memo-builder

# Memo Builder ## Skill Configuration ### Common Skill Instructions MANDATORY: Before searching connectors, retrieving evidence, or drafting output, read and apply the shared runtime contract in `../public-equity-investing/SKILL.md## Cross-Skill Runtime Contract`. Then check the router skill map and `../../shared/plugin-routing-playbook.md` for adjacent skills that should be sequenced with this workflow. Do not run user-context setup or inspection during ordinary workflow work; route only explicit saved-context, source-setup, onboarding, or automation-setup requests to `../user-context/SKILL.md`. ### Source Resolution Load `../../shared/workflow-source-resolution.md`. Resolve only the categories needed for this workflow: `company_filings_ir`, `earnings_transcripts_presentations`, `internal_research`, `portfolio_models_trackers`, and `market_data_estimates`. Use the shared runtime contract to map each attempted category to an available app, connector, file, export, or user-provided input. ## Relevant Dependency Categories These are the source categories most likely to matter for this workflow. Use the router contract to resolve only the categories the task actually needs, prefer user-named sources first, and state any material source limitation. - `Internal Research` - `Portfolio Models & Trackers` - `Market Data & Estimates` - `Company Filings & IR` - `Earnings Transcripts & Events` ## Deliverable Intake Before source gathering or analysis for a new substantive hero deliverable, load `../../shared/deliverable-intake-policy.md` and use its adaptive `request_user_input` preflight for materially unresolved format, depth, audience/use, or focus choices. Reuse resolved preferences in downstream steps; when acting only as input to an owning workflow, do not re-prompt. ## Goal Produce decision-grade Public Equity Investing memos that are evidence-led, numerically auditable, and ready for PM, committee, client, or internal research review. This skill owns formal written artifacts and synthesis. It does not own raw trade construction; use `long-short-pitch` first when the user needs variant perception, trade expression, sizing considerations, add/trim/exit/cover rules, or pair-trade mechanics. ## Use When Use this skill when the user asks for: - IC memo, investment memo, committee note, client memo, research note, PM update, or equity event note - memo rewrite, upgrade, shortening, pressure test, or QA review - synthesis of outputs from models, comps, scenarios, earnings, event-driven equity, balance-sheet-risk, or long/short pitch work - committee-ready framing with source posture, valuation, scenarios, risks, catalysts, open items, and recommendation Credit-first memos route out: Use Credit Markets for credit instruments, creditworthiness, restructuring, distressed, recovery, spreads, yields, covenants, and debt security analysis. This skill may synthesize Credit Markets output only when the final deliverable is a common-equity or listed-equity memo. Do not use when the user asks to "pitch this long/short," "make this investable," "build a pair trade," "what is the variant perception," or "what would make us cover/add/trim." Use `long-short-pitch`. ## Artifact Boundary - Default deliverable: produce the full decision-grade memo in the selected or high-confidence inferred surface. For formal investment committee, PM, committee, client, or research memo circulation, use a real Word document (`.docx`) as the normal recommended surface. For high-confidence source-heavy web-style investment reports, use a polished standalone HTML report following `../../shared/html-artifact-standard.md`. For quick or conversational memo updates, inline response can be the recommended surface. - If the surface is semi-ambiguous, use `../../shared/deliverable-intake-policy.md` before drafting and let the chosen or timeout-resolved option control the hero deliverable. - Chat can be the presentation surface for narrow or conversational memo requests, but Markdown is formatting, not the deliverable contract. - This skill ships `scripts/build_memo_package.py`, which can generate a real DOCX memo and an HTML companion, or HTML as primary with a DOCX companion. If DOCX is selected, generate the full memo in DOCX; do not treat DOCX as a thin companion to HTML. - Route workbook-backed outputs to the model/workbook skills and create `.xlsx` artifacts only when the user selects or clearly implies a model, tracker, valuation, or workbook surface. - Use `dashboard-builder` only when the user explicitly asks for a standardized dashboard, reusable dashboard template, PM cockpit, or structured payload-driven render. On that path, load `references/DASHBOARD_PACK.md`; `memo-builder` owns memo synthesis, narrative architecture, evidence selection, and audience framing, while `dashboard-builder` owns the standardized shell/rendering/QA. Build a `public_equity_investing_dashboard.v1` payload as an internal renderer input. ## Source And Mode Selection Use the strongest available source tier and label assumptions, stale data, proxies, and unsupported figures. Load `references/source-policy.md` for source priority and minimum inputs. Default to the full memo mode that best fits the decision. Use a compressed mode only when the user explicitly asks for a `summary`, `short`, `quick`, `one-page`, `brief`, `TL;DR`, or review-only output: - `ic-note` - `investment-memo` - `event-driven-committee-note` - `pm-update` - `client-research-note` - `qa-review` Load `references/memo-modes.md` for mode selection and `references/output-contracts.md` for section-level contracts. ## When To Invoke Support Load `shared/support-layer-routing-contract.md` for source/data/QC/style support. Use `financial-source-of-truth` before final claims, metrics, citations, or management assertions. Use `financials-normalizer` before building from messy financials, consensus exports, guidance tables, segment data, share count, net debt, or capital allocation inputs. Use `excel-data-cleaner` before relying on malformed tables or mixed fiscal periods. Use `deck-report-qc` before circulating a memo-derived deck/report or source-heavy packet, and use `style-guide-adapter` only after substance is locked. These support artifacts stay secondary; `memo-builder` owns the memo and investment judgment. ## Sub-agent decomposition For complex medium/large requests, use sub-agents where available; otherwise emulate the split as named workstreams. Suggested lanes: evidence, model/valuation, risks and catalysts, narrative synthesis, and memo QA. Keep this skill as the lead: reconcile conflicts, source labels, assumptions, open items, final QA, and the user-facing answer. ## Default Memo Spine Use this full spine unless another memo mode is clearly requested: 1. `Recommendation / Decision Ask` 2. `Executive Summary` 3. `Thesis and Evidence` 4. `What Must Be True` 5. `Valuation / Scenario Work` 6. `Risks, Disconfirmers, and Mitigants` 7. `Catalysts and Monitoring` 8. `Implementation Considerations` 9. `Open Items / Data Requests` If the memo is screen-grade because source context is thin, say so before the first metric table. ## Handoffs Upstream inputs this skill can synthesize: - source/evidence: `financial-source-of-truth`, `financials-normalizer`, `company-tearsheet` - models/valuation: `equity-model-update`, `dcf-model-builder`, `three-statement-model-builder`, `comps-valuation`, `scenario-sensitivity-generator` - earnings/research: `earnings-preview`, `earnings-deep-dive`, `initiating-coverage` - trade/event/risk: `long-short-pitch`, `event-driven-analyzer`, `portfolio-risk-management` - monitoring and QC: `thesis-tracker`, `catalyst-calendar`, `deck-report-qc`, `style-guide-adapter` Use `sector-context-overlay` only when issuer economics clearly match a supported sector; load `references/sector-overlays.md` only for memo-specific sector framing. ## Rendering And Compression - Lead with the decision ask or conclusion. - For a standalone HTML investment committee memo, make the first-read layer a compact memo opening: recommendation / decision ask, decision hinge, what is priced in, valuation or scenario skew, and any source limitation that changes the action. Do not repeat the same recommendation across a dashboard-style hero, summary tile, and full decision panel. - Keep background proportionate; preserve recommendation, scenario/downside, source posture, catalysts, monitoring, and open items. - Use tables only when they compress decision-critical information. - For long memos, put source caveats and open items near the end, but do not bury data limitations that affect the recommendation. - Sparse source context should produce a full screen-grade memo skeleton with assumptions, missing evidence, and upgrade path, not a shortened memo. - State source readiness in decision-specific terms when evidence supports only one direction of action, such as `sufficient to decline initiation today; insufficient to support initiation`, rather than using broad readiness labels that imply the entire underwrite is decision-grade. - Keep citations traceable but readable in HTML: do not fragment tickers, prices, EPS ranges, percentages, dates, multiples, or metric labels into separately linked tokens. - Before delivering standalone local HTML, visually inspect the opening viewport and decision-critical downstream sections through local headless-browser screenshots, not the in-app Browser plugin. Check desktop and narrow-screen legibility, clipping, citation density, whitespace, and that the page does not acquire horizontal overflow. - When delivering DOCX, ensure the document itself contains the full memo spine, including recommendation, decision hinge, thesis/evidence, valuation or scenario view, risks, catalysts, source posture, and open items. The HTML report may be a companion, but the DOCX must be usable as the first-read artifact on its own. ## QA Before finalizing, verify: - recommendation and decision hinge are clear - material numbers are sourced, model-derived, or explicit assumptions - downside is mechanistic, not generic - `what must be true` items are measurable - valuation/pricing/recovery support matches memo type - a multi-year valuation using forward-period earnings and an exit or terminal multiple shows either present value using an explicit discount-rate / required-return assumption or annualized return / IRR versus a stated hurdle; undiscounted terminal price appreciation alone does not support initiation - disconfirmers and monitoring triggers are present - trade-construction asks are handed to `long-short-pitch` Load `references/quality-workflow.md` for the full quality bar and workflow. ## Script Map When deterministic packaging from structured memo inputs is appropriate, use `scripts/build_memo_package.py --primary-format docx` or `scripts/build_memo_package.py --primary-format html` according to the selected memo surface. The script writes both reader-facing surfaces plus a manifest; the selected primary format controls the manifest's first-read artifact. ## Public Equity PM Judgment Layer For substantial memos, load `shared/pm-judgment-heuristics.md` before finalizing. Audience modes: `long_only_pm`, `long_short_hf`, `sell_side_research`, `etf_index_diligence`, `public_equity_diligence`. Memo modes to support: `buy-side-investment-memo`, `pm-update`, `sell-side-research-note`, `client-research-note`, `etf-index-diligence-note`, and `public-equity-diligence-memo`. Required PM judgment: - Every full memo needs a decision hinge, what must be true, what is priced in, downside mechanism, measurable disconfirmers, source posture, and action discipline. - Build-from-scratch memo mode must include an intake checklist, source packet requirements, first-pass house view, variant wedge, priced-in debate, estimate path, valuation/skew, downside mechanism, catalysts, disconfirmers, action rules, and exact evidence needed to upgrade from screen-grade to decision-grade. - Sell-side notes need rating/target-price debate, estimate revision bridge, variant versus Street, risk-to-rating, and compliance-safe wording. - ETF/index notes need mandate, methodology, holdings/weights, tracking-error or benchmark implications, factor exposure, liquidity, and rebalance/corporate-action risk.