draft-a-legal-document

Draft a legal document for you, like an NDA, a customer contract, an offer letter, a privacy policy, terms of service, a board decision, a reply to a customer data request, or a brief to send a real lawyer. I work from your existing templates if I have them, or from market-standard wording with a clear note. Drafts only, never sent or signed.

Jump to install

Repository: ThomasMoreAI/legal-skills-open
Last source activity: June 1, 2026 at 17:27
Detected SKILL.md language: English
Stars: 40
Forks: 4

Install options

The review-first prompt is selected by default. You can switch to a direct command or download a local copy.

Review the source files

Read SKILL.md and any companion files shown by SkillsMP before deciding whether to install.

Back to search Back to installation

name	draft-a-legal-document
title	Draft a Legal Document
description	Draft a legal document for you, like an NDA, a customer contract, an offer letter, a privacy policy, terms of service, a board decision, a reply to a customer data request, or a brief to send a real lawyer. I work from your existing templates if I have them, or from market-standard wording with a clear note. Drafts only, never sent or signed.
author	gethouston
author_url	https://github.com/gethouston/houston/tree/main/store/agents/legal/.agents/skills/draft-a-legal-document
license	MIT
version	0.1.0
execution_mode	open
jurisdiction	general
practice	contracts
language	en

Draft a Legal Document

One skill for every first-draft founder need. type param picks template + structure + citations. "Drafts only, never send / file / sign" discipline shared.

Parameter: `type`

Commercial paper (reads template library first):

nda - bilateral or one-way NDA anchored on your template.
consulting - consulting / contractor agreement anchored on CIIAA + deliverables + term.
offer-letter - employee offer letter anchored on 409A + compensation + vesting + at-will language.
msa - customer-facing master services agreement.
order-form - order form tied to existing MSA.
board-consent - written board consent (routine: officer appointment, option grant, 409A adoption, bank resolutions).

Privacy / policy:

privacy-policy - full Privacy Policy with AI-training disclosure, SCCs, subprocessor list, legal-basis citations.
tos - Terms of Service (usage, IP, acceptable use, liability cap, dispute forum).

Regulatory response:

dsr-response - GDPR Art. 15 / CCPA DSR first-touch packet: acknowledgment + identity-verification request + export cover note (3 files).

Escalation:

escalation-brief - structured brief for outside counsel: 2-3 sentence matter summary, specific questions for lawyer, quoted excerpts with cite, deadline, recommended firm type (corporate / commercial lit / privacy / IP / employment). Never names specific firms.

User names type in plain English ("draft NDA with Acme", "write our privacy policy", "package this for counsel") → infer. Ambiguous → ask ONE question naming 10 options grouped by bucket.

When to use

Explicit: "draft {type}", "write our privacy policy", "respond to this DSR", "escalate this to counsel".
Plain-English asks map to a type: "draft an NDA for {counterparty}" → nda; "draft a contractor / consulting agreement" → consulting; "draft a customer contract" / "MSA for {customer}" / "order form" → msa or order-form; "draft a board consent" / "board decision for {action}" → board-consent; "draft / update our privacy policy" → ; "draft our terms of service" → ; "a customer asked for their data" / "respond to a DSR / GDPR / CCPA request" → ; "package this for a lawyer" / "prep a handoff to outside counsel" / "trademark handoff to a lawyer" → .

privacy-policy

tos

dsr-response

escalation-brief

Read ledger + legal context. Gather missing required fields per above. Write atomically.
Discover tools via Composio only when type needs one: googledocs / notion for mirror-copy (optional), googledrive for reading template library, firecrawl for landing-page scrape (privacy-policy, tos).
Branch on type.
- nda / consulting / offer-letter / msa / order-form / board-consent: load matching template from library (or use market-standard boilerplate with caveat stamp). Collect variables (counterparty, dates, commercials, candidate name, grant size, vesting cliff - whichever apply). Substitute variables. Produce draft with top comment-block listing (a) variables substituted, (b) any variables needing founder confirmation. Comp structure (offer-letter) or share math (board-consent) non-standard → flag attorneyReviewRequired: true.
- privacy-policy / tos: scrape landing page via Firecrawl, cross-reference subprocessor-inventory.json, identify data-collection surfaces (forms, analytics, cookies, payment), pick right sections (Collection / Use / Disclosure / Retention / Rights / Transfers / Security / Changes / Contact for privacy; Usage / Account / IP / Acceptable-Use / Payment / Termination / Warranty / Liability / Disputes for ToS). Cite GDPR articles for EU-inclusive geography, CCPA/CPRA for US. AI-training disclosure explicit (opt-in or opt-out, state it). Produce sectioned markdown draft.
- dsr-response: compute statutory clock (GDPR Art. 15 → 30 days, CCPA → 45 days). Produce three files: acknowledgment.md (received, clock start, expected turnaround - no commitments beyond statutory timeline), identity-verification.md (what we need to confirm it's them), export-cover-note.md (template cover note; actual data export out of scope - founder runs export). Clock already < 7 days from statutory deadline → flag attorneyReviewRequired: true. Write as folder dsr-responses/{request-id}-{YYYY-MM-DD}/.
- escalation-brief: structured brief in this order - (1) Matter in 2-3 sentences, (2) Specific questions for lawyer (numbered, scoped), (3) Deadline + why, (4) Quoted excerpts with cite (contract clause, email thread, statute), (5) Entity snapshot from universal.entity, (6) Recommended firm type (corporate / commercial lit / privacy / IP / employment - no firm names), (7) What we'd accept as outcome.
Write draft atomically (*.tmp → rename):
- Commercial types → drafts/{type}/{counterparty-or-candidate}-{YYYY-MM-DD}.md.
- privacy-policy → privacy-drafts/privacy-policy-{YYYY-MM-DD}.md.
- tos → privacy-drafts/tos-{YYYY-MM-DD}.md.
- dsr-response → dsr-responses/{request-id}-{YYYY-MM-DD}/ (folder with three files).
- escalation-brief → escalations/{matter-slug}-{YYYY-MM-DD}.md.
Optional Google Docs mirror. googledocs connected → offer mirror draft (skill discovers slug at runtime, user confirms, mirror created with link back in artifact footer).
Append to outputs.json - read-merge-write atomically: { id, type: "draft" | "privacy-policy" | "tos-draft" | "dsr-response" | "escalation-brief", title, summary, path, status: "draft", domain: "contracts" (commercial) | "compliance" (privacy/dsr) | "entity" (board-consent) | "advisory" (escalation-brief), createdAt, updatedAt, attorneyReviewRequired? }.
Summarize to user. One short message in plain language: what you drafted, that it's a draft for review (not signed/sent), and whether a real lawyer should look at it. Never mention file names, paths, or the underlying procedure.