only
• Identify any specific comparison targets (benchmark companies)

P2: Dispatch + Investigate

Subagents execute tasks using references/subagent_prompt.md and output evidence packets in references/research_notes_format.md.

With Subagents (Claude Code / Cowork / DeerFlow)

1. Dispatch Group A tasks in parallel (max 3 concurrent)
2. Each subagent searches, fetches, and tags source types
3. Every source line includes Source-Type and As Of
4. Wait for Group A completion
5. Dispatch Group B (can read Group A notes)

Subagent Output Requirements

Each task-{id}.md must contain:

• Question status: answered / contradicted / unknown, with the stopping evidence
• Sources section: stable locators from actual retrievals with source type, accessibility, date, and source-family identity
• Claim-evidence table: claim, evidence excerpt/locator, scope, confidence, and whether the original was opened
• Counter-evidence and unknowns: what was sought, what was found, and what remains unresolved

Without Subagents (Degraded Mode)

Lead agent executes tasks sequentially, acting as each specialist. Preserve raw search noise outside the final evidence packet; retain a query log when reproducibility matters.

Enterprise Research: Six-Dimension Collection

Follow references/enterprise_research_methodology.md for:

• Detailed collection workflow per dimension (query strategies, data fields, validation)
• Data source priority matrix (P0-P3 ranking)
• Claim-specific corroboration and conflict-handling rules

Key principles:

• Evidence-driven: every conclusion must trace to a citable source
• Corroboration: a second source adds weight only when it is independent of the same underlying disclosure or dataset
• Restrained judgment: mark speculation explicitly, avoid unsubstantiated claims
• Structured presentation: complex information via tables, lists, hierarchies

Run L1 quality check after completing each dimension (see enterprise_quality_checklist.md).

Status per task: [P2 task-{id} complete] {N} sources, {M} findings. Status all: [P2 complete] {N} tasks done, {M} total sources. Building registry.

E3: Select Analysis Frameworks Only When Useful

Load references/enterprise_analysis_frameworks.md only when the user's decision benefits from a framework. Use SWOT for strategic option framing, a risk matrix for decisions with explicit probability/impact inputs, and scoring only when weights and scales are defensible. Omit the framework rather than fabricate entries or precision.

Run L2 quality check after analysis is complete.

E4: Quality Control

Three-level checks from references/enterprise_quality_checklist.md:

• L1 (Data): Source count, attribution, cross-validation, timeliness
• L2 (Analysis): Decision-question coverage, claim support, counter-evidence, and framework fitness when a framework is used
• L3 (Document): Structure compliance, format consistency, readability, appendices

E5: Draft Using Enterprise Template

Use the 7-chapter enterprise report template from enterprise_quality_checklist.md only when it matches the requested decision. Otherwise adapt the structure around the question map.

1. Company Overview
2. Business & Product Structure
3. Market & Competitive Position
4. Financial & Operations Analysis
5. Risks & Concerns
6. Recent Developments
7. Comprehensive Assessment & Conclusion

Plus appendices: Data Source Index, Glossary, Disclaimer.

E3-E7: Enterprise Analysis, Drafting, and Review

• E3: Structured Analysis — Select a framework from references/enterprise_analysis_frameworks.md only when it improves the decision and its inputs are defensible; otherwise use a claim-evidence table
• E4: Quality Control — Run L1/L2/L3 checks per references/enterprise_quality_checklist.md
• E5: Draft — Use 7-chapter enterprise template
• E6-E7: Multi-Pass Drafting and Review — Same as P4-P7 below

P3: Citation Registry + Source Governance

Lead agent reads all task notes and builds unified registry.

Registry Process

1. Read every task file's claim-evidence table and sources
2. Merge sources; deduplicate URLs but also group multiple publications derived from the same study, filing, press release, dataset, interview, or sponsor as one evidence family
3. Assign sequential [n] numbers by first appearance
4. Tag: source_type, accessibility, as_of date, evidence family, authority, independence limits, and task id
5. Build a claim-coverage matrix: supporting evidence, disconfirming evidence, decisive original checked, and remaining unknown
6. Record excluded sources with reasons. Do not exclude a source merely for failing an arbitrary score; restrict it to claims it can support

Registry Output Format

Critical rule: These [n] are FINAL. P5 may only cite from Approved list. Dropped sources never reappear.

Authorized first-party handling: When researching the user's own organization or assets:

1. Use authorized original records for internal business facts they directly record
2. Label them authorized-first-party and state whose record it is
3. Seek an external source only when the claim requires external corroboration
4. Keep the conclusion explicit: internally established, externally corroborated, conflicted, or externally unknown

Exclusive source handling: When user EXPLICITLY PROVIDES their paid subscriptions or private APIs for third-party research (e.g., "Use my Crunchbase Pro to research competitors"), you SHOULD:

1. Accept it as "exclusive-user-provided" accessibility
2. Use it as competitive advantage
3. Cite it properly in registry
4. If no independent equivalent exists, preserve the source's valid first-party scope and mark the external claim unknown

Report: [P3 complete] {answered}/{total} questions answered. {N} load-bearing claims supported, {M} unresolved. Source totals are diagnostics.

Handling Information Black Box

When researching entities with no public footprint:

What an external researcher would find:

• WHOIS: Privacy protected → No owner info
• Web search: No news, no press releases
• Social media: No company pages
• Business registries: No public API or requires local access
• Result: Complete information black box

Correct response:

DO NOT:

• ❌ Describe an internally established fact as independently externally corroborated
• ❌ Assume the company exists based on domain registration alone
• ❌ Fill missing data with speculation
• ❌ Discard an authorized first-party record when it directly establishes an internal business fact

DO:

• ✅ Clearly state what an external researcher can/cannot verify
• ✅ Report authorized first-party facts as internally established, separately from external visibility
• ✅ Document all failed search attempts
• ✅ Mark claims as [unverified] or omit entirely
• ✅ Narrow or stop when evidence cannot answer the decision question
• ✅ Recommend direct contact for due diligence

P4: Evidence-Mapped Outline

Lead agent reads evidence packets + registry to build the outline, then reopens decisive originals.

1. Identify cross-task patterns
2. Design sections topic-first, not task-order-first
3. Map each section to specific findings with source numbers
4. Flag sections needing counter-review
5. Mark recency-sensitive claims with AS_OF checks
6. Mark every load-bearing claim as supported / contradicted / unknown

Outline format:

P5: Draft from Notes

Write section by section using references/report_template_v6.md, adapting it to the user's format contract.

Rules:

• Every factual claim needs citation [n]
• Numbers/percentages must have source
• Add confidence marker per section: High/Medium/Low with rationale
• Add counter-claim sentence when evidence conflicts
• New sources may enter only through the same registry and verification path
• Use [unverified] for unsupported statements

Anti-hallucination:

• Lead agent never invents URLs; every locator must come from an actual retrieval
• Lead agent never treats notes as proof; reopen the original for load-bearing claims, conflicts, exact numbers/dates, and quotations
• Lead agent never fabricates data; unsupported claims remain unknown or are omitted

Status: [P5 in progress] {N}/{M} sections, ~{words} words.

P6: Counter-Review (Mandatory)

For each major conclusion, perform opposite-view checks. These checks do not automatically require another agent or a team; use independent reviewers only when the user request or applicable workspace instructions call for them:

1. Could the conclusion be wrong?
2. Which high-impact claims depend on one evidence family, even if many domains repeat it?
3. Which claims lack a source that can directly observe the fact?
4. Are stale sources used for time-sensitive claims?
5. Report only evidence-backed issues; zero findings is a valid outcome. State unresolved uncertainty explicitly. Do not invent issues or repeat a completed check solely to reach an issue count.

Using Counter-Review Team (Optional)

For comprehensive parallel review, use the Counter-Review Team:

See references/counter_review_team_guide.md for detailed usage.

Manual Counter-Review (Default)

When a review team has not been selected, perform these evidence checks directly. Obtain individual independent review if the user request or applicable workspace instructions require it:

• Verify every load-bearing claim against its decisive original
• Check whether corroborating sources are genuinely independent and able to observe the claim
• Verify AS_OF dates on time-sensitive claims
• Document opposing interpretations

Output

Include only evidence-backed controversies in the final report. Use numbered entries only when such controversies exist. If none are established, state that explicitly; never fill placeholder disputes to satisfy the template. Report unresolved uncertainty separately, or state that none remains.

The example above applies only when both statements are supported by the completed checks; otherwise list the actual controversies or unresolved questions.

Report: [P6 complete] {N} issues found: {critical} critical, {high} high, {medium} medium.

P7: Verify

Cross-check before finalization:

1. Registry cross-check: List every [n] in report vs approved registry
2. Load-bearing check: Trace every decisive conclusion, exact figure/date/quotation, and disputed fact to the original source
3. Sample low-impact claims: Use spot checks only as a diagnostic; expand to the full affected class when one fails
4. Validate no dropped source resurrected
5. Check evidence-family concentration for key claims

Report: [P7 complete] {N} spot-checks, {M} violations fixed.

Output Requirements

• Match the requested language and tone
• Preserve technical terms in English
• Respect the report spec and formatting rules
• Include a references section or bibliography

Reference Files

Core V6 Pipeline References

General Research References

Enterprise Research References (load when in Enterprise Research Mode)

Anti-Patterns

• Single-pass drafting without parallel complete passes
• Splitting passes by section instead of full report drafts
• Ignoring the format contract or user template
• Claims without citations or evidence table mapping
• Mixing conflicting dates without calling out discrepancies
• Copying external AI output without verification
• Deleting intermediate drafts or raw research outputs
• Lead agent trusting notes as authority — use packets for routing, then reopen decisive originals
• Inventing URLs — only use URLs from actual search results
• Resurrecting dropped sources — dropped in P3 never reappear
• Missing AS_OF for time-sensitive claims — always include source date
• Skipping evidence checks — complete P6, report only supported findings, and allow zero issues when no issue is established.
• FIRST-PARTY OVERCLAIM — authorized records can establish internal business facts but cannot impersonate external validation
• IGNORING EXCLUSIVE SOURCES — when user provides Crunchbase Pro etc. for competitor research, USE IT

Next Step: Verify and Deliver

After completing research, suggest verification and output: