Reflect 3 Attempt 02

M&A Deal Workbench — Counsel Task Skill

When to use

This skill applies whenever you are given a structured M&A transaction counsel task with a deal workbench that exposes deal records, draft terms, playbooks, policies, and diligence data through HTTP APIs. The task will ask you to prepare a structured JSON answer — an issue register, closing package, committee escalation memo, transition review, or deviation matrix — by gathering facts from the workbench and comparing them against a playbook or policy.

Core workflow

1. Orient from the prompt

Read the task prompt to extract these invariants:

• Role and side: Are you buyer-side or seller-side counsel? This determines which playbook or policy to apply and whether draft terms that are absent or one-sided are treated as pro-buyer or pro-seller issues.

• Deal identifier: The deal_id (e.g. PRJ_XXXXX) that anchors all API lookups.

• Transaction type: Asset purchase agreement, stock purchase agreement, public company merger, or carveout APA. The type drives which terms are expected and which template sections are in play.

• Task category: One of a small set of recurring shapes —

  Category	What you produce
  Issue register	Per-issue comparison of draft terms against a playbook, with risk ratings, recommended actions, quantified deltas, and a negotiation priority order.
  Closing / economics package	Holder-level consideration allocation, indemnity and escrow mechanics, NWC adjustment, required consents, employee and restrictive covenants, D&O tail, regulatory status, and closing-readiness classification.
  Committee escalation memo	Filter to only the current draft terms that exceed a policy threshold requiring committee approval; exclude in-policy, stale, and non-committee terms. Provide policy comparison, delta, benchmark context, exposure, recommendation, and aggregate summary.
  Carveout transition review	Focus on separation terms: IP/domain transition, TSA scope/duration/fees, purchase-price allocation, transfer taxes, employee continuity, outside date, governing law.
  Deviation matrix	Position-by-position matrix of buyer positions on indemnity, survival, materiality scrape, escrow, consents, HSR, and material contracts, with final-position enums, closing blockers, and risk totals.

2. Read the answer template first

Every task provides an answer_template.json (or schema embedded in the prompt). Read it before gathering data. It defines:

• The exact top-level keys your answer must contain.
• The allowed enum values for every categorical field (risk_rating, issue_status, recommended_action, final_position, etc.).
• The list of valid issue_id values — use only these.
• The precision rules: integer dollars, decimal percent-points to n places, integer months, holder percentages to four decimals, dates as YYYY-MM-DD.
• Which fields are nullable vs. required in every object.

Never add keys that are not in the template. If a field is not applicable, use null (JSON null), not a placeholder string or zero — unless the template explicitly calls for a zero.

3. Gather data from the workbench

Use the documented API routes. The base URL is provided as <TASK_ENV_BASE_URL>. The minimum set of endpoints to hit for any task:

When the prompt mentions read-only SQL, use POST /api/query with the provided token for cross-table verification — but the REST endpoints above are the primary data source and are always sufficient.

4. Compare draft against playbook or policy

This is the central analytical step. For each category present in the playbook or policy, locate the corresponding draft term (match by category field).

Determine issue_status:

Seller vs. buyer orientation flips the direction of "exceeds":

• Seller-side: wants lower caps, shorter survival, less escrow, no financing condition. Draft values above the playbook preferred are draft_exceeds_playbook.
• Buyer-side: wants higher caps, longer survival, more escrow, full materiality scrape. Draft values below the playbook preferred are draft_below_playbook.

For policy tasks (committee escalation):

• Only terms whose restricted_flag is "yes" and whose draft value exceeds the threshold_value are escalated. In-policy terms, stale terms, and terms approved at a lower level are excluded. The template's excluded_in_policy_terms and excluded_in_policy_categories fields should list what you considered but did NOT escalate — this proves you reviewed everything.

Treat absent terms as issues when the data supports it: A missing term (source_term_ids: []) is an issue when the surrounding deal data shows the term is needed. Examples: employees exist but no employee-continuity term; required consents exist but no consent-closing-condition term; HSR is required but no HSR covenant.

Use stable identifiers from the workbench:

• source_term_ids must contain the actual term_id values returned by the API (e.g. "TERM_PRJ_XXXXX_NN"). Use an empty array [] only for missing terms.
• Consent IDs, employee IDs, finding IDs, contract IDs, and note IDs must match the workbench exactly.

5. Calculate amounts

Purchase price basis: Use headline_value from the deal record as the purchase-price basis for all percentage-to-dollar conversions unless the term or playbook rule explicitly states a different basis (e.g. "enterprise value", "upfront cash", "equity value").

Dollar amounts: percentage × headline_value ÷ 100, rounded to the nearest integer dollar.

draft_amount_dollars = round(draft_percent / 100 * headline_value)

Deltas: The gap between the draft and the fallback position.

delta_to_fallback_dollars = draft_amount - fallback_amount
delta_to_fallback_months  = draft_months - fallback_months   (seller: draft exceeds → positive delta)

Exposure aggregation: Sum the exposure_low and exposure_high values from /risk-estimates across the relevant categories. Use these for total_quantified_exposure_low_dollars and total_quantified_exposure_high_dollars.

Employee totals: Sum count and pto_liability across all employee groups in /employees. Count required_for_closing: "yes" consents for required_closing_consent_count.

6. Assign risk, action, and priority

Risk rating defaults to the playbook's risk_default for that category. Override based on deal-specific factors:

• HIGH: Required consents from top-revenue customers with termination leverage; terms where the draft is at the far extreme; regulatory conditions that can block closing; financing conditions without adequate break fees.
• MEDIUM: Terms where the draft exceeds playbook but a negotiated middle ground is likely; missing terms with quantifiable but manageable exposure.
• LOW: Standard terms with small deltas; governing law / forum preferences; items where counterparty leverage is weak.

Recommended action (use only values from the template's enum):

Priority order: Sort issues from highest negotiation priority (must-have protections, closing blockers, largest dollar exposure) to lowest (nice-to-have positions, small deltas). The deal notes often signal what the business team considers must-have vs. tradeable.

7. Return JSON — and only JSON

The prompt will say "Return only valid JSON" or "Do not include narrative outside the JSON." Follow this literally. Your entire response should parse as a single JSON object conforming to the answer template.

Before returning, verify:

• Every issue_id is from the template's possible_issue_ids (or stable_issue_ids).
• Every enum field uses one of the template's allowed_enums values.
• Dollar amounts are integers; percentages are decimal numbers at the specified precision.
• source_term_ids match actual term IDs from the /terms endpoint.
• priority_order contains every issue_id from issue_register exactly once.
• summary_metrics counts are consistent with the issue_register contents.
• No extra keys, no missing required keys.

Task-type quick reference

Issue register (seller or buyer APA/SPA review)

1. Get deal → terms → playbook rules.
2. For every playbook rule, find the matching draft term by category.
3. For every draft term not in the playbook, decide if it's benign or an issue.
4. For every playbook rule with no matching draft term, decide if the deal context requires it (→ missing_required_term) or if it's not applicable (→ omit).
5. Populate the percent/month/amount fields that apply; leave others null.
6. Build priority_order from highest to lowest negotiation urgency.
7. Compute summary_metrics from the register.

Closing / economics package (buyer SPA)

1. Get deal → terms → cap-table → consents → employees → material-contracts → regulatory → diligence-findings → playbook rules.
2. Economics: Distribute headline value across holders by fully-diluted percentage. Cash/stock split follows the deal's upfront_cash and stock_value.
3. Indemnity: Compare draft cap and survival against buyer playbook minimums.
4. Escrow: Determine basis, required percentage, release timeline, and trigger.
5. NWC: Check diligence findings for working-capital items; propose collar or fixed-price treatment.
6. Consents: Classify each consent as closing_condition, notice_only, or post_closing_covenant based on required_for_closing and contract type.
7. Material contracts: Same classification; attach annual_revenue.
8. Employees: Total continuing employee count, PTO liability, service-credit and WARN-risk employee IDs, required action.
9. Restrictive covenants: Whether non-compete/non-solicit is required; which holder groups must be covered.
10. D&O tail: Whether required, period, cost allocation.
11. Regulatory: HSR requirement, hell-or-high-water, closing-condition need.
12. Closing readiness: READY / READY_WITH_CONDITIONS / NOT_READY. Classification depends on blocker count and severity.

Committee escalation (public company merger, policy-driven)

1. Get deal → terms → policy thresholds → benchmarks → risk estimates → notes.
2. Filter strictly: only terms flagged restricted_flag: "yes" in the policy AND whose draft values exceed the threshold_value.
3. Exclude: in-policy terms, stale terms, non-committee categories, and distractor terms (similar names, different deals).
4. For each escalated term, build the full comparison: draft_metric, policy_metric, delta, benchmark, exposure, recommendation, required_conditions.
5. Benchmark position: Compare the draft value to the benchmark median and upper quartile. Classify as at_or_below_median, between_median_and_upper_quartile, at_upper_quartile, or above_upper_quartile.
6. Aggregate summary: Include both the escalated count and explicit lists of what was excluded — the committee needs to know you reviewed everything.

Carveout transition review (seller APA)

1. Get deal → terms → playbook rules → employees → consents → material-contracts → regulatory → benchmarks → risk estimates.
2. Focus on the eight standard transition issues: customer-consent termination risk, employee continuity/PTO, governing law/forum, IP/domain transition, outside-date extension, Section 1060 allocation, transfer-tax split, TSA scope/duration/fees.
3. For each issue, capture both the transition_issues entry (analysis) and the corresponding required_redlines entry (the specific contractual fix).
4. draft_value_normalized and required_position_normalized capture the comparison in a structured form; use the same units and shape for both.
5. operational_risk aggregates exposures: stranded cost gap, PTO liability, consent amount at risk, top-customer annual revenue at risk, transition disruption high estimate.
6. Map each exposure to a business_outcomes_protected value.

Deviation matrix (buyer SPA)

1. Get deal → terms → playbook rules → consents → material-contracts → regulatory → diligence findings → benchmarks → risk estimates.
2. The seven standard positions: indemnity cap & basket, survival & knowledge, materiality scrape, escrow/holdback/release, consent closing condition, HSR condition, material contracts.
3. For each position, provide draft_percent/draft_months/draft_amount_usd, the corresponding preferred_* and fallback_* values, and the shortfall_to_fallback_usd / shortfall_to_preferred_usd.
4. final_position must be one of the template's position enum values.
5. Closing blockers: List every consent, regulatory clearance, or material contract that must be satisfied before closing. Each gets a blocker_type, must_be_satisfied_before_closing: true, and a risk rating.
6. Risk totals: Aggregate across all positions. indemnity_cap_shortfall_to_fallback_usd is the gap between the draft cap amount and the buyer's fallback cap amount. total_modeled_exposure_low_usd and total_modeled_exposure_high_usd sum the risk-estimate ranges for relevant categories.

Common pitfalls

1. Wrong basis for percentage calculations: Always check the basis field on each term and playbook rule. Most are "purchase price", but some are "enterprise value", "equity value", or "upfront cash". Using the wrong basis produces wrong dollar amounts.

2. Confusing percent values with decimal fractions: The workbench uses percentage-point notation. A draft cap of 18% is 18.0, not 0.18. Percentages round to the decimal places specified in the template's units.

3. Including stale or distractor terms: Check staleness_flag on draft terms; stale terms should not be treated as current draft positions. Similarly, playbook rules and policy thresholds from unrelated deal types should not be applied.

4. Omitting the "excluded" lists in committee memos: The template has explicit fields for excluded_in_policy_terms and excluded_in_policy_categories. Populate these — they demonstrate completeness.

5. Skipping missing terms: When the playbook requires a provision and the draft is silent, this is a missing_required_term issue. Do not omit it just because there is no source_term_id to cite.

6. Mismatched priority_order: Every issue_id in issue_register must appear in priority_order exactly once, and vice versa. The lengths must match.

7. Integer vs. float for dollars: Dollar amounts must be integers. Round to the nearest dollar, not to thousands or millions.

8. Inventing IDs: Never fabricate term IDs, consent IDs, employee IDs, or other stable identifiers. Use only the values returned by the workbench APIs.