agentleFS
corebits-iocorebits-io/deal-origination-playbook/skills/business-research/SKILL.md

business-research

Skill0 starsChanged 2 months agoMITView on GitHub

Research each target company properly before anyone speaks to it — what the business does, scale evidence, tenure and history, succession and ownership-change signals quoted and dated, what the owner appears to care about, and an explicit list of what is not known. Produces a one-page target brief per company. Triggers on: research this company, research the owners, target brief, brief me on, pre-call research, succession signals, deal brief, prepare for the call, what do we know about this company, read the signals.

3 other projects use a version of this.

Adopted by

Most recently taken up by

The file

Show its rules and text

We split this skill into the rules it gives an agent

15 rules. 0 are common, 9 are distinctive, and 6 are very specific to this project.

  • 0

    Common

    Most of the 3 edited versions we read keep these.

  • 9

    Distinctive

    This author's own choices. Worth a look.

  • 6

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Read mandate-profile.md — section 2 for the thesis, section 10 for what the user is cleared to say about themselves

    line 23
  2. Distinctive:

    Read the market map — the brief must describe the company in the map's segment language

    line 25
  3. Very specific:

    Never brief a Not a Fit or Near Miss row: If the user asks for one specifically, do it and say why the row is not in the keep pile.

    line 30
  4. Distinctive:

    Ask how many to brief. Recommend 10 to 20 at a time.

    line 33
  5. Very specific:

    All three, or it is Not found.

    line 89
  6. Distinctive:

    Before recording any signal, check it against this list.

    line 96
  7. Very specific:

    Use templates/target-brief.md, one file per company, into briefs/ in the workspace.

    line 117
  8. Distinctive:

    Three specifics only someone who read about them would know: Not flattery, not generic sector observations.

    line 120
  9. Distinctive:

    Not a discovery checklist

    line 123
  10. Distinctive:

    Keep the brief to one page of substance.

    line 131
  11. Very specific:

    Update est_revenue / est_ebitda only with a published figure or a labeled estimate with the basis stated inline.

    line 141
  12. Very specific:

    If research contradicted the earlier classification — sponsor ownership found, family rather than founder, a different segment — say so …

    line 144
  13. Distinctive:

    Every company where you found no signal at all

    line 155
  14. Distinctive:

    Every contradiction with the existing row

    line 156
  15. Distinctive:

    This is where the playbook stops: No outreach copy, no sequences, nothing sent

    line 217

“Common” is measured: kept by at least 6 in 10 of the 3 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

Business Research

Produces the one page you take into a first conversation: what this company is, what the evidence says, what the signal is, and — the part that matters most — what you do not know.

Read first:

  • references/signals-and-evidence.md — signal classes, the evidence standard, what is noise, the line you do not cross
  • references/ownership-taxonomy.md — ownership context and succession dynamics
  • references/research-ethics.md — public professional sources only
  • references/scoring-rules.md — anti-fabrication rules

Step 0 — Load and scope

  1. Read mandate-profile.md — section 2 for the thesis, section 10 for what the user is cleared to say about themselves
  2. Read the market map — the brief must describe the company in the map's segment language
  3. Read targets.csv
  4. Filter to fit_score = Strong Fit or Fit

Never brief a Not a Fit or Near Miss row. If the user asks for one specifically, do it and say why the row is not in the keep pile.

Ask how many to brief. Recommend 10 to 20 at a time. A brief is 30 to 60 minutes of real research, and a rushed brief is worse than none — it produces a confident first conversation built on inference.


Step 1 — Research the business

Sources, in order of value:

  1. The company's own site — services, segments, locations, history, team
  2. Registry filings — formation, officers, changes, addresses
  3. Licence and certification rosters — dated proof of operating standard
  4. Association listings and committee roles
  5. Trade press and local press — anniversary pieces, award profiles, expansion coverage
  6. Job postings — what they are hiring tells you what they are doing
  7. Review platforms — volume and age, for scale and tenure proxies
  8. Procurement and contract award records, where public

What to establish:

FieldStandard
What it doesIn the map's segment language, no unsourced adjectives
ScalePublished figure, or labeled estimate with basis, or Not found
OwnershipAlready classified — verify it still holds, note if it changed
Tenure and historyDated events with sources
LocationsOpened, closed, or acquired, with dates
Structural changesName changes, new entities, holdco insertions

Rules:

  1. Quote, do not paraphrase, anything you will treat as a signal
  2. Never infer financials from headcount and present them as fact
  3. A dated website is not a finding about the business. See the non-signal table in references/signals-and-evidence.md
  4. Record what you checked and found nothing in. That belongs in the brief
  5. Date everything. evidence_date on every claim that has one

Step 2 — Read the signals

Work the four classes in references/signals-and-evidence.md:

  • A. Succession — tenure with no visible successor, retirement language, a step back to Chairman, a second generation visibly present or visibly absent, association roles handed over
  • B. Ownership change — officer changes, PSC updates, new registered agent, a holdco inserted, share transfers, name changes
  • C. Operating change — new facility, certification added, hiring burst, large contract, capability added
  • D. Stress — licence lapse, closure, overdue filings, litigation of record

For each signal found: the exact quote or filed fact, the URL, and the date. All three, or it is Not found.

Score signal_strength per the table in references/signals-and-evidence.md — high, medium, low, or Not found.

🚧 HARD GATE — the inference line

Before recording any signal, check it against this list. If it comes from any of these, do not record it and do not mention it:

  • An inference about anybody's health, age, or family circumstances
  • A personal social media account
  • Anything obtained by pretexting — posing as a customer, supplier, applicant, or journalist
  • Family members who are not in the business
  • A photograph read for age or infirmity

A signal is something a person said or filed in public, in a professional context. Beyond the ethics, inferred personal signals are usually wrong, and acting on one in a first conversation is unrecoverable.

Also check independence: three articles quoting one press release are one signal, not three. Record it once and cite the primary source.


Step 3 — Write the brief

Use templates/target-brief.md, one file per company, into briefs/ in the workspace. Every section, including:

  • Three specifics only someone who read about them would know. Not flattery, not generic sector observations. Concrete facts about this company
  • Questions to ask — the real ones this research raised and could not answer. Not a discovery checklist
  • What we do not know — explicit, itemized, and never filled in with an inference at the last minute

The last section is the point of the whole document. A brief that hides its gaps produces a confident conversation built on an invention, and owners in small sectors remember that.

Keep the brief to one page of substance. Length is not thoroughness; a brief nobody reads before the call has failed.


Step 4 — Write back to targets.csv

Fill signal, signal_source, evidence_date, signal_strength, and set brief_status = briefed and status = briefed.

Update est_revenue / est_ebitda only with a published figure or a labeled estimate with the basis stated inline.

If research contradicted the earlier classification — sponsor ownership found, family rather than founder, a different segment — say so loudly and flag the row back to ownership-classification or mandate-fit-scoring. Do not quietly overwrite a scored row.

🚧 HARD GATE — before writing

Show:

  • Which companies you briefed
  • One complete brief in full, so the user can judge the standard
  • The signal found for each, with strength, quoted
  • Every company where you found no signal at all
  • Every contradiction with the existing row
  • Anything you checked and could not establish

Get approval. Then write.


Step 5 — Report

Business research complete — 14 companies briefed

Signals found:
  high        4   (2 owner-stated succession, 2 filed ownership change)
  medium      6   (long tenure + no visible successor, both sourced)
  low         2   (operating change only)
  Not found   2   ← approach on mandate fit and merit, no angle

By class:
  A succession        7
  B ownership change  3
  C operating         5
  D stress            1   (overdue filings — diligence note, not an angle)

Evidence recency:
  Under 12 months     8
  1-3 years           4
  Over 3 years        0   ← none presented as current intent

Rejected as non-signals:      11
  (dated websites 4, no profile 3, small headcount 2, sector consolidation 2)
Rejected on the inference line: 2
  (an inferred retirement age, a personal social post — neither recorded)

Contradictions found:
  [Company A] — holdco inserted 2025, likely sponsor deal → back to classification
  [Company B] — second generation runs ops → family-owned, not founder-owned

Scale evidence:
  Revenue published    2
  Labeled estimate     6
  Not found            6

Re-score recommended: 10 rows now have a sourced signal and may reach Strong Fit.

Notes

  • Fit and signal are different questions. This skill answers the second. Never let a strong signal argue a company into the mandate
  • Not found on signal is fine. It means you go on mandate fit and general merit rather than a specific angle. It does not lower fit_score
  • Recency bands matter. A 2018 quote about retiring soon is a 2018 fact — and if it happened, the company may be somebody's add-on now. Check
  • Re-score after a research batch. Rows blocked from Strong Fit only by a missing signal can now be reassessed
  • Briefs go stale. Re-check before any second cycle
  • Never commit briefs. They hold personal data and later conversation notes. The .gitignore covers the workspace patterns; keep it that way
  • This is where the playbook stops. No outreach copy, no sequences, nothing sent

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.