A drafting-and-review copilot for a founders' / co-founders' agreement — the terms fixing equity, vesting, IP, roles, control, deadlock, and departure between cofounders. Jurisdiction-agnostic, anchored on the Delaware C-corp default. Two modes: DRAFT (intake → equity & vesting → clauses → blocker triage → pre-signature check) and REVIEW (audit an existing agreement against an 18-clause checklist and red-flag scan). It handles the highest-dispute terms first-class: the equity split as documented reasoning (not a fake calculator), reverse vesting and the 83(b) clock, present-tense IP assignment (the Stanford v. Roche trap), leaver buyback and dead equity, and the deadlock clause most tools omit. It drafts for the venture, never one founder against another. Not legal advice.
A Claude skill that turns the drafting and review of a founders' / co-founders' agreement into a repeatable method — the document (or set of terms) that fixes equity, vesting, IP ownership, roles, decision-making, deadlock, and departure between the people starting a company.
Jurisdiction-agnostic, anchored on the Delaware C-corp default, with the highest-dispute terms handled first-class — the equity split, reverse vesting and the 83(b) clock, present-tense IP assignment, leaver mechanics, and the deadlock clause most tools skip.
What it does
Runs in two modes.
DRAFT — five phases, fourteen steps, from intake to signature:
| Phase | What you produce |
|---|---|
| 1 — Intake & founder mapping | Founder-and-role map, entity/jurisdiction determination, contribution inventory |
| 2 — Equity & vesting architecture | The split with a written rationale (not a fake calculator), the vesting schedule, the 83(b) flag, acceleration |
| 3 — Core clause drafting | IP assignment, roles & deadlock, leaver & buyback, transfer/ROFR, and the supporting terms — into the right instrument |
| 4 — Conflict & blocker triage | Desirable-vs-blocking, plus divergent-interest points routed to independent counsel |
| 5 — Pre-signature finalisation | The "clean, investable cap table" diligence dry-run; open blockers closed as conditions |
REVIEW — audit an existing agreement against an 18-clause checklist and a red-flag scan, and output a triaged gap report (Critical / Important / Optional).
The spine running through every step: vesting is the mechanism, not the split; document the rationale, not just the number; present-tense IP assignment or nothing; every share must have a home on departure; design the deadlock before it happens; draft the terms into the right instrument and make them expire cleanly.
Install
Drop the folder into your Claude skills directory:
git clone https://github.com/sboghossian/founder-agreement-drafting.git \
~/.claude/skills/founder-agreement-drafting
Then invoke it in Claude Code with /founder-agreement-drafting, or just
describe a founder deal and it triggers on phrases like "draft a founders'
agreement", "split equity between founders", "set up founder vesting", "founder
IP assignment", "founder deadlock clause", "review this founders' agreement."
Files
SKILL.md— the executable workflow (the skill itself).REFERENCE.md— the research backbone, with every clause, the case law (Stanford v. Roche), the equity-split data (Wasserman / NBER), the jurisdiction table (Delaware / LLC / UK / MENA), and primary-source citations.
Scope
This is a drafting method, not legal, tax, or financial advice. It tells you where each founder term must live and how it must behave — it does not certify that a term is enforceable in your jurisdiction, decide who "deserves" more equity, or recommend a tax election. It drafts for the venture as a whole, the way company counsel does: each founder should have independent counsel before signing. Jurisdiction-specific terms (non-compete enforceability, MENA onshore forfeiture, LLC profits-interest tax, the 83(b) decision) are flagged for local / tax counsel, not supplied. Prompts to a public AI tool are not privileged; work with abstracted placeholders.
Credit
Part of a series of open legal skills from HAQQ Legal AI,
initiated with Abbas (Chief Legal Officer). The method here is synthesised
from public best-practice sources — Y Combinator, Cooley GO, Clerky, Carta,
Orrick, Wilson Sonsini, SeedLegals, Noam Wasserman / HBS, and named case law
(all cited in REFERENCE.md). Packaged as a Claude skill by Stephane
Boghossian (Head of Growth, HAQQ Legal AI).
License
AGPL-3.0. Anyone who builds this method into a hosted or distributed product must open-source the derivative.