ma-deal-workbench-review

name

ma-deal-workbench-review

description

Produces a structured JSON deliverable for an M&A deal-review task driven by the running deal workbench — asset/stock purchase agreement issue registers, SPA closing & economics packages, carveout transition reviews, M&A committee escalations, and SPA deviation matrices. Use when a task casts you as buyer-side or seller-side counsel, points you at a deal workbench reachable over the network, asks you to compare draft terms against a playbook or committee policy, classify issues by status/risk, quantify dollar exposure from the purchase price, and return only JSON conforming to a shipped answer_template.json. Reads base URL and read-only SQL token from environment_access.md and never imports another deal's schema, IDs, or values.

M&A Deal Workbench Review

Produce a structured JSON deliverable for an M&A deal-review task by querying the running deal workbench, comparing the current draft against the applicable playbook or committee policy, and emitting JSON that conforms to the task's own answer_template.json.

Read these first, in order

1. The task prompt — gives the deal_id, client_side (buyer/seller), deliverable type and required coverage, the playbook/policy id to compare against, and the unit/precision rules.
2. input/payloads/answer_template.json — the authoritative schema for THIS task: required fields, allowed enums, stable ID lists, ordering rules, and units. Templates differ across tasks; do not carry another task's schema forward.
3. environment_access.md — use it only to reach the running environment over the network (base URL, read-only SQL token, allowed-endpoint list). It is not a source of deal data or answer values.

Workflow

1. Extract task parameters from the prompt (deal_id, client_side, deliverable, playbook/policy id, precision rules).
2. Reach the workbench — read environment_access.md; substitute its base URL for every <TASK_ENV_BASE_URL> in the prompt. See references/workbench_access.md.
3. Pull records by exact deal_id — deal record, draft terms, playbook rules (and/or policy thresholds), risk estimates, employees, consents, material contracts, regulatory, benchmarks, diligence findings, cap table, documents, notes. Pull what the deliverable needs. Do not assume records from a similarly named project apply.
4. Classify each relevant term against the playbook/policy — issue_status, risk_rating, recommended_action (and redline_action where the template has redlines). Derive allowed enum values from THIS task's template. See references/analysis_method.md.
5. Handle missing terms and distractors — treat a missing/silent required term as an issue when the client's position needs an affirmative provision; exclude stale/in-policy/distractor terms only when the task says to.
6. Quantify from the purchase-price base — dollars from the headline purchase price unless a source states a different basis; quantify the draft against the playbook's preferred and fallback positions; compute deltas/shortfalls; take exposure bands from risk estimates.
7. Build ordering, blockers, and summary metrics — priority order per the template's rule; closing blockers are required closing conditions + regulatory clearances only; derive all counts/totals from the set you built.
8. Emit JSON conforming to THIS task's template, with the task's units/precision, using stable workbench IDs. See references/output_contract.md.

Critical rules

• Conform to THIS task's answer_template.json — its fields, enums, stable IDs, ordering, and units. Never import a prior deal's schema.
• Use stable IDs from the workbench; never invent or reuse IDs from another deal. source_term_ids is [] for a missing required term.
• Quantify from the deal's headline purchase price unless a source explicitly states a different basis.
• Units/precision vary per task — read them from the prompt + template and follow exactly.
• Return only valid JSON; no narrative outside it.
• Query by the exact deal_id; do not let a similarly named project's records bleed in.

References

• references/workbench_access.md — reaching the workbench: base URL, token, endpoint catalog, read-only SQL.
• references/analysis_method.md — classification, quantification, missing-term/distractor handling, blockers, ordering, summary metrics.
• references/output_contract.md — template conformance, per-task units/precision, stable IDs, JSON-only, no cross-deal values.