company-researcher
Skill3 starsChanged 4 months agoMITView on GitHub
Use when the user names a company they are applying or interviewing with, asks whether a startup is safe to join, wants to know if a company's AI claims are real, or needs talking points and questions before a call.
1 other project uses a version of this.
Adopted by
Most recently taken up by
- Sach1ng/PM-operating-OStheir own version, first seen Oct 24, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- Sach1ng/PM-operating-OS★ 3same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository6 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- cold-outreach-writerSkill
- jd-decoderSkill
- mock-interviewerSkill
- resume-tailorSkill
- salary-negotiatorSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 4 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 23
If cash, burn, or revenue are unknown, write "unknown," recommend the user ask, and never invent runway.
- Distinctive: line 27
Engineering blog: do they publish real technical decisions?
- Distinctive: line 28
Eval methodology: do they describe how they measure model quality?
- Distinctive: line 54
If research surfaces something serious (quarterly layoffs, founder controversy, Builder.ai-pattern claims), lead with it.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Company Researcher
Produce the brief a well-connected friend would give: what the company actually does, whether it is financially and operationally credible, whether its AI is real, and what to say in the room. Candidates should diligence a startup like a junior investor because they are investing years of their career. High-profile startup failures repeatedly show that famous investor logos are weak evidence; operating artifacts are stronger evidence.
What you need
Company name (plus URL if obscure), the target role, and the stage: about to apply, screen scheduled, or onsite scheduled. Depth scales with stage. If you have web access, research live. If not, say so and ask the user to paste the about page, JD, and recent news, then work from that.
Process
Step 1: The business in plain words
What they sell, to whom, and what the buyer pays for. If you cannot state it in two sentences, that is a finding, and "how do you make money" becomes a great interview question.
Step 2: Survival math (borrow the VC lens)
- Runway estimate: last raise (amount, date) minus elapsed burn. Rough burn proxy: headcount × fully loaded cost, plus compute for AI companies. State it as an estimate with the inputs shown.
- Capital efficiency: if revenue is real, the diagnostic question is "how much net burn per dollar of net new ARR?" Do not apply a universal stage benchmark without a current, comparable source. The user can ask a softened version: "How has the plan changed since the last raise?"
- A company 18+ months past its last raise with no revenue story hires differently than one fresh off a round. Say which pattern this is.
- If cash, burn, or revenue are unknown, write "unknown," recommend the user ask, and never invent runway. Diligence reduces risk; it does not eliminate private-company opacity.
Step 3: The four-artifact AI-washing test
Classify AI-native vs AI-forward vs AI-washed using artifacts, not branding (regulators now fine companies for fake AI claims; candidates should apply the same skepticism):
- Engineering blog: do they publish real technical decisions?
- Eval methodology: do they describe how they measure model quality?
- Model/data ownership: is it honestly stated (own models, fine-tunes, or API layer)? API-layer is fine; hiding it is not.
- Production customer outcome: named customers with concrete results?
Also run the hiring-mix check: a "frontier infra" company hiring mostly sales with a thin engineering core does not match its claim. Many eval/agent/infra roles = the work is real.
Step 4: The team and the moment
Recent 6 months: launches, big customers, layoffs, leadership changes. Who leads the org this role sits in; their public writing. Open-roles pattern: a cluster of similar openings means a funded push; a lone re-posted role can mean churn. Check a layoffs tracker before believing a hiring spree.
Step 5: Sources and their weights
Levels.fyi for compensation and level comparables. Blind and Glassdoor for repeated themes only; anonymous sentiment is not fact, and one review is an anecdote. LinkedIn headcount graph for growth and the engineering-vs-GTM mix. Engineering blog, GitHub org, changelog, and status page for the AI-washing test. Label every claim in the brief: measured, reported, or anecdote.
Output format
- Two-line summary (business + stage)
- Survival math: runway estimate with inputs, momentum, risks, all dated
- AI classification with the four artifacts checked
- The team you would join: known facts vs what to verify in the room
- Talking points: 3 specific references (a launch, a blog post, a design decision) that show real research; specific beats flattering
- Questions to ask them: 3-5, at least one that surfaces something the user genuinely needs (runway, roadmap ownership, why the role is open). Phrase red flags as questions: "How has the plan changed since the last raise?" extracts more than an accusation.
- Red flags, if any: stated plainly, sourced, dated
Hard rules
- Date every claim; "raised $40M" means nothing without "in March 2026."
- Separate verified facts from inference from anecdote, explicitly, every time.
- If research surfaces something serious (quarterly layoffs, founder controversy, Builder.ai-pattern claims), lead with it. The user is choosing where to spend years.
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
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.

