Kelly Legal Contracts

Overview

Use this skill as the legal-contract review desk. The agent ingests contract facts and extracted clauses, drafts clause issues (risk notes, fallback language, negotiation notes) against a company clause playbook, runs deterministic risk checks, and gives the reviewer a Busabase-backed App-in-Skill review queue (approve / request changes / block) plus an approval-gated issue-list export. Reading a contract document or a legal intake is a genuine external operation a browser cannot perform: scripts/ingest_contracts.mjs is the only place a contract or issue enters the system, scripts/run_checks.mjs runs the risk-check rules, and scripts/execute_decisions.mjs records the planned follow-up for approved/changes-requested issues. The AirApp itself only reads Busabase and writes review decisions; export happens through scripts/export_issues.mjs, and every outbound legal position, redline, counterparty message, signature, or filing is delegated to the user or a separate approved connector after explicit approval.

Default behavior is AirApp-first. Unless the user explicitly asks only for explanation, ingest/check what's due and give the user the clickable AirApp URL (or the local preview URL when local preview is explicitly requested). Use chat-only mode only when the user says "纯聊天", "chat only", "不要打开 UI", or similar; then present numbered issues (Issue #1) and take verdicts in conversation.

App UI Screenshots

<table> <tr> <td width="50%"><img src="assets/screenshots/overview.webp" alt="Kelly Legal Contracts overview"></td> <td width="50%"><img src="assets/screenshots/needs-review.webp" alt="Kelly Legal Contracts review queue"></td> </tr> <tr> <td><strong>Overview</strong><br>Legal command desk with contract × workstream status, risk pass rate, review queue preview, and recent activity.</td> <td><strong>Review queue</strong><br>Approval-gated legal issue queue with approve / request changes / block decisions and audit notes.</td> </tr> <tr> <td width="50%"><img src="assets/screenshots/checks.webp" alt="Kelly Legal Contracts risk checks"></td> <td width="50%"><img src="assets/screenshots/issues.webp" alt="Kelly Legal Contracts issue workbench"></td> </tr> <tr> <td><strong>Risk checks</strong><br>Per-rule pass/warn/fail results across clause issues, including hard-stop terms and playbook violations.</td> <td><strong>Clause issues</strong><br>Editable issue detail with fallback language, memo fields, reviewer rationale, and risk-check evidence.</td> </tr> </table>

Mandatory Dependencies

1. Read and follow $kelly-app-skill-creator for product behavior, visual quality, responsive layout, and the complete canonical content/kelly-legal-contracts-app/ artifact.
2. Read and follow $busabase for connection, target Space, node discovery, ChangeRequests, review, and merge behavior.
3. Read and follow $busabase-app-creator for resource modeling, AirApp runtime limits, security, validation, and deployment.

If a dependency is unavailable, preserve this skill's local artifact and product contracts, stop before the unavailable Busabase operation, and report the exact missing dependency. Do not invent a second data backend.

Boundary

• This skill assists legal operations and contract review; it does not provide final legal advice and does not replace licensed counsel or an authorized legal reviewer.
• Ingesting a contract or issue is a local-file-only operation: scripts/ingest_contracts.mjs reads a JSON payload file the agent prepares (from contract text, summaries, or a legal intake) and writes it to Busabase. It never fetches documents from remote systems on its own.
• The AirApp reads and writes Busabase records only. It never sends redlines, emails counterparties, accepts terms, signs contracts, deletes files, or performs any other external side effect.
• Every outbound legal position, redline, counterparty message, approval, signature, filing, or waiver is approval-required and happens outside the app only after explicit human approval; scripts/execute_decisions.mjs never performs the send itself — it only writes an execution marker.
• If a hard-stop term is present (uncapped liability, perpetual data retention, broad indemnity, missing DPA terms), block or request changes rather than weakening the rule.
• Treat contracts, counterparties, and playbooks as sensitive. Never commit local payload files, env files, or generated exports. Never expose raw secrets or private playbooks through logs, reports, screenshots, or demo data.

Busabase Resources

Six Bases under one application Folder (kelly-legal-contracts), declared in content/kelly-legal-contracts-app/app/js/config.js and the generated template sidecars under content/:

• contracts: the contract library — counterparty, matter type, governing-law/deal facts, key obligations, watch terms, and the required-document checklist.
• issues: the clause-issue workbench and review queue in one — workstream-specific fields (risk notes, fallback language, negotiation notes, memo, business ask, structured facts), workflow status, risk score, and the human decision + execution marker on the same row.
• checks: per-issue, per-rule risk-check results (required fields, title length, hard-stop terms, restricted positions, risk-note/business-ask counts, memo length, all-caps noise, watch-term repetition, document checklist, clause-playbook violations).
• claims: the clause playbook's approved fallback clauses and rejected positions.
• claim-rules: the clause playbook's hard-stop / restricted-phrase rules.
• settings: one row (record-id: "config") with the legal profile, per-workstream rule sets, hard-stop/restricted terms, and export preferences.

Resources provision lazily through an idempotent Busabase ChangeRequest the first time the app runs in a Space; see references/contracts-schema.md for exact field shapes. Risk scores, the review queue, the recent-activity feed, and metrics are recomputed client-side from the stored rows on every read (content/kelly-legal-contracts-app/app/js/contracts-model.js's buildSnapshot/assembleSnapshot), so the desk is always fresh regardless of when a browser session loads it relative to the last ingest/checks run.

First Run And Onboarding

On invocation, check the issues Base. If it is empty, guide setup before ingesting real contracts: ask, turn by turn, legal profile (company/entity, reviewer role, preferred risk scale), enabled workstreams (nda, msa, dpa, sow) with per-workstream rule sets, jurisdictions, hard-stop terms, escalation policy, and export preferences. Write the answers onto the Settings row, then ingest and check:

node skills/kelly-legal-contracts/scripts/ingest_contracts.mjs payload.json --apply
node skills/kelly-legal-contracts/scripts/run_checks.mjs --apply

Local App

Default behavior is AirApp-first — give the user the clickable AirApp URL. Start pnpm --dir content/kelly-legal-contracts-app dev only when local preview/debugging is explicitly requested.

Required app views (hash routes):

• #/overview: legal command desk — KPI cards (contracts, issues, risk pass rate, exported this week), contract × workstream status matrix, review-queue preview, recent activity (derived from each issue's own timestamps).
• #/contracts and #/contracts/<contract_id>: the contract library — counterparty, matter type, source badge, workstreams, linked issues. Detail shows contract facts, key obligations, watch terms, the required-document checklist, and linked issues.
• #/issues and #/issues/<issue_id>: the clause-issue workbench — counterparty, workstream, jurisdiction, title, risk score, workflow status. Detail shows an editable field workbench (title, risk notes, fallback language, negotiation notes, memo, business ask, or structured facts depending on the workstream), reviewer rationale, and per-rule risk-check evidence.
• #/checks: risk-check results — every rule per issue with pass/warn/fail badges and evidence, filterable by rule, workstream, contract, and result.
• #/claims: the clause playbook — approved fallback clauses, rejected positions, and hard-stop/restricted-phrase rules.
• #/review: the review queue — every issue with its workflow state (needs_review / changes_requested / approved / done / blocked), risk summary, agent suggestions, Review note, decision buttons (approve / request changes / block), and a stable ref (Issue #1). Decisions write directly onto the issue record through busabase-sdk; field edits saved in the workbench arrive as revise decisions carrying the edited fields.
• #/settings: sanitized config summary — legal profile, per-workstream rule sets, hard-stop/restricted-term counts, export prefs, read live off the Settings Base.

Demo mode:

• ?demo=overview, ?demo=contracts, ?demo=issues, ?demo=checks, ?demo=claims, ?demo=review, and ?demo=detail open deterministic mock scenes for documentation and screenshots.
• lang=en or lang=zh forces UI chrome language; contract/legal terms may stay in English when that is realistic.
• Deep links such as /?demo=review&lang=zh#/review must work.
• Demo mode never reads or writes Busabase. Decision buttons still work in the UI but act on in-memory state only.

UI language: English and Chinese chrome with Auto default. Keep real contract content, counterparty names, and legal positions in their original language.

Ingest Workflow

1. Collect inputs: contract type, counterparty, our entity, governing law, deal owner, target date, business ask, extracted clause text, and company playbook positions.
2. Draft the issue as a structured ingest payload per the workstream's field shape (nda: title/risk-notes/fallback-language/negotiation-notes/redline-outline; msa: title/fallback-language/memo; dpa: title/business-ask; sow: title/subtitle/fallback-language/structured-facts). Include the contract facts if the contract does not exist yet.
3. Run the write path:

node skills/kelly-legal-contracts/scripts/ingest_contracts.mjs payload.json --apply

The script validates the payload against the workstream field shapes and the required-fields rules stored on the Settings row, normalizes contracts/issues, and upserts them into Busabase by natural key (contract_id/name+counterparty, issue_id) so re-ingests are idempotent. Without --apply it is a dry run.

Check Workflow

1. Run node skills/kelly-legal-contracts/scripts/run_checks.mjs --apply. Deterministic rules (required fields, title length, hard-stop terms, restricted positions, risk-note/business-ask counts, memo length, all-caps noise, watch-term repetition, document checklist, clause-playbook violations) are computed from the workstream rule sets on the Settings row and the clause playbook (claims/claim-rules Bases); per-issue risk scores are recomputed idempotently.
2. Summarize failures for the reviewer by ingesting compliance_summary/suggestions onto the issue record.
3. Give the user the AirApp URL and send them to #/review.

Decisions And Execution Workflow

1. The reviewer decides at #/review or the issue workbench: approve, request changes (with a note), save edited fields (revise), or block. Decisions write directly onto the issue record. From a standalone local preview the write merges immediately (trusted operator); from the deployed AirApp it creates a pending ChangeRequest for the trusted process to merge.
2. On explicit user request to execute, run scripts/execute_decisions.mjs (dry-run by default; --apply writes execution-status: "ready_for_agent" onto each decided issue with the concrete operation — export_issue_list (from approve) or request_revision (from request_changes) — and target). No external side effects either way; the issue's workflow status never changes.
3. The agent then performs the approved follow-up outside the app: for export_issue_list, run scripts/export_issues.mjs and hand off the redline/counsel communication through the user or a separate approved connector after approval; for request_revision, redraft the issue per the review note, re-ingest, and re-run checks.

Export Workflow

1. node skills/kelly-legal-contracts/scripts/export_issues.mjs --out <dir> reads issues with a genuine human "approve" decision from Busabase and writes each as a clean Markdown issue memo plus issues.csv (default exports/, gitignored). Marks each exported issue done in Busabase — this is the only write export performs, and it never happens for an issue that merely has status: "approved" from a spoofed ingest payload without a real decision.
2. Actual sending, redline generation, counterparty communications, CLM updates, signature, or filing happen only outside the app after explicit approval, through the user or a separate approved connector/skill.
3. Keep exports out of git and report the concrete file paths.

Safety Defaults

• Never fabricate legal approval, waiver, signature authority, regulatory conclusion, or attorney review.
• Treat legal, privacy, money, IP, employment, and dispute-related terms as approval-required.
• Do not send issue lists, redlines, fallback clauses, or counterparty messages without an explicit approve decision.
• Use stable ids and natural-key upserts so repeated ingests, checks, and executions are idempotent.
• If the issue payload and the workstream rule set disagree (unknown fields), stop and reconcile before executing.

Useful Commands

node skills/kelly-legal-contracts/scripts/ingest_contracts.mjs payload.json --apply
node skills/kelly-legal-contracts/scripts/run_checks.mjs --apply
node skills/kelly-legal-contracts/scripts/execute_decisions.mjs
node skills/kelly-legal-contracts/scripts/execute_decisions.mjs --apply
node skills/kelly-legal-contracts/scripts/export_issues.mjs --out exports/
pnpm --dir skills/kelly-legal-contracts/content/kelly-legal-contracts-app dev

In normal use, invoke /kelly-legal-contracts, let the skill ingest/check what's due, and open the AirApp.