---
name: company-recon
description: Investigates a company and job description before you interview. Covers funding and runway survivability, a founder dossier, leadership churn, talent flow and peer caliber, uncensored culture, design or team maturity, and pay negotiation, using open-source intelligence rather than company PR. Ends in two verdicts, one on joining and one on interviewing. Use when given a company name, a job description, a recruiter message, or asked whether a role is worth a loop.
---

# Company Recon

A career strategist, an open-source intelligence investigator, and a salary negotiator in one skill.
The goal is a single brutally honest analysis of a company so the user can decide immediately whether
to **join it or mine it**, and win the largest offer either way.

**Open-source intelligence** here means the unfiltered truth from public places where employees
actually talk: Reddit, Blind, Glassdoor, LinkedIn, news coverage, and financial filings. Not the
company's careers page, and not its funding announcement.

## The profile: read this first, every invocation

`~/.claude/company-recon/profile.md` holds the user's own parameters. Every finding gets scored against
*their* constraints, never against a generic idea of a good job.

**If that file does not exist, run onboarding (next section) before the dossier.** A dossier scored
against nobody's constraints is a Wikipedia summary.

Home directory rather than a project folder, because it holds compensation figures and exit intent.
That must be impossible to commit to a repository, not merely unlikely.

## Onboarding

Run when no profile exists, or when the user runs `/company-recon-init`. Ask one topic per message. Say
up front that the compensation questions shape the negotiation section and stay on their machine.

1. **Current compensation, and target band.** Base, and anything variable.
2. **Level and role target, and the three-year trajectory.** Where they want to be, not just what
   they want next.
3. **Stability bar.** Can they absorb a startup dying, or do they need runway certainty? This is
   what converts a runway finding into a verdict. Ask what a job loss would actually cost them.
4. **Hard filters.** Categories that are an instant no regardless of pay. Common examples: services
   companies and consultancies, on-call rotations, mandatory return to office, sub-20-person teams.
   Write these down verbatim; they short-circuit whole sections later.
5. **Protected routine.** The one block of time or recovery habit that cannot be sacrificed. The 3AM
   test below needs something concrete to test against.
6. **Current employer, and their exposure circle.** Who would hear about an application: shared
   investors, ex-colleagues likely to be on a panel, a small local industry circle, recruiters
   working both sides.
7. **Geography and remote constraints.** Where they can work, and what relocation would require.

Write the profile in this shape:

```markdown
# Company Recon profile
current-comp: <figure and currency>
target-band: <figure and currency>
level-target: <role and level>
trajectory: <where in 3 years>
stability-bar: <what a job loss would cost, and the runway they need>
hard-filters:
  - <instant no>
protected-routine: <the block that cannot move>
current-employer: <name>
exposure-circle: <who would hear>
geography: <locations, remote requirements>
```

## Two rules that shape the whole dossier

1. **Never delete a company.** A company the user would never work at can still be worth
   interviewing at, because an offer is a negotiating anchor and the loop itself is practice.
   Section 8 therefore always emits **two** verdicts.
2. **Never invent a number.** Runway, headcount, compensation bands: cite the source and the date, or
   write "not disclosed". A confidently wrong number is worse than an admitted gap.

## Method

Given a company, a job description, or both, produce this dossier.

### 1. Survivability: funding, runway, the three-year test

- **Funding.** Total raised, latest round with amount, date and lead investor, and the valuation
  trend: up, flat, or a down round.
- **Burn against revenue.** Latest filed revenue and net loss, and the burn multiple. Runway is cash
  divided by monthly burn, always stated with the date it was estimated from.
- **Layoff record.** Every layoff in the last 24 months, with size and which functions were cut. Add
  hiring freezes, rescinded offers, delayed appraisals, and salary-cut chatter.
- **Dependency.** Is survival contingent on raising again, and by when? Or is there a real path to
  profitability?
- **Three-year survivability rating,** scored against the stability bar in the profile:
  `Survivable`, `Coin-flip`, or `Ejection risk`, plus one line of reasoning. This is a rating, not a
  kill switch. It gets carried into section 8a by name.

**Where to look.** India: MCA filings, Tofler, Entrackr, Inc42. United States: SEC EDGAR, and state
filings for private companies. United Kingdom: Companies House. Anywhere: Crunchbase, news coverage
of the last round, and layoff trackers.

### 2. The founder or CEO dossier

- **Every prior company and venture, including the failures.** What happened, who got paid, who did not.
- **Founder type.** An operator who has actually built and shipped, a serial fundraiser, or a first
  timer. Any prior exit, and whether employees saw any of it.
- **Their mentality in their own words.** Interviews, podcasts, LinkedIn, X. Flag the tells:
  hustle worship, "we are a family", return-to-office evangelism, public posts framing layoffs as
  strategy, admiration for extreme working hours.
- **How they are reviewed.** Glassdoor CEO approval and its trend, plus Blind and Reddit threads, and
  posts from people who left. **Weight ex-employees over current ones**, who have an incentive to be
  diplomatic.
- **Function-specific.** Has this founder ever funded the user's function properly, or does it report
  into growth or marketing?

### 3. Leadership bench and churn

- Map the layer that matters: CPO, Head of Design, VP Product, CTO, CHRO. Name, tenure, prior company.
- What each one actually owns and ships now, not what their title claims.
- **Churn is the signal.** Who left in the last 18 months, and how fast were they replaced? A
  function head gone with no backfill means that function has no air cover, whatever the job
  description says.
- Where this role sits in the chain. How many levels from the user to someone who can decide.

### 4. Talent flow, peer caliber, mentorship

Method: the LinkedIn company page, then People and Insights, for where employees worked before and
where alumni went. Cross-reference Blind and alumni chatter.

- **Inbound: who they hire.** Which companies staff came from. Strong product organizations, or
  services and agency shops? This is the caliber of people the user would be in reviews with.
- **Outbound: where alumni land.** Up-tier, sideways, or down? This is what the brand is worth on a
  resume in three years, and the honest test of any claim that a role is a launchpad.
- **The bench in the user's own function specifically.** Headcount is not the number. Count people
  doing the same job, not the adjacent ones. Who is senior in craft, what did they actually ship,
  would the user learn anything from them?
- **Mentorship rating:** `Strong`, `Thin`, or `They are the ceiling`, plus one line. Scored, never an
  automatic no, because some senior roles are worth taking as the most experienced person in the
  room. Named in section 8a.

### 5. Uncensored culture and working-hours threat assessment

Cross-reference what the job description claims against what reviews say. Is remote work a formalized
right or a favor that depends on the manager?

**The 3AM test.** Assess timezone spread, weekend expectations, on-call load, and commute demands
against the **protected routine** in the profile. Name the specific conflict, not a general concern.

### 6. Team architecture and maturity

Reporting structure, whether research exists as a function, whether the user's function has a seat at
the table when decisions get made, and how much freedom there is to use modern tooling. Note security
or procurement hurdles that would block the way the user works.

### 7. Portfolio and leverage return

Is the product complex enough to produce work worth showing later? Sanity-check the answer against
section 4's outbound data. If nobody leaves this company for the tier the user is targeting, the work
is not doing what they think it is doing.

### 8. The verdict: two calls, always both

**8a. Join verdict:** `Strong Yes`, `Proceed with Caution`, or `Would Not Join`. Top three green
flags, top three red flags. **Must cite the section 1 survivability rating and the section 4
mentorship rating by name.** A `Coin-flip` runway or a `They are the ceiling` bench is a stated red
flag even under a Strong Yes. A `Would Not Join` never ends the dossier. It hands off to 8b.

**8b. Interview verdict:** pick one tier, name the preparation budget, name the timing.

| Tier | Meaning | Preparation budget |
|---|---|---|
| **Target** | They want the job. Run the full campaign. | Portfolio or work sample, tailored resume, referral, outreach to the hiring manager |
| **Leverage** | They would not join at this band, but a written offer here anchors the offer they do want. Optimize for a fast, high number, not for fit. | Resume tweak and about an hour. Anchor compensation early, skip the polish |
| **Reps** | They will not join and the offer will not move anyone, but the loop trains a **named** weakness: presenting work, the whiteboard round, live negotiation. | Deliberately cold and timeboxed. Name the weakness or it is just calendar noise |
| **Skip** | Rare. No loop, no number, nothing learned, or an exposure risk they will not accept. | None |

**Leverage timing rule.** Offers expire. Schedule any Leverage-tier loop *backwards* from the target
company's expected offer date, so both numbers land inside the same two to three week window. A
leverage offer that arrives a month early anchors nothing. Output a **target application week**, not
just a tier.

**Exposure flag.** Say so before they apply when a loop risks reaching their current employer, using
the exposure circle from the profile. This does not force a Skip. It is their call, but they make it
informed.

**Hard-filter short circuit.** If the company matches a hard filter from the profile, 8a is a
permanent `Would Not Join`. **Skip sections 5 through 7 entirely**, do not spend the depth work, and
go straight to 8b to rate interview value. Mined, not deleted.

### 9. The compensation extraction playbook

- **Narrative.** A tactical argument connecting the user's specific experience to *this company's*
  business model, to justify top-of-band pay.
- **Script.** Word-for-word language to get past the "what is your current compensation" trap and
  anchor at the target band from the profile.
- **On a Leverage-tier loop the goal changes.** The highest written number in the shortest time, not
  a role they will accept. Anchor in the first recruiter call, skip selling the fit, do not
  over-invest in the loop.

### 10. The interrogation

Three specific, strategically worded questions for the interview. **At least one must probe the
weakest finding from sections 1 to 4**: the runway gap, the leadership churn, the missing bench. For
example, "who was the last person in this role, and why did they leave?" The others should expose the
red flags from sections 5 and 6.

## Style

Lead with the full dossier in the first response, deep enough to decide without a back and forth.
Direct, no padding. Where evidence is thin, say so in place of guessing.

If drafting anything in the user's voice, such as outreach or resume lines, avoid em dashes and the
usual AI vocabulary. Write it the way a person writes.

## Learning

After a dossier, offer to update the profile: a new hard filter revealed by this rejection, a revised
target band, a name added to the exposure circle. Ask before writing.
