Dossier — Decision-Grade Entity Research — Research

Dossier is a decision-grade entity research skill that produces a hypothesis-tested dossier on a specific company, person, nonprofit, or government org. It forces the user to state their hypothesis upfront, then searches for supporting and disconfirming evidence to deliver a verdict. The output is an editable Word document with identity facts, a 12-month activity timeline, network and reputation signals, red flags, conversation hooks, and a source-provenance audit log.

research claude-code

📂 View skill on GitHub ↗ 🚀 How to use

✨ What it does

Forces a hypothesis via mandatory intake question Q4, ensuring the dossier tests rather than confirms beliefs.
Allocates at least 30% of search budget to disconfirming evidence to counter confirmation bias.
Generates a 9-section Word document with verdict, identity facts, timeline, network, reputation, red flags, conversation
Tiers every citation as primary, secondary, or tertiary, and surfaces tier next to each flag.
Includes a red-flag pass for litigation, regulatory actions, unusual departures, and financial signals.
Produces conversation hooks tied to specific findings with suggested framing.

🎯 When to use it

Preparing for a sales meeting or partnership pitch with a specific company or person.
Conducting investment or acquisition due diligence on a target entity.
Doing journalism or personal vetting (e.g., hiring, dating) where sensitivity exclusions are needed.
Preparing for a job interview or competitive intelligence on a specific organization.

🚀 How to use

Trigger by asking for background research, diligence, or meeting prep on a specific entity, e.g., 'prep me for a meeting with [person/company]' or 'due diligence on [company]'. The skill will ask six forcing questions one at a time (identity, type, purpose, hypothesis, depth, sensitivities). Provide exact identifiers (LinkedIn URL, domain, EIN) and a clear hypothesis. Example prompts:

I'm pitching Microsoft Tuesday. My hypothesis is they're consolidating AI spend on their first-party Foundry platform. Validate or disprove, and give me three conversation hooks tied to what you find.

Due diligence on Acme Corp. I believe their overhead ratio is sketchy. Check the 990s.

📄 Output: A Word document (.docx) with nine sections including executive summary, hypothesis test, timeline, network, reputation, red flags, conversation hooks, and audit log.

📦 Add this skill to Claude Code

# 1. Get the skills repo
git clone --depth 1 https://github.com/alirezarezvani/claude-skills /tmp/claude-skills

# 2. Copy this skill into your project (or ~/.claude/skills for all projects)
mkdir -p .claude/skills
cp -r /tmp/claude-skills/research/dossier/skills/dossier .claude/skills/dossier

Skill source: research/dossier/skills/dossier/SKILL.md

⚠️ Good to know

Requires WebSearch and WebFetch, Node.js with docx package, and optionally bash_tool and curl for free APIs; BYOK MCPs are optional enhancements. The skill refuses to proceed on ambiguous names and requires a hypothesis, falling back to an implicit hypothesis if the user refuses.

❓ FAQ

What happens if I don't have a hypothesis?

The skill pushes back once, asking you to guess and commit to a position you can update later. If you still refuse, it falls back to an implicit hypothesis ('what's the most surprising thing I could find?') and flags the fallback in the audit log.

Can I use this for personal vetting?

Yes, but for journalism or personal vetting contexts, the skill asks about sensitivities to exclude (e.g., medical, family, political history) and honors those exclusions.

What sources does it use?

It uses WebSearch and WebFetch, plus free APIs like SEC EDGAR, GitHub, and ProPublica. Optional BYOK MCPs (LinkedIn, Crunchbase, Apollo, etc.) can enhance coverage, and findings from those are marked as BYOK-sourced.

🤖 Overview, features, install steps and FAQ were generated from the project's SKILL.md on Sep 4, 2026. Always check the original source before running commands.