git:20260614.5208b4c to git:20260615.1734111

- description: "Business-model analysis is a structural snapshot. Default lookback is 1 fiscal quarter (the most recent disclosure). Max 8 quarters when analyzing channel mix evolution or management changes over a 2-year window."

- !curl -s -o /dev/null -w "%{http_code}" --max-time 2 https://mcp.agentii.ai/mcp/health 2>/dev/null || echo "UNREACHABLE"

-

- **Ticker resolution **: Before any data retrieval, resolve the ticker via the three-layer fallback per retrieval.md Pre-Flight Step 0: (1) exact match via `search_companies(ticker=<input>)`, (2) pg_trgm fuzzy alias match via `gold.entity_aliases` (6,721 rows), (3) share class normalization for multi-class tickers (GOOG/GOOGL→GOOG, BRK.A/BRK.B→BRK.B). Return canonical ticker, match method, and confidence indicator.

-

- **Workspace style.md override check **: Check `./style.md` in the workspace root for per-workspace overrides (`default_lookback_quarters`, `reporting_currency`, `sector_focus`, `output_verbosity`, `peer_universe`). Apply overrides to output formatting and temporal scope. Precedence: workspace `style.md` > package `style.md` > skill defaults.

-

- **Agent Call Tracing**: The first tool you call will return a `_run_id` in its result. On every subsequent tool call, include HTTP header `X-Agentii-Trace: agent={skill_name}; parent={caller_name}; instance={instance_label}`. The MCP server will inject run_id, depth, and user_id automatically. When spawning parallel sub-agents of the same type, assign each a unique instance label (e.g., equity-research-1, equity-research-2). See `contracts/x-agentii-trace-header.md` for the full contract.

- | lookback_quarters | 1 | Single most-recent quarter snapshot for structural analysis |

- The Neon production database and `api.agentii.ai` REST/MCP surfaces are LIVE and AUTHORITATIVE as of 2026-05-25. All retrieval planning MUST treat these as ground truth. Production scale: 4.17M `gold.xbrl_facts` (with `is_primary` partial index), 11,575 `pipeline.src_documents` (100% non-null `description`, GIN-indexed `secondary_labels`), 243K `pipeline.src_silver_pages` (covering ALL 5 SEC form types — 8-K/10-K/10-Q/6-K/20-F; `labels` JSONB with `general.description` + `general.keywords`), 4,653 `pipeline.earnings_calendar` rows, 79 `gold.launch_ticker_registry` tickers at 100% processing. Always call `get_ticker_coverage/{ticker}` before retrieval planning.

- Follow the retrieval strategy decision tree in `retrieval.md`:

-

- - **Branch (a)** — `search_xbrl_facts` for revenue concentration, gross margin profile, segment-level P&L (concepts `Revenues`, `GrossProfit`, `OperatingIncomeLoss`, `SegmentReportingInformation`).

- - **Branch (b)** — `search_cross_period` when analyzing channel-mix evolution over 2+ years (mode 1_2) or management-change continuity (mode 1_5).

- - **Branch (c)** — single-period three-layer for the most recent 10-K/10-Q business-overview pages (modes 1_1, 1_3, 1_4).

-

- **Layer 1 narrowing — `secondary_labels` filter **: For 8-K-driven business-model-relevant disclosures, prefer `?secondary_label=other_events_8_01` (item 8.01 covers business-strategy events) AND `?secondary_label=financial_results_2_02` (item 2.02 surfaces segment-level revenue commentary). The GIN-indexed filter on `pipeline.src_documents.secondary_labels` is the preferred narrowing axis BEFORE Layer 2.

-

- **Layer 2 page-relevance signal**: Score pages using `labels->>'general'->>'description'` (~100-char LLM-generated page summary) AND `labels->>'general'->>'keywords'` (extracted entity terms). Both fields are populated on 96%+ of `pipeline.src_silver_pages` rows. For business-model analysis, prefer pages whose `keywords` contain entity terms like product names, segment names, geographies, channel partners, executive names.

- Default: 1 fiscal quarter (max 8). Business-model analysis is a structural snapshot — historical lookback only when explicitly tracking channel-mix evolution (mode 1_2 default 12 quarters) or management changes (mode 1_5 default 4 quarters).

- See frontmatter `allowed_tools` — 12 tools declared:

- 1. ** Pre-flight (mandatory)**: call `get_company_fiscal_calendar/{ticker}` for fiscal orientation, then `get_ticker_coverage/{ticker}` to discover which data sources are populated. Route based on coverage: `sec_filings` populated → standard three-layer protocol; `xbrl_facts`-only → structural inferences from segment data only, flag `data_availability: degraded`.

-

- 2. **Layer 1 Document Discovery**: `search_documents(ticker={T}, form_type=["10-K","20-F"], limit=3)` to find the most recent annual report (richest business-overview content). Add `?secondary_labels=financial_results_2_02,other_events_8_01` to also surface relevant 8-Ks. Document identifiers returned in `citation_id` form (e.g., `sec135`) — pass these directly to Layer 2.

-

- 3. **Layer 2 Page Map**: `read_source_outline/{ticker}/{citation_id}` — scan `description` + `keywords` for each page. Identify pages covering: Business overview / Item 1 (mode 1_1), Distribution & sales channels (mode 1_2), Revenue by segment & geography (mode 1_3), Industry / market context (mode 1_4), Directors & executive officers (mode 1_5). Bare `page_no` integers are forbidden in LLM-facing output — always use `{ticker} {citation_id} page<N>` format (e.g., "LLY sec135 page12").

-

- 4. **Layer 3 Deep Read**: `read_source_pages/{ticker}/{citation_id}?row_numbers=page<N1>,page<N2>` — load full `page_content` for ONLY the 3-5 pages identified in Layer 2.

-

- 5. **XBRL retrieval**: `search_xbrl_facts(ticker, concept=["Revenues","GrossProfit","OperatingIncomeLoss"], fiscal_year=[<latest>])` — returns `is_primary: true` rows by default ( superseded; `?include_all_sources=true` only for audit). The `source_authority` field (3=10-K, 2=10-Q, 1=8-K) is returned for fact-provenance transparency.

-

- 6. **Evidence-pack handoff**: produce `evidence-pack.json` + `evidence-digest.md` . All citations use the v1.0 frozen format with `{ticker} {citation_id} page<N>` references.

- ### Mode: business-model-classification (1_1 — anchor)

-

- **Display name**: Business Model & Offerings Classification

-

- <!-- ported_from: references/prompts/1/1_1.md -->

-

- **Objective**: Determine business-model type (product / service / platform), core offering, and market positioning (low-end / mid-tier / high-end) using the most recent quarter's disclosures, sell-side research, and media sentiment.

-

- **Output structure**:

-

- - **Business Model**: [Product / Service / Platform]

- - **Core Offering**: [e.g., Connected Wearable, Diagnostic Consumables, SaaS Subscription, Drug Pipeline, Cloud Platform]

- - **Positioning**: [Low-end / Mid-tier / High-end] + brief rationale (e.g., "High-end based on >70% gross margin and premium ARPU vs. peers")

- - **Citation density**: ≥1 citation per 200 words, format `{ticker} {citation_id} page<N>`.

-

- ### Mode: distribution-channel-analysis (1_2)

-

- **Display name**: Distribution Channels & Go-to-Market Analysis

-

- <!-- ported_from: references/prompts/1/1_2.md -->

-

- **Objective**: Assess primary distribution model (direct sales / channel partners / hybrid), channel mix evolution (3-year trailing), and strategic implications for pricing power and customer intimacy.

-

- **Output structure**:

-

- - **Distribution Model**: [Direct Sales / Channel Partners / Hybrid]

- - **Distribution Partners**: [list disclosed channel types and representative partners]

- - **Current Channel Mix**: Direct : Indirect = 1 : XX (latest available data)

- - **Historical Channel Mix Trend (Trailing 3 Years)**:

- - Year -2: [Direct : Indirect = 1 : XX]

- - Year -1: [Direct : Indirect = 1 : XX]

- - Current: [Direct : Indirect = 1 : XX]

- - **Strategic Implication**: e.g., "Shift toward direct sales has enhanced pricing control and customer intimacy but increased SG&A".

-

- ### Mode: revenue-composition-and-concentration (1_3)

-

- **Display name**: Revenue Composition & Concentration Risk Analysis

-

- <!-- ported_from: references/prompts/1/1_3.md -->

-

- **Objective**: Decompose revenue by product line / customer type / geography / end market, identify concentration risk (any single product or client >20% of total revenue), and trace temporal mix evolution.

-

- **Output structure**:

-

- - **Latest Quarter Revenue Breakdown**:

- - By Product Line: top 3 contributors with %

- - By User Type (B2B vs. B2C): mix with %

- - By Geography (NA / EMEA / APAC): top 3 regions with %

- - By End Market: top 3 markets with %

- - **Concentration Risk Matrix**: any product/client >20% flagged.

- - **Temporal Comparison**: vs. prior quarter and YoY same-period.

-

- ### Mode: market-sizing-and-relative-growth (1_4)

-

- **Display name**: Market Sizing (TAM/SAM/SOM) & Relative Growth

-

- <!-- ported_from: references/prompts/1/1_4.md -->

-

- **Objective**: Quantify TAM, SAM, SOM for key verticals, project 3-5 year CAGR, assess relative growth (company vs. addressable market) and historical SOM evolution.

-

- **Output structure**:

-

- - **Market Sizing & Growth**:

- - TAM (Current Year): USD XXX bn

- - SAM (Current Year): USD XXX bn

- - SOM (Current Year): XX%

- - TAM CAGR (Past 3 Years): XX%

- - TAM CAGR (Forward 3-5 Years): XX%

- - **Relative Growth Table**: Company Revenue Growth vs. TAM CAGR for past 3Y and next 3Y.

- - **Historical SOM Trajectory**: 3-year evolution with execution-strength assessment.

-

- ### Mode: management-and-leadership (1_5)

-

- **Display name**: Management Team & Leadership Analysis

-

- <!-- ported_from: references/prompts/1/1_5.md -->

-

- **Objective**: Assess executive team backgrounds, track records, and recent leadership changes (CEO/CFO/COO/CMO) for strategic-execution implications.

-

- **Output structure**:

- - **Key Executives & Track Record**: [Name, Role, Tenure, Notable Prior Experience, Industry Expertise, Capital Allocation Record]

- - **Recent Management Changes (Trailing 1-2 Quarters)**: [Name, Role, Effective Date, Reason, Successor Background]

- - **Strategic Implications of Changes**: e.g., "New CFO brings strong M&A background, suggesting a shift toward inorganic growth".

- | Tool | Failure Mode | Fallback Action | Coverage Annotation |

- |------|-------------|-----------------|---------------------|

- | `read_source_pages` | SQL error / PROXY_ERROR | Use `search_keyword_in_source(document_id, keyword)` if document_id known; otherwise `search_documents` with same query | "source file unavailable; used keyword search instead" |

- | `read_source_deep_outline` | PROXY_ERROR / 404 | Use lightweight `read_source_outline` and flag `deep_outline_degraded: true` | "deep outline unavailable; used lightweight page map instead" |

- | `read_source_outline` | PROXY_ERROR / 404 | Use `list_sources` for document-level metadata | "page map unavailable; used document listing instead" |

- | `list_xbrl_concepts` | Timeout / 503 | Use direct `search_xbrl_facts` with standard US-GAAP concepts (Revenues, GrossProfit, OperatingIncomeLoss) | "concept discovery skipped due to timeout; using standard US-GAAP concepts" |

- | `get_company_fiscal_calendar` | Cross-validation failed | Use XBRL-derived period grid from `search_xbrl_facts` `period_end` dates | "fiscal calendar mismatch; using XBRL-derived period grid" |

- | `search_xbrl_facts` | Empty (concept not populated) | Fall back to narrative analysis from 10-K Item 1 (Business Overview) | "segment-level P&L not in XBRL; structural inference only" |

- | `batch_search` | PROXY_ERROR | Use sequential individual calls (one per sub-query) | "batch search unavailable; used sequential calls" |

-

- Tool errors are retried ONCE with the fallback action before escalating to the retrieval gaps failure policy. If both Layer 2 and Layer 3 tools are unavailable, enter document access degradation mode (structured data + metadata only, flag output as degraded).

- Write the final deliverable to `{{ticker}}/{{YYYY-MM-DD_HHMM}}_business-model_product-line-decomp.md` .

- ## Output Structure

- The final deliverable MUST be written as a markdown file to the workspace using the convention :

- ```

- {ticker}/{YYYY-MM-DD_HHMM}_business-model_{affix}.md

- ```

- Where `affix` is a short descriptive slug (e.g., `product-line-decomp`, `channel-mix`, `market-sizing`, `management-changes`). Examples:

- - `LLY/2026-05-25_1430_business-model_product-line-decomp.md`

- - `NVDA/2026-05-25_1545_business-model_platform-classification.md`

- The deliverable file MUST contain (in order):

- 1. **Executive Summary** (≤200 words) — business-model classification + headline structural insights.

- 2. **Business Model Type** (mode 1_1) — Product / Service / Platform classification with rationale.

- 3. **Product Line Decomposition** (mode 1_3) — revenue breakdown by product, top-3 contributors, concentration risk.

- 4. **Distribution Channel Analysis** (mode 1_2) — direct vs. indirect mix, partners, 3-year evolution.

- 5. **Customer Segment Analysis** (mode 1_3) — B2B vs. B2C, geography, end markets, concentration risk.

- 6. **Revenue Model** — recurring vs. one-time, pricing power, unit economics, gross-margin profile.

- 7. **Business Unit Performance** — segment-level P&L where available (XBRL or narrative).

- 8. **Market Sizing & Competitive Positioning** (mode 1_4) — TAM/SAM/SOM, relative growth vs. market.

- 9. **Management & Leadership** (mode 1_5) — executive team, recent changes, strategic implications.

- 10. **Coverage Gaps & Citations** — list of dimensions not retrievable + full citation index in `{ticker} {citation_id} page<N>` format.

- **Citation density**: ≥1 citation per 200 words. Bare `page_no` integers are forbidden — always use `{ticker} {citation_id} page<N>`. **Citation link format **: use clickable links: `[📄 {ticker} {form_type} p.{N}](https://agentii.ai/v/{ticker}/{citation_id}/{N})`. Example: `[📄 LLY 10-K p.42](https://agentii.ai/v/LLY/sec175/42)`.

- **agentii.md append **: After writing the output file, append a YAML block to `agentii.md` at the workspace root with `ticker`, `date`, `skill`, `output_file`, and `key_conclusions`. Create the file with a `# Project Memory Index` heading if it doesn't exist. See `contracts/agentii-md-schema.md`.