company-recon
Skill0 starsChanged 27 days agoMITView on GitHub
Fast strategic recon on any B2B company from a single URL. Use this skill whenever someone pastes a company website and wants to understand the business quickly. Triggers include "run recon on this company," "what can you tell me about [URL]," "give me a snapshot of [company]," "research this prospect," "pull up this website," "company overview," "scout this account," "who are these people," "break down this company," "check out this vendor," or any variation where a URL arrives and the reader wants to know what the company does, who it sells to, how it is positioned, what marketing it is running, and whether it matters to us. This is the fast pre-call read, not a deep multi-company study. Trigger it even when the user pastes a bare URL with no other words.
1 other project uses a version of this.
Adopted by
Most recently taken up by
- rishb0/resume-harnesstheir own version, first seen Oct 15, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- rishb0/resume-harness★ 0same 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 repository9 other files it gives its agents.
- brand-kit-setupSkill
- brand-visualsSkill
- competitive-white-spaceSkill
- content-writerSkill
- google-ads-auditSkill
- house-styleSkill
- linkedin-ads-auditSkill
- social-generatorSkill
- transcript-to-articleSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
16 rules. About 0 are common, 6 are distinctive, and 10 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈6
Distinctive · estimate
This author's own choices. Worth a look.
10
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 20
Read these first. They are the source of truth and this skill does not invent around them.
- Very specific: line 29
If brand-kit/ is missing or still has its template placeholders, stop and tell the user to run brand-kit-setup.
- Distinctive: line 74
If a page 404s or bounces to the homepage, drop it and move on.
- Very specific: line 191
Trap exposure: check Named Traps in brand-kit/positioning.md.
- Very specific: line 195
Never name a competitor in this section who is not in the Competitors table or found in your own research for this specific company.
- Very specific: line 209
Write these the way brand-kit/voice.md says to write: direct, specific, mechanism over outcome, no consulting filler.
- Distinctive: line 226
No hedging: If it is an inference, say "signals suggest" once and move on.
- Very specific: line 235
Use the bundled docx skill and let the host resolve it.
- Very specific: line 246
If brand-visuals is installed, offer to turn the markdown into a branded PDF instead.
- Very specific: line 249
Styling comes from the brand kit, not from this file: Read the Palette table in brand-kit/brand.md and map it:
- Very specific: line 260
Use the fonts from Typography in the same file.
- Distinctive: line 278
Run every fetch and every search in parallel.
- Distinctive: line 280
If a page or search returns nothing, skip it silently.
- Distinctive: line 283
After the document exists, give a two to three sentence summary in chat: the single most useful finding, and the Section 6 classification.
- Very specific: line 293
competitive-white-space: if the recon turned up a competitive set worth studying, or the user says "now run the competitors." That skill … - Very specific: line 296
content-writer: if a gap in Section 5 is worth answering with an asset.
≈ “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 Recon
Before you write anything
Read these first. They are the source of truth and this skill does not invent around them.
brand-kit/brand.md: name, palette, fonts, CTAsbrand-kit/voice.md: house style and banned constructionsbrand-kit/positioning.md: the frame, pillars, traps, stats, vocabularybrand-kit/icp.md: who you sell to, and your competitor setbrand-kit/capabilities.md: read before making any product claim
If brand-kit/ is missing or still has its template placeholders, stop and tell
the user to run brand-kit-setup. Do not proceed with invented positioning.
What this produces
A seven-section strategic snapshot of any B2B company, built from public sources in about five minutes, delivered as a formatted Word document.
The reader should finish it knowing what the company does, who it sells to, how it is positioned, what marketing it is actually running, where its funnel is thin, and whether it is a prospect, a partner, a competitor, or noise.
When to use it
- Pre-call research before a sales, partner, or investor conversation
- Scouting an inbound lead, an event contact, or a name from a campaign list
- A quick read on one player in your category before deciding to study it further
- Partner evaluation: reseller, systems integrator, or technology alliance
- Any time the only input is a URL
When not to use it
- Three to five companies compared against each other. Use
competitive-white-space. - A paid-channel performance review. Use
google-ads-auditorlinkedin-ads-audit. - Long-form content built from the findings. Recon first, then
content-writer.
Workflow
Step 1: Fetch the website, in parallel
Fire these fetches simultaneously. Do not ask the user which pages to visit.
From the base URL:
- Homepage (the URL provided)
- About or company page: try
/about,/about-us,/company - Product or services page: try
/products,/services,/solutions,/platform - Pricing: try
/pricing,/plans. Often absent, which is itself a signal. - Blog or resources index: try
/blog,/resources,/insights. Index only, for content themes. - Customers or case studies: try
/customers,/case-studies,/success-stories
If a page 404s or bounces to the homepage, drop it and move on. Do not retry.
Step 2: Run targeted searches, in parallel
Four to six searches to fill the gaps a website will never admit to:
[company] LinkedIn company page: follower count, posting cadence[company] reviewsOR[company] G2OR[company] Capterra: outside perception[company] competitors: who the market compares them against[company] funding OR revenue OR employees: scale signalssite:[company domain] [category terms]: run this probe using the Category terms fromSearch Termsinbrand-kit/icp.md. This is what tells you whether the company plays in your market at all. If that list is empty, use the category named inbrand.mdunderCompanyand say in the document that the probe was approximate."[company]" case study OR testimonial: what proof they have in market
Step 3: Produce the document
Build the seven sections below, then hand off to the document step.
Document sections
Section 1: What they do
Two to three sentences. The problem they solve, who has it, and the business model: software, services, marketplace, platform, agency. Frame it around the buyer's problem, not their feature list.
Section 2: Who they sell to
A tight table or bullet list:
- Industries: verticals visible in case studies, logos, and copy
- Company size: enterprise, mid-market, SMB, or a mix. Infer from pricing, logo wall, and the vocabulary of the site.
- Buyer roles: who signs. Infer from messaging, demo CTAs, content topics.
- Geography: footer, office locations, currency, case studies
Section 3: How they are positioned
- Core value proposition: one sentence, pulled from the hero or H1
- Claimed differentiators: three to five bullets on what they say is unique
- Proof in market: logos, case studies, certifications, awards, data claims
- Pricing signal: model and range if public. "Sales-led, custom pricing" if not, which tells you about deal size and cycle.
Section 4: What marketing they are running
Walk every visible channel:
- Website and SEO: content depth, blog volume, resource library, site quality, visible conversion paths
- LinkedIn: followers, posting frequency and themes, whether leadership posts
- Paid: evidence of paid search or paid social from search results and public ad libraries
- Content and thought leadership: webinars, podcasts, reports, ebooks, video
- Email: newsletter signup, gated assets, lead magnets
- Outbound and events: conference presence, partner directories, sponsorships
Close the section with a Marketing Maturity Rating, bolded so it stands out:
| Rating | What it looks like |
| Early | Thin content, no visible paid, no funnel |
| Building | Some content, first paid experiments, one conversion path |
| Established | Consistent content, multi-channel paid, a clear funnel |
| Advanced | Attribution, personalization, full-funnel coverage, segmented offers |
State the rating and give the one-line basis for it.
Section 5: Messaging by buying stage
Map what they have, and what they are missing, at each stage:
- Unaware (problem education): content that teaches the market before mentioning the product. If none exists, that is the finding.
- Aware (solution differentiation): how they argue their approach is different. Comparison pages, methodology content, positioning pieces.
- Engaged (conversion): what converts. Demo, trial, assessment, consultation, download. Note the primary CTA and whether there is more than one path for different intent levels.
Where a stage is thin, add one [Suggestion]: a concrete, realistic idea for a company of that type and size. One per stage, maximum.
Section 6: Relevance to you
The judgment section. Classify the company into exactly one bucket and defend it in two to three sentences.
Read before writing this section:
The Fit Matrixinbrand-kit/icp.md: score the company against the ICP column on seniority, size, function, and industry. A company that clears three of four dimensions is a prospect. Two of four is mixed. One is not.Categories We Are Notinbrand-kit/positioning.md: if the company lives in one of those adjacent categories, it is far more likely a partner or a non-competitor than a rival, no matter how similar the homepage sounds.What We Do Not Doinbrand-kit/capabilities.md: the overlap test cuts both ways. A company doing something you explicitly do not do is not a competitor.
The four buckets:
| Classification | Test |
| Prospect | They match the ICP column of the Fit Matrix and show the problem you solve. Note which buying-committee role from Buying Committee is the entry point. |
| Partner | They sell to your ICP but solve an adjacent problem. Name the specific handoff or integration that would make the partnership real. |
| Competitor | They pitch your buyer on your problem with a comparable claim. Say whether they are the kind you lose deals to or the kind that shows up in search, using the tier language in Competitors. |
| Not relevant | Wrong buyer, wrong problem, or wrong scale. Say so in one line and stop. Do not stretch. |
Then add two short reads:
- Category posture: how they talk about the category your
Category termsprobe covers. Building in it, buying into it, or ignoring it entirely. - Trap exposure: check
Named Trapsinbrand-kit/positioning.md. Which named trap, if any, does this company's stack or messaging show symptoms of? Name the trap and the symptom. If none, say none. Do not force it.
Never name a competitor in this section who is not in the Competitors table or
found in your own research for this specific company.
Section 7: Strategic observations
Five to seven bullets. This is the "so what."
- What is working that they should do more of
- The biggest gap or missed opportunity
- How a company with this model and market typically builds pipeline
- Red flags or headwinds visible from the outside
- One non-obvious insight from connecting dots across the research
- A restatement of the Section 6 classification in one line, with the reason
Write these the way brand-kit/voice.md says to write: direct, specific,
mechanism over outcome, no consulting filler. Every bullet should survive
someone asking "so what do I do with that?"
Writing rules
- Audience: executive. Written for someone who will act on it today.
- No em-dashes. Commas, periods, colons.
- No emojis.
- No links or URLs in the body. The reader already has the URL.
- No code blocks.
- Paragraphs of two to three sentences, or bullets.
- Bold sparingly, for the rating, the classification, and nothing else.
- Tables for anything comparative.
- Do not invent data. Unknown is "Not visible from public research."
- No hedging. If it is an inference, say "signals suggest" once and move on.
- Apply the banned words and AI tells from
brand-kit/voice.mdbefore delivery.
Output
Produce a Word document.
Use the bundled docx skill and let the host resolve it. Follow that skill's
own implementation rules rather than restating them here: explicit page size,
real bullet numbering rather than unicode characters, table widths declared on
both the table and its cells, cell margins for readability, heading styles
overridden by their built-in IDs, and separate paragraphs instead of newline
characters inside a run. Validate the file with whatever validation step that
skill provides before presenting it.
If no docx skill is available: output the same seven sections as clean
markdown in the conversation, keep the tables, and tell the user plainly that no
document skill was found so they got markdown instead. Do not silently downgrade.
If brand-visuals is installed, offer to turn the markdown into a branded PDF
instead. Nothing in the analysis changes; only the format does.
Styling comes from the brand kit, not from this file. Read the Palette
table in brand-kit/brand.md and map it:
| Document element | Palette role |
| Title, section headers | Primary |
| Subsection headers, body text | Body text |
| Divider rules, table header fill | Accent, or Accent soft for fills |
| Table borders, secondary labels | Rule / border, Muted text |
| Callouts for good news / problems | Positive and Problem, with their tints |
Use the fonts from Typography in the same file. If those fonts are not
guaranteed on a recipient's machine, fall back to a common system font and note
the substitution once. The footer string is the Document Footer value from
brand.md, set right-aligned in the muted text color.
Name the file after the company and the analysis. Where it goes depends on the host, so do not assume a path. If the session has a designated output or deliverables folder, write it there. If it does not, which is the normal case in a plain command line session on a laptop, write it to the current working directory. Either way, say in one line where you put it. A file written to a folder that does not exist on this machine is a silent failure, and the user finds out by not finding it.
Speed notes
- Do not ask clarifying questions. The URL is the whole input.
- Run every fetch and every search in parallel. Serial research is why recon takes twenty minutes instead of five.
- If a page or search returns nothing, skip it silently. Absence is only worth reporting when it is strategic: "no case studies anywhere on the site" is a finding, "the /team page 404s" is not.
- After the document exists, give a two to three sentence summary in chat: the single most useful finding, and the Section 6 classification. Do not reproduce the document in chat.
What comes next
Optional handoffs. Each one runs on its own if the user goes straight to it.
competitive-white-space: if the recon turned up a competitive set worth studying, or the user says "now run the competitors." That skill will use this recon's findings if they are in the conversation, and gather its own if not.content-writer: if a gap in Section 5 is worth answering with an asset.google-ads-auditorlinkedin-ads-audit: if the subject is your own company and Section 4 raised questions about paid spend.
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.

