agentleFS
TheCraigHewittTheCraigHewitt/cowork-starter-pack/skills/research-brief/SKILL.md

research-brief

Skill5 starsChanged 6 months agoMITView on GitHub

Produces a one-page research brief on a topic, person, or company — what they do, what matters, what's changed recently, sources. Use when the user asks to research, look into, brief me on, or get context on a specific topic, company, competitor, or person. Pulls from the web, any project context, and recent emails involving the subject.

We split this skill into the rules it gives an agent

11 rules. 0 are common, 10 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 18 edited versions we read keep these.

  • 10

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Ask the user what they want a brief on if it's not clear.

    line 12
  2. Distinctive:

    Ask what they need the brief *for.* A sales call?

    line 13
  3. Very specific:

    Read context/preferences.md — source citation style, confidence level tagging if relevant

    line 14
  4. Distinctive:

    Check the project folder — have you researched this subject before?

    line 15
  5. Distinctive:

    Search the web — company site, news, LinkedIn if it's a person, press releases, recent podcast/video appearances for public figures

    line 36
  6. Distinctive:

    Do not regurgitate their "About" page.

    line 44
  7. Distinctive:

    Do not invent specifics. If you can't find revenue numbers, don't estimate them.

    line 45
  8. Distinctive:

    Do not include information that's more than 18 months old unless it's foundational (company history, product roadmap).

    line 46
  9. Distinctive:

    Do not produce a 10-page report.

    line 47
  10. Distinctive:

    For a sales brief: Add a section: "Likely objections" — based on their stage, budget, existing stack.

    line 80
  11. Distinctive:

    For a competitive brief: Add a section: "Where they win, where they lose" — honest assessment.

    line 81

“Common” is measured: kept by at least 6 in 10 of the 18 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Research Brief

Turn "I need to know about X before my 2pm call" into a one-page brief the user can skim in three minutes and feel prepared.

Before starting

  1. Ask the user what they want a brief on if it's not clear. A company? A person? A topic? A competitor?
  2. Ask what they need the brief for. A sales call? A strategic decision? A blog post? Same subject, different briefs.
  3. Read context/preferences.md — source citation style, confidence level tagging if relevant
  4. Check the project folder — have you researched this subject before? Build on it, don't redo it.

What to produce

A markdown file saved to /output/research-brief-{subject}-{YYYY-MM-DD}.md with four sections:

  1. The 30-second version — three to five sentences. If the user only reads this section, what do they need to know?
  2. What they do / what it is — core facts. Company: product, size, customers, how they make money. Person: role, background, what they're known for. Topic: the basics.
  3. What matters right now — recent news, strategic shifts, public statements, anything timely. What's changed in the last 90 days.
  4. Sources & confidence — what you drew from, and where you're guessing vs. citing. Label each claim with a confidence marker if the user asked for it in preferences.md.

Inputs

  • Web search (required)
  • The user's prior research on this subject in the project folder (if exists)
  • Recent emails mentioning the subject (optional — useful for competitive briefs)
  • Any docs the user uploaded with the request

How it works

  1. Understand the subject and the purpose (ask if unclear)
  2. Search the web — company site, news, LinkedIn if it's a person, press releases, recent podcast/video appearances for public figures
  3. Cross-reference with any existing project context
  4. Draft the four sections. Short. Dense with information. No padding.
  5. Flag anything uncertain or contested — never smooth over ambiguity
  6. Save the file, tell the user one-line what stood out ("Strong product, weak financials — the gap is interesting if you're pitching to them")

What not to do

  • Do not regurgitate their "About" page. Synthesize, don't copy-paste.
  • Do not invent specifics. If you can't find revenue numbers, don't estimate them. Say "not publicly disclosed."
  • Do not include information that's more than 18 months old unless it's foundational (company history, product roadmap). Prioritize recent.
  • Do not produce a 10-page report. A one-page brief is the format. Respect the format.

Example output structure

# Research Brief — Northwind Inc. — 2026-04-18

## The 30-second version
Northwind is a 200-person B2B workflow SaaS targeting mid-market ops teams, about $18M ARR, series B last year ($24M led by Accel). They're pivoting from horizontal workflow to vertical (finance ops first). New CRO joined in January. Growth was strong through 2024 but has slowed visibly in the last two quarters — this is likely why they're engaging on our pricing conversation now.

## What they do
- **Product:** workflow automation for ops teams — think Zapier but with more structured primitives
- **Customers:** ~600 SMB and mid-market, mostly US, growing international presence
- **Pricing:** $500–$3,000/month per team, custom for enterprise
- **Team:** 200 FTEs, HQ in Austin, 60% eng/product

## What matters right now
- **Vertical pivot.** Announced in March — finance ops as the beachhead vertical. Hiring finance-ops-specific AEs.
- **Slower growth.** Q4 2025 growth rate was 8% QoQ, down from 14% in Q2. Public hiring slowdown in engineering.
- **New CRO.** Marcus Alvarez, ex-Greenhouse, joined January. Known for aggressive deal closing.
- **Competitive pressure.** Retool and Zapier both pushed into their mid-market segment in Q1.

## Sources & confidence
- Company site, March 2026 press release (high confidence — primary source)
- Crunchbase funding data (high confidence — primary source)
- LinkedIn for headcount and CRO details (high confidence)
- ARR estimate from The Information's March coverage (medium confidence — reported, not disclosed)
- Growth rate QoQ from their annual all-hands leak on Blind (low confidence — unverified)
- Vertical pivot and competitive context from recent Latent Space podcast interview with their CEO (high confidence — on the record)

Customization notes

  • For a sales brief. Add a section: "Likely objections" — based on their stage, budget, existing stack.
  • For a competitive brief. Add a section: "Where they win, where they lose" — honest assessment.
  • For a prospect brief (person). Focus sections 2 and 3 on the individual — background, what they've said publicly, any connection to the user's network.
  • For a market brief (topic). Replace sections 2 and 3 with "The landscape" and "What's shifting" — same structure, broader scope.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.