company
Skill1 starsChanged 3 months agoApache-2.0View on GitHub
Create or review customer-facing, commercial, product, support, and visual work using the governed COMPANY.md, CUSTOMER.md, OFFER.md, VOICE.md, and linked DESIGN.md context. Use when the user invokes $company, says /company, asks for on-brand work, or asks an agent to follow company context; do not use as a substitute for missing business facts or approvals.
5 other projects use a version of this.
Adopted by
Most recently taken up by
- ckaico/reversiumtheir own version, first seen Nov 15, 2026
- tandregbg/core-claude-skillstheir own version, first seen Nov 1, 2026
- varunmoka7/resume-hypergraphtheir own version, first seen Oct 2, 2026
- devmesh-in/claude-companytheir own version, first seen Aug 25, 2026
- MasatoshiSano/seigi-companytheir own version, first seen Mar 22, 2026
Other versions5 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 5
- devmesh-in/claude-company★ 8same name
- tandregbg/core-claude-skills★ 4same name
- MasatoshiSano/seigi-company★ 1same name
- ckaico/reversium★ 0same name
- varunmoka7/resume-hypergraph★ 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 repository2 other files it gives its agents.
- integrations/AGENTS.mdAGENTS.md
- integrations/CLAUDE.mdCLAUDE.md
The file
We split this skill into the rules it gives an agent
7 rules. 0 are common, 2 are distinctive, and 5 are very specific to this project.
0
Common
Most of the 3 edited versions we read keep these.
2
Distinctive
This author's own choices. Worth a look.
5
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 16
When business context already exists and the user is adopting Company.md, run the same skill runner with adopt <workspace> --format pretty …
- Very specific: line 24
Use the VCS root as the workspace boundary when present.
- Distinctive: line 34
Use the narrowest sufficient profile:
- Distinctive: line 52
Before returning the work:
- Very specific: line 54
check Claims to Avoid, commercial guardrails, and customer exclusions.
- Very specific: line 57
if visual, confirm DESIGN.md tokens and prohibitions were used after the business context.
- Very specific: line 63
After writing the receipt, invoke the skill runner with artifact verify <receipt> --root <authorized-root> --format pretty.
“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.
Files
Company
Turn a plain-language request into a deliverable using the company or product explicitly resolved for that task.
Start or adopt a pack
Resolve the absolute path of scripts/run-companymd.mjs beside this file and execute that runner for Company.md CLI calls. It selects COMPANYMD_CLI, a workspace-local install, an installed executable, or the pinned GitHub release. Inspect the runner source when diagnosing a failure, not as a prerequisite to running it. Do not execute arbitrary npm cache files.
When the user asks to set up Company.md and no pack exists, read references/authoring.md. Initialize drafts, derive what the authorized sources support, keep unknowns explicit, and lint the result. Ask for facts that remain necessary after inspecting supplied sources. Do not invent evidence or treat drafting as approval to activate documents.
When business context already exists and the user is adopting Company.md, run the same skill runner with adopt <workspace> --format pretty first. Map useful sources into the four-file minimum; do not rewrite or delete the originals.
Resolve the request
Identify the deliverable, audience, purpose, customer inputs, output format, and whether visual design is in scope. Ask only for missing information that would materially change the result. Treat facts supplied about a named customer as request context, not as permanent Company.md truth.
Load the narrowest context
Use the VCS root as the workspace boundary when present; otherwise use the initial working directory. An explicit user-authorized workspace containing several repositories can be the boundary instead. Pass it as --workspace-root; do not expand it merely to make a failed link load.
Invoke context directly on the supplied workspace or current directory before a recursive file inventory: the resolver discovers the nearest registry or pack. Pass any explicitly named product or company as --subject <id-or-alias> without first checking for a registry; the CLI also validates a simple pack's identity. Omit the selector when no identity was named. Run one context command, which resolves and validates the same snapshot. Use --compact to elide identical inherited prose while retaining source hashes and constraints.
Read the resolved context and binding paths. Do not enumerate sibling packs, historical fixtures, or all installed skills to choose a product or template. Read other related files when the task needs them, after resolving the identity. An unknown or ambiguous identity is a routing error, never a reason to borrow a nearby pack. For registry setup, product changes, or template binding, read references/product-routing.md.
node <absolute-skill-directory>/scripts/run-companymd.mjs context <workspace-or-pack> --subject <id-or-alias> --profile <profile> --compact --workspace-root <authorized-root> --output <temporary-context-file> --receipt <temporary-receipt-file>Use the narrowest sufficient profile:
core: internal company reasoning.customer: customer research or qualification.commercial: packaging, proposals, or sales strategy without authored copy.communications: emails, proposals, landing pages, scripts, support, or product copy.visual: decks, pages, ads, or other designed artifacts. Add--require-designwhen the requested result must use the brand design. For slides, pass--artifact presentation, which requires visual context and design.
Read the generated context and receipt before authoring. Stop dependent work on validation errors; use lint for diagnosis or CI, not as a mandatory duplicate read before context. Add --allow-draft only when the user knowingly requests work with draft context, including creating and testing a new pack. Existing authorization is sufficient; make the draft status clear in the handoff. Never bypass a classification failure.
Create the deliverable
Use the artifact capability appropriate to the requested output. Apply business meaning first, customer relevance second, offer and claim boundaries third, voice fourth, and visual rules last. Never invent prices, customer facts, proof, promises, availability, certifications, or design rules. Make assumptions visible and keep unresolved questions out of audience-facing copy unless the deliverable requires them.
For a sales presentation or pitch deck, read references/sales-deck.md and honor its binary-artifact completion contract. For a before/after validation request, read references/evaluation.md.
Verify and hand off
Before returning the work:
- check
Claims to Avoid, commercial guardrails, and customer exclusions; - require evidence for qualitative time or performance claims such as “hours,” “half a day,” or “instant,” including narrative baselines, just as for numeric ROI or percentages;
- confirm the writing follows
VOICE.mdrules rather than copying example wording mechanically; - if visual, confirm
DESIGN.mdtokens and prohibitions were used after the business context; - distinguish verified facts, user-provided client facts, and assumptions;
- preserve the context receipt beside the deliverable when the output is a file.
For artifact traceability, run scripts/create-receipt.mjs after generation with --context-receipt <temporary-receipt-file> to bind the resolved identity, design, template, and source hashes. Pass --root <authorized-root> so paths remain portable, --contract generic/v1 or the required specific contract, client sources, intermediates, unresolved facts, and verification gates. For a native cloud artifact, also pass --remote-url, --provider, --revision, and --mime-type, export metadata when available, and the artifact/access and artifact/revision gates. When the requested artifact cannot be created, use --expected-deliverable with blocked gates and a concrete cause. Keep the receipt beside the artifact or its local manifest.
After writing the receipt, invoke the skill runner with artifact verify <receipt> --root <authorized-root> --format pretty. Fix verification failures before handoff. Hash checks establish source integrity; declared gates require actual artifact inspection and do not prove visual conformance by themselves. Remote verification is offline unless the provider was independently checked.
Return the artifact first, followed by a concise Company check naming the profile, source files, unresolved facts, and any approval still required. Do not expose the generated context bundle as a second source of truth.
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.

