Contract review against a playbook

The brief is the playbook: what you accept, what you push back on and with what wording, what you never sign. One model maps every clause of the contract to the rule it meets or breaks, writes the redlines in the playbook’s wording, and writes the negotiating note; a model from another family checks each redline against the playbook and sends the set back when a clause is missing or a redline goes further than its rule. The lawyer decides what is sent to the other side.

npm install @obversa/runtime @obversa/engine-claude-cli @obversa/engine-codex-cli

In that directory, create package.json with "type": "module", put the brief and the sample inputs from examples/use-cases/business/ beside it, then run the copied file.

The file

import { claude } from '@obversa/engine-claude-cli';
import { codex } from '@obversa/engine-codex-cli';
import {
  briefFromFile,
  formatEvent,
  person,
  run,
  stage,
  workflow,
  type TeamSeat,
} from '@obversa/runtime';

interface ContractPlaybookEngines {
  readonly claude: (model: string) => TeamSeat;
  readonly codex: (model: string) => TeamSeat;
}

const realEngines: ContractPlaybookEngines = { claude, codex };

/**
 * Contract review against a playbook. The brief is the playbook: what we
 * accept, what we push back on, what we never sign. One model maps every
 * clause of the contract to the rule it meets or breaks, writes the
 * redlines, and is read by a model from another family that checks each
 * redline against the playbook and sends the work back when one is missing
 * or goes further than the playbook allows. A lawyer decides what is sent.
 */
function createContractPlaybook(engines: ContractPlaybookEngines = realEngines) {
  return workflow('contract-playbook', {
    brief: briefFromFile('briefs/playbook.md'),
    options: { timeout: '15m' },

    roles: {
      review: engines.claude('claude-sonnet-4-5'),
      'playbook-check': [engines.codex('gpt-5.6-luna')],
      lawyer: person('Send these redlines to the other side?'),
    },

    stages: [
      stage('clauses', {
        agent: 'review',
        writes: 'review/clauses.md',
        desc: 'Read contracts/msa.md and list every clause with the playbook rule it meets or breaks, one line each.',
        gate: 'Every numbered clause of the contract is on the list.',
      }),

      stage('redline', {
        agent: 'review',
        writes: 'review/redlines.md',
        desc: 'For each clause that breaks a rule, write the replacement wording the playbook allows, with the rule it comes from.',
        gate: 'Every breaking clause has a redline, no redline goes further than its rule, and a checker from another family has accepted the set.',
        reviewedBy: 'playbook-check',
        retry: 3,
      }),

      stage('positions', {
        agent: 'review',
        writes: 'review/positions.md',
        desc: 'Write the negotiating note: what to hold, what to concede and to what, and what is a walk-away.',
        gate: 'Every redline has a position, and a reviewer from another family has accepted them.',
        reviewedBy: 'playbook-check',
        // Three attempts: the allowance matches how open-ended the work is.
        // Deciding what to hold, what to concede and what is a walk-away is
        // the most judgement-heavy step here, and it had none while listing
        // clauses had three.
        retry: 3,
      }),

      stage('negotiate', {
        input: 'lawyer',
        desc: 'Put the redlines and the positions in front of the lawyer.',
        gate: 'The lawyer has decided.',
        sendsBackTo: 'redline',
      }),
    ],
  });
}

const result = await run(createContractPlaybook(), {
  onEvent: (event) => console.log(formatEvent(event)),
});
console.log(JSON.stringify(result.outcome, null, 2));

Every event of the run prints as one line as it happens, the usage lines included, and the outcome prints last as JSON. The run stops at negotiate with the redlines and the positions in review/. A no goes back to redline with the lawyer’s note as the finding.

What a run printed

The output below is from one real run of this file, with the two command line tools signed in on the machine. The transcript is trimmed at both ends: a reader’s own run prints a ▸ run event line before the first stage and — unless the process is killed first — a ◂ run line carrying the outcome and the run’s token total after the last stage. The machine that captured this also had an extra local search tool available to the first seat, which this example does not configure, so those calls are elided and a reader’s own run will not make them. The run stops at the person, which is what every example here does before anything is sent or filed.

contract-playbook ▸ dag (4 nodes)
contract-playbook · node clauses: start
contract-playbook › clauses • clauses
contract-playbook › clauses engine:thinking
contract-playbook › clauses engine:text
… 4 more event lines …
contract-playbook › clauses engine:thinking
contract-playbook › clauses engine:text
contract-playbook › clauses   tool Read use
contract-playbook › clauses   tool tool result
contract-playbook › clauses engine:thinking
contract-playbook › clauses engine:text
contract-playbook › clauses   tool Write use
contract-playbook › clauses   tool tool result
contract-playbook › clauses engine:thinking
contract-playbook › clauses   tool Bash use
… 128 more event lines …
contract-playbook › positions › positions-review   tool tool result
contract-playbook › positions › positions-review engine:thinking
contract-playbook › positions › positions-review engine:text
contract-playbook › positions › positions-review   claude-sonnet-4-5-20250929: 1544326/6105 tok
contract-playbook › positions › positions-review • positions: fail
contract-playbook › positions › positions-review · until met: positions writes: true
contract-playbook › positions › positions-review › review-panel • positions
contract-playbook › positions › positions-review › review-panel • positions-1
contract-playbook › positions › positions-review › review-panel engine:text
contract-playbook › positions › positions-review › review-panel   gpt-5.6-luna: 90370/1942 tok
contract-playbook › positions › positions-review › review-panel • positions-1: pass
contract-playbook › positions › positions-review › review-panel • positions: pass
contract-playbook › positions › positions-review · review: pass
contract-playbook › positions › positions-review ◂ pass (2 iter)
contract-playbook · node positions: done (pass)
contract-playbook · node negotiate: start
contract-playbook › negotiate • negotiate
contract-playbook › negotiate • negotiate: paused
contract-playbook · node negotiate: done (paused)
contract-playbook ◂ dag paused

{
  "status": "paused",
  "summary": "waiting for a person: Send these redlines to the other side?",
  "data": {
    "clauses": {
      "status": "pass",
      "summary": "All 7 contract clauses reviewed against playbook. 1 accepted (governing law), 4 require pushback (term auto-renewal, payment terms and price increases, liability cap too low, data processing without proper controls), 2 are never-sign blockers (uncapped indemnity, exclusivity requirement)."
    },
    "redline": {
      "status": "pass",
      "summary": "Review panel: 1/1 reviewer(s) cleared."
    },
    "positions": {
      "status": "pass",
      "summary": "Review panel: 1/1 reviewer(s) cleared.",
      "data": {
        "response": "Done. Fixed all six issues in review/positions.md:\n\n**Changes made:**\n\n1. **Clause 1 (Term)** - line 30: Changed from allowing 45 days notice concession to requiring both renewal by written agreement AND 60 days notice with no concession.\n\n2. **Clause 2 (Payment)** - lines 35-39: Changed from allowing 21 days concession to requiring 30 days minimum with no concession.\n\n3. **Clause 2 (Price)** - lines 43-48: Changed from allowing 7% cap and 60 days notice to requiring 5% or CPI (whichever is lower) and 90 days notice with no concession.\n\n4. **Clause 3 (Liability)** - lines 52-57: Changed from allowing nine or six months concession to requiring twelve months with no concession.\n\n5. **Clause 5 (Data)** - lines 61-66: Changed from allowing 60 days deletion to requiring 30 days deletion with no concession.\n\n6. **Negotiation Strategy** - lines 70-76: Rewrote to state all six redlines are required as written with no middle ground, removing the previous strategy that suggested accepting concessions.\n\nAll positions now match playbook requirements exactly."
      }
    },
    "negotiate": {
      "status": "paused",
      "summary": "waiting for a person: Send these redlines to the other side?",
      "data": {
        "requestId": "negotiate#1#8cf706409cc45450f55cea668f1ab7a7c224e4efb2740004a63b091e92c22264",
        "gateId": "negotiate",
        "gateVersion": 1,
        "digest": "8cf706409cc45450f55cea668f1ab7a7c224e4efb2740004a63b091e92c22264",
        "decisionText": "Send these redlines to the other side?",
        "responseSchema": {
          "type": "object",
          "properties": {
            "approved": {
              "type": "boolean"
            },
            "note": {
              "type": "string"
            }
          },
          "required": [
            "approved"
          ]
        },
        "input": {
          "positions": "Review panel: 1/1 reviewer(s) cleared."
        },
        "presentation": {}
      }
    }
  }
}

The team’s shape

step	does	done when
clauses	Read contracts/msa.md and list every clause with the playbook rule it meets or breaks, one line each.	Every numbered clause of the contract is on the list.
redline	For each clause that breaks a rule, write the replacement wording the playbook allows, with the rule it comes from.	Every breaking clause has a redline, no redline goes further than its rule, and a checker from another family has accepted the set.
positions	Write the negotiating note: what to hold, what to concede and to what, and what is a walk-away.	Every redline has a position, and a reviewer from another family has accepted them.
negotiate	Put the redlines and the positions in front of the lawyer.	The lawyer has decided.

Where a stage’s gate names a file, the package checks the file after the job. Where it names an exit code, the command decides. Where it names a reviewer, the reviewer’s decision decides. Where it names a person, the run pauses until the person answers. A model’s own report of its work never passes a stage.

Gotchas

The checker holds the redlines to the playbook, both ways. A never-sign clause with no redline sends the set back, and so does a redline that asks for more than its rule allows.
The contract is a workspace file the brief names. files: ["contracts/msa.md"] in the brief’s front matter tells every seat the file exists and forbids the review seat from writing it.
The playbook is the whole brief. Everything a seat knows about what is acceptable comes from briefs/playbook.md; a rule that is not written there is not applied.

Source

The file is examples/use-cases/business/contract-playbook.ts; its proof, brief and sample inputs sit beside it. The runtime package page lists every input.

Start

Concepts

Use cases

Examples

Reviewing

Recording runs

Driving runs

Writing workflow shapes

Memory

Workspace

Compare

Glossary

Core packages

Plugins

Contract review against a playbook

The file

What a run printed

The team’s shape

Gotchas

Source

​The file

​What a run printed

​The team’s shape

​Gotchas

​Source

The file

What a run printed

The team’s shape

Gotchas

Source