Skip to main content

Documentation Index

Fetch the complete documentation index at: https://sherlock-osint.vercel.app/docs/llms.txt

Use this file to discover all available pages before exploring further.

When you launch an AI investigation in Sherlock, you’re not just handing the AI a question — you’re configuring the entire investigative frame. Scopes are how you do that. A scope is a domain configuration that tells Sherlock what kind of investigation you’re running, what sources are most relevant, what output sections make sense, what persona the AI should adopt, and how to interpret and present findings. Choosing the right scope is one of the most effective ways to get sharper, more targeted investigation output.

​
How scopes shape a run

When you select a scope during run setup, it influences the investigation in several ways:
  • Domain context — background knowledge about the domain that is passed into the AI prompt to orient it toward the right frameworks and terminology
  • Investigation objective — a default framing for what the run should accomplish, adapted to the domain
  • Suggested sources — a curated list of domain-relevant sources (databases, publications, agencies) surfaced as chips during run setup
  • Categories — the feed and discovery categories that are active for the workspace, matching the domain
  • Personas — a set of role-based personas (journalist, analyst, researcher, etc.) that tune the AI’s voice, depth, and focus
  • Starter prompts — example prompts shown during run setup to help you get started in the domain
  • Output sections — the set of artifact sections generated by default (e.g., a government-fraud run surfaces anomalies prominently; a scientific-research run includes a literature review)
  • Label profile — domain-appropriate labels for workspaces, artifacts, signals, and follow-ups throughout the UI

​
Built-in scopes

Sherlock ships with nine built-in scopes covering the most common investigation domains:
Investigates federal spending, procurement irregularities, lobbying activity, and public corruption. Suggested sources include USASpending, SAM.gov, FEC, FPDS, GAO, IGNET, FOIA.gov, OpenSecrets, ProPublica, and PACER. Best for journalists, watchdogs, and researchers tracking public-sector financial flows.
Covers SEC filings, corporate records, litigation history, ownership structures, and financial reporting. Suggested sources include SEC EDGAR, OpenCorporates, Crunchbase, CourtListener, Bloomberg, Reuters, and the Financial Times. Suited for investment analysis, M&A research, and vendor vetting.
Examines foreign policy developments, international relations, armed conflict, and diplomatic activity. Suggested sources include the U.S. State Department, United Nations, CSIS, CFR, Brookings, the Atlantic Council, RAND, Bellingcat, SIPRI, and Foreign Affairs. Designed for analysts, policy researchers, and intelligence practitioners.
Investigates vulnerabilities, threat actors, indicators of compromise (IOCs), and threat intelligence. Suggested sources include NVD, CVE, MITRE ATT&CK, CISA US-CERT, VirusTotal, Mandiant, CrowdStrike Blog, and KrebsOnSecurity. Built for security researchers, incident responders, and threat-intel analysts.
Covers market analysis, product comparison, patent landscapes, and competitive positioning. Suggested sources include Crunchbase, PitchBook, CB Insights, Google Patents, USPTO, PatentsView, Gartner, G2, TechCrunch, and Wired. Useful for product teams, strategy groups, and business analysts.
Synthesizes academic papers, clinical studies, and research findings across scientific domains. Suggested sources include Google Scholar, PubMed, arXiv, Semantic Scholar, Nature, Science, Cell Press, PLOS, NIH, WHO, and CDC. Ideal for researchers, science journalists, and policy analysts working with technical literature.
Tracks AI model developments, research lab activity, and industry trends across the technology sector. Suggested sources include OpenAI, Anthropic, Google DeepMind, Meta AI, arXiv, Hugging Face, Papers with Code, The Information, TechCrunch, and Semafor. Best for technology analysts, AI researchers, and venture investors.
Covers legislation, regulatory actions, compliance developments, and policy analysis across jurisdictions. Suggested sources include the Federal Register, Congress.gov, EUR-Lex, FTC, SEC, CISA, Brookings, CSIS, and Lawfare. Suited for policy analysts, compliance teams, legal researchers, and government affairs practitioners.
An unconstrained scope with no fixed source list and no domain-specific framing applied to the prompt. Use this when your topic doesn’t fit a specific domain, when you want the AI to range freely across source types, or when you’re in early exploratory research and haven’t yet narrowed the focus.

​
Personas

Each scope ships with a set of personas — role-based configurations that adjust the AI’s voice, analytical lens, and depth of focus. A persona shapes how the AI frames its output: a journalist persona emphasizes newsworthy angles and public-interest framing; an analyst persona emphasizes structured assessment and confidence calibration; a researcher persona emphasizes methodological rigor and literature grounding. You select a persona during run setup. The selected persona is stored with the run record and reflected in the artifact’s provenance metadata, so you can always see which persona produced a given report.

​
Custom scopes

If none of the built-in scopes fits your use case, you can create your own. Go to Settings → Scopes to define a custom scope with your own domain context, investigation objective, suggested sources, categories, and personas. Custom scopes are stored locally alongside the built-ins and are available in run setup immediately after you save them.
1

Open Settings

Navigate to Settings → Scopes from the app navigation.
2

Create a new scope

Click New Scope and fill in the name, description, domain context, investigation objective, and any suggested sources or categories relevant to your domain.
3

Add personas (optional)

Define one or more personas to tune the AI’s voice for your domain. Each persona needs an ID, a label, and an instruction that describes the role and analytical style.
4

Save and use

Save the scope. It will appear alongside the built-in scopes the next time you set up a run.
If you’re not sure which scope to use, start with Open Investigation. It imposes no domain constraints, so you can explore freely. Once you have a clearer sense of what the investigation requires, switch to the matching domain scope for a more targeted follow-up run — the domain context and source suggestions will sharpen the output considerably.