stock-analyser

Stock Analyser

Use this skill to generate a deep, source-backed, insight-rich fundamental analysis of an Indian listed stock for long-term investors. The final output must be a single self-contained Markdown document ready to save as an .md file.

Operating objective

Produce a high-quality Indian stock research note covering: business quality, growth, financial health, returns, valuation, peers, ownership and governance, sector-cycle/macro sensitivity, liquidity, and clear pros/cons. The skill must emphasize insight, not just data collection.

Non-negotiable rules

Use Indian market context only.
No price prediction, no target price, no buy/sell/hold language, and no recommendation of any kind.
Do not fabricate any number. If unavailable, write DATA UNAVAILABLE.
Every material metric must cite a source such as NSE, BSE, Screener.in, annual report, earnings transcript, investor presentation, credit rating note, or credible financial news.
Prefer live web data. If live data is unavailable, clearly state DATA UNAVAILABLE or Figures from limited context; verify independently on NSE/BSE/Screener.in.
Every key financial metric must be cross-referenced from at least 2 independent sources during research. If only 1 source is available, mark it [single-source] in the report. If 0 sources, mark it DATA UNAVAILABLE.
Separate verified facts from interpretation.
Keep the tone factual, analytical, and long-term oriented.
Scenario analysis may be included only as historical-trend-based illustration, not as a forecast.
Do not use vague hype words like "multibagger", "sure winner", "must buy", or "hidden gem".
Do not hide uncertainty. Missing or weakly supported points must be called out explicitly.
You MUST save the final report as a .md file using the Write tool (or platform equivalent). Printing the report to chat without saving it to a file is a skill violation. The file name must follow the pattern {TICKER}_{YYYY-MM-DD}.md in the current working directory.
You MUST use parallel research agents for data collection (Phase 1). Do not collect all data sequentially in the main agent. Single-agent data collection is a skill violation.
Every metric in the final report must carry its as-of date or reporting period. Do not present numbers without time context. See "Data freshness" section for format.

Trigger behavior

When invoked, ask exactly one question and wait:

Which stock? Company name or NSE/BSE ticker.

Do not begin analysis until the answer is provided. If the stock identifier is ambiguous, ask a clarifying question before research.

Input contract

Required: stock_identifier (company name or NSE/BSE ticker).

No other inputs. The skill always analyses across two fixed horizons — 5 years and 10 years. This ensures both medium-term trajectory and long-term structural risks are covered. The user does not choose a horizon — the skill reveals what both timeframes show.

Research workflow

Research is a phased funnel. Each phase gates the next. Do not skip ahead.

Preferred sources

Use these in priority order. Higher-ranked sources win when numbers conflict.

NSE India, BSE India (official filings, shareholding, corporate actions)
Screener.in (consolidated financials, ratios, peer sets)
Tickertape (valuation, technicals, mutual fund holdings)
Moneycontrol (earnings, news, analyst commentary)
Annual reports, investor presentations
Earnings call transcripts
Credit rating agencies (CRISIL, ICRA, CARE, India Ratings)
Reputable financial news (ET, Livemint, Business Standard)

Source fetching guide

Do not blindly web-search "NSE INFY financial data". Use direct URL patterns and API endpoints. This saves search calls and returns structured data.

Verified source availability (tested April 2026):

Source	Fetch method	Notes
NSE India	Works — via Bash curl with session cookies	Official source. Requires 2-step cookie flow. See recipe below.
Screener.in	Works — via WebFetch	Predictable URLs, rich consolidated data
Tickertape	Works — via WebFetch	Slug varies, may need search
Moneycontrol	Unpredictable URLs	Always needs web search + WebFetch

NSE India (primary — official data via curl)

NSE uses Akamai bot protection. WebFetch will not work. Use the Bash tool with curl instead. The flow requires two steps: establish a session, then hit API endpoints.

Critical curl requirements (all three are mandatory):

--compressed flag — NSE sends gzip-encoded responses
Accept-Encoding: gzip, deflate — do NOT include br (brotli). Most curl builds lack brotli support and will receive garbled data.
Session cookies — first request to https://www.nseindia.com establishes cookies, subsequent API calls reuse them.

NSE curl template — base form for both steps:

curl -s --compressed \
  -b /tmp/nse_cookies.txt -c /tmp/nse_cookies.txt \
  -H 'User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/146.0.0.0 Safari/537.36' \
  -H 'Accept: application/json, text/javascript, */*; q=0.01' \
  -H 'Accept-Language: en-IN,en-GB;q=0.9,en-US;q=0.8,en;q=0.7' \
  -H 'Accept-Encoding: gzip, deflate' \
  -H 'Referer: https://www.nseindia.com/get-quotes/equity?symbol={SYMBOL}' \
  -H 'Sec-Ch-Ua: "Chromium";v="146", "Not-A.Brand";v="24", "Google Chrome";v="146"' \
  -H 'Sec-Ch-Ua-Mobile: ?0' \
  -H 'Sec-Ch-Ua-Platform: "macOS"' \
  -H 'Sec-Fetch-Dest: empty' \
  -H 'Sec-Fetch-Mode: cors' \
  -H 'Sec-Fetch-Site: same-origin' \
  'https://www.nseindia.com{ENDPOINT}'

Step 1 — Session init (run once): Remove -b flag (keep only -c), add -o /dev/null, change Accept to text/html,application/xhtml+xml,application/xml;q=0.9,image/avif,image/webp,image/apng,*/*;q=0.8, change Sec-Fetch-Dest to document, Sec-Fetch-Mode to navigate, replace Sec-Fetch-Site with none, add Sec-Fetch-User: ?1 and Upgrade-Insecure-Requests: 1, remove Referer, set URL to https://www.nseindia.com.

Step 2 — API calls: Use the template as-is, substituting {ENDPOINT} and {SYMBOL}.

Verified working endpoints (tested April 2026 with TCS):

Endpoint	Data returned	Use for
`/api/quote-equity?symbol={SYMBOL}`	CMP, open/high/low/close, 52w range (with dates), P/E, sector P/E, P/B, market cap, face value, ISIN, industry, all index memberships, pre-open data	Primary price & valuation
`/api/quote-equity?symbol={SYMBOL}&section=trade_info`	Total traded volume, traded value (₹ Cr), total market cap (₹ Cr), free-float market cap (₹ Cr), impact cost, delivery %, daily & annual volatility, VAR margin	Liquidity analysis — impact cost and delivery % are unique to NSE
`/api/equity-meta-info?symbol={SYMBOL}`	Company name, ISIN, industry, F&O eligibility, SLB eligibility, listing date, active series	Entity resolution — confirms ticker, F&O status
`/api/NextApi/globalSearch/equity?symbol={QUERY}`	Search results with symbol, company name matches	Ticker resolution from user input
`/api/NextApi/apiClient/GetQuoteApi?functionName=getSymbolData&marketType=N&series=EQ&symbol={SYMBOL}`	Order book, CMP, OHLC, 52w range, total market cap, free-float market cap, face value, impact cost, delivery %, traded volume/value	Richest single endpoint — superset of quote-equity + trade_info
`/api/NextApi/apiClient/GetQuoteApi?functionName=getMetaData&symbol={SYMBOL}`	Company name, symbol, ISIN, industry, listing date, F&O status	Metadata
`/api/NextApi/apiClient/GetQuoteApi?functionName=getYearwiseData&symbol={SYMBOL}EQN`	Price change %: 1W, 1M, 3M, 6M, 1Y, 2Y, 3Y, 5Y + same for benchmark index	Historical price performance — vs Nifty 50
`/api/NextApi/apiClient/GetQuoteApi?functionName=getIndexList&symbol={SYMBOL}`	All indices the stock belongs to (Nifty 50, Nifty IT, etc.)	Index membership for liquidity classification
`/api/NextApi/apiClient/GetQuoteApi?functionName=getSymbolChartData&symbol={SYMBOL}EQN&days=1D`	Intraday price data points	Chart data (optional)
`/api/equity-stockIndices?index=NIFTY%20IT`	All constituents of the index with their last price, change %, P/E, 52w range	Sector P/E comparison — every stock in the index

Note on identifier format: Some NextApi endpoints use {SYMBOL}EQN (e.g., TCSEQN). This is {NSE_SYMBOL}{SERIES}{MARKET} where series is EQ and market is N (NSE).

Not available via API (use Screener.in/Tickertape instead):

Shareholding pattern (NSE corporate filing pages return HTTP2 errors as of April 2026)
Financial results (quarterly/annual P&L — not exposed via API)
Board meeting filings
Corporate actions history

If the session expires (API returns empty or HTML instead of JSON), re-run Step 1 to get fresh cookies. Sessions typically last 5-15 minutes.

Screener.in (secondary — fills gaps NSE doesn't cover)

Predictable URLs via WebFetch — no search needed:

Main: https://www.screener.in/company/{SYMBOL}/consolidated/ (use /consolidated/ by default; drop it for standalone-only companies)
Peers: append #peers to the above URL

Contains: P/E, P/B, market cap, dividend yield, ROCE, ROE, face value, book value, 10Y financials (revenue, profit, EPS, OPM), quarterly results, balance sheet, cash flow, shareholding, peer comparison, segment revenue breakdown. Primary source for multi-year financials, quarterly EPS, balance sheet, cash flow, ROE/ROCE, shareholding trends, and peer comparison — data NSE's API does not expose.

Tickertape (tertiary — cross-reference)

Always resolve the slug first — do not guess it. Web-search site:tickertape.in {NSE_SYMBOL} stock to get the canonical URL, then fetch that URL. The slug format ({company-name}-{SYMBOL}) is inconsistent across stocks and guessing will produce 404s.

Contains: current price, P/E, P/B, EV/EBITDA, market cap, dividend yield, free cash flow, peer comparison with sector averages, mutual fund holdings count, shareholding pattern, 52-week range.

Moneycontrol (for qualitative data)

URL unpredictable — always web-search: site:moneycontrol.com {COMPANY NAME} stock overview. Best for: earnings call transcripts, management commentary, news, analyst estimates — qualitative data not on NSE/Screener.in/Tickertape.

General fetching rules

NSE curl: official price, valuation, liquidity (most authoritative). Screener.in: multi-year financials, ratios, shareholding, peers. Tickertape: cross-ref valuation and MF holdings. Moneycontrol: qualitative data (transcripts, commentary, news).
Record the actual URL or endpoint fetched as the source citation — not just "NSE India".

Phase 0 — Entity resolution (main agent, mandatory first step)

Before any data collection, resolve and lock the entity:

Establish an NSE session first (run the Step 1 curl from the Source fetching guide). Then resolve the ticker:
- If user gave a ticker: fetch /api/equity-meta-info?symbol={TICKER} via curl. This returns company name, ISIN, industry, F&O status, and listing date.
- If user gave a company name: fetch /api/NextApi/globalSearch/equity?symbol={NAME} via curl. Pick the best match, then fetch equity-meta-info to confirm.
- Cross-check with Screener.in: https://www.screener.in/company/{SYMBOL}/consolidated/ (WebFetch) to confirm the entity and get additional data. From these, lock: NSE symbol, full company name, ISIN, sector/industry, F&O eligibility, listing date.
Lock the reporting basis: consolidated or standalone. Screener.in shows both — if /consolidated/ returns data, use consolidated. This choice is final and must be passed to all 4 agents. Every agent must fetch data matching this basis only. If a source only shows the other basis, mark those values [wrong basis — discard].
Confirm the sector and industry classification from the NSE equity-meta-info response (industry field) and Screener.in.
Classify the company's segment structure using the latest annual report or investor presentation:
- Single-segment: One business line contributes >85% of revenue. Treat as a pure-play. No segment-level analysis needed.
- Multi-segment: 2+ segments each contributing >15% of revenue. Segment-level analysis is mandatory (see "Multi-segment company rules" section).
- Record for each material segment: segment name, revenue (₹ Cr), % of total revenue, EBIT or EBITDA if disclosed, and primary sector classification.
Note the latest reported quarter and annual report year available.
Check for corporate actions (stock splits, bonus issues, mergers/demergers) in the last 10 years. Screener.in's company page and NSE corporate actions data both show these. Record any split/bonus with date and ratio (e.g., "10:1 split Sep 2018", "1:1 bonus Jun 2022"). This list is passed to Agent B and used in Phase 2 to verify historical data is adjusted.

If any ambiguity remains (e.g., multiple entities with similar names, recent merger/demerger), ask the user before proceeding.

Only after entity resolution is complete, proceed to Phase 1. The segment classification from this phase controls whether multi-segment rules activate in all downstream sections.

Phase 1 — Parallel data collection (multi-agent)

NSE curl ownership: Only Agent A uses NSE curl endpoints. It inherits the session cookie file (/tmp/nse_cookies.txt) created during Phase 0. Agents B, C, and D use Screener.in, Tickertape, Moneycontrol, and web search only — no NSE curl. This avoids cookie contention from parallel writes to the same file.

Dispatch 4 research agents in parallel using the Agent tool (or platform equivalent). Each agent works independently and returns structured data. All 4 agents must be launched in a single parallel tool call — do not run them sequentially. The main agent must pass the resolved NSE symbol, company name, and cookie file path (for Agent A only) so agents can construct URLs directly (see "Source fetching guide").

Every agent must follow these rules:

Use the Source fetching guide URL patterns to fetch data directly. Do not blindly web-search when a direct URL is available.
For each metric collected, record: metric name, value, source URL (the actual URL fetched, not just the site name), as-of date or reporting period.
If a metric is not found, return DATA UNAVAILABLE — do not estimate or leave blank.
Search at least 2 different sources for every key metric. Record both values.
If sources disagree, report both values and flag the conflict. Do not resolve it — the main agent resolves conflicts in Phase 2.
Do not interpret or analyse. Just collect, attribute, and return.

Agent A — Price, size, and valuation

Collect:

Live CMP, 52-week high, 52-week low, market cap (₹ Cr), face value
P/E (TTM), P/B, EV/EBITDA — current values
P/E, P/B, EV/EBITDA — sector averages
P/E, P/B, EV/EBITDA — company's own 5-year high, low, and median
Dividend yield (current), dividend per share (last 3 years), payout ratio
Average daily trading volume (1-month, 3-month)
Average daily traded value in ₹ Cr (1-month, 3-month)

Primary: NSE quote-equity + trade_info + sector index. Secondary: Screener.in consolidated. Cross-ref: Tickertape (EV/EBITDA, sector averages, MF holdings). See Source fetching guide for all URLs.

Agent B — Financials and health

Consolidated-level (always collect):

Revenue, net profit, EBITDA — last 5 annual periods
EPS (diluted) — last 8 quarters with YoY change
EBITDA margin, net profit margin — last 5 annual periods
Revenue CAGR (3Y, 5Y), net profit CAGR (3Y, 5Y), EPS CAGR (3Y, 5Y)
Free cash flow — last 3 annual periods
Debt-to-equity ratio, interest coverage ratio, current ratio — current and 5-year trend
ROE and ROCE — current, 3-year average, 5-year average
Any material one-time items, other-income spikes, or non-cash adjustments in recent results

Corporate action adjustment (mandatory): The main agent passes a list of splits/bonuses from Phase 0. For all per-share metrics (EPS, DPS, book value):

Verify that historical values from Screener.in are split-adjusted. Screener.in usually adjusts automatically — confirm by checking if EPS shows a sudden discontinuity around a known split date. If pre-split EPS is ~10x (or Nx for an N:1 split) the post-split value, the data is NOT adjusted.
If unadjusted data is detected, flag it: [unadjusted: {split ratio} on {date}]. Do not compute CAGRs from unadjusted per-share data.
Revenue, net profit, EBITDA, and margins are NOT affected by splits/bonuses — only per-share metrics are.

Segment-level (collect only if Phase 0 classified the company as multi-segment):

For each material segment (>15% of revenue):
- Segment revenue — last 3 annual periods
- Segment EBIT or EBITDA (whichever the company discloses) — last 3 annual periods
- Segment margin (EBIT/revenue or EBITDA/revenue) — last 3 annual periods
- Segment revenue CAGR (3Y)
- Segment capital employed or identifiable assets, if disclosed
Source: annual report segment reporting note (Ind AS 108 / AS 17), investor presentations

Primary: Screener.in consolidated (10Y financials, quarterly results, balance sheet, cash flow, ROE, ROCE). Secondary: Tickertape (margins, ratios, FCF). Moneycontrol for quarterly commentary and one-time items. For segment data: web search annual report PDF / investor presentation. No NSE curl — Agent A handles NSE. See Source fetching guide for URLs.

Agent C — Peers and sector

Collect:

Peer identification — 3-step selection process:

Step 1 — Build a candidate list:

Start with Screener.in's peer comparison page for the target company. This gives sector peers pre-filtered by the platform.
Supplement with Tickertape's peer/competitor list for the same company.
Combine into a single candidate list. Typically 8-15 names.

Step 2 — Score each candidate on 3 dimensions:

Business-model similarity (most important): Do they earn revenue the same way? Same customer type (B2B/B2C/government)? Same core activity (manufacturing/services/platform/distribution)? A horizontal IT services company's peer is another horizontal IT services company, not an embedded software niche player.
Market-cap proximity: Within 0.3x-3x of the target. Closer is better.
Revenue-mix overlap: What % of the candidate's revenue comes from the same segment as the target's primary segment? Higher overlap = better peer.

Step 3 — Select the top 3:

Rank candidates by business-model similarity first, then break ties by market-cap proximity, then revenue-mix overlap.
For each selected peer, state in one line why it was chosen (e.g., "closest pure-play competitor in the same segment and market-cap band").

Special cases:

Conglomerates / multi-segment companies: If the target has 2+ segments each contributing >25% of revenue, select peers for the primary segment (largest revenue share). Note in the report that peer comparison reflects the primary segment only.
Monopolies or near-monopolies (e.g., IRCTC, CDSL, Coal India): If fewer than 2 legitimate business-model peers exist in India, do the following:
1. Use whatever domestic peers exist (even if only 1 or 0).
2. Fill remaining slots with the closest adjacent business-model peers and label them [adjacent peer] — e.g., for IRCTC, an online travel platform like MakeMyTrip/EaseMyTrip, noting the business model differs.
3. State clearly: True listed peers: [N]. Adjacent peers used to fill comparison: [N]. Do not force-fit unrelated companies just to reach 3.
Recently demerged or restructured entities: If the company was demerged in the last 2 years, peers should reflect the post-demerger business, not the pre-demerger parent's sector.

For each selected peer, collect:

Company name, NSE ticker, market cap
P/E, P/B, ROE (%)
Revenue growth: 1Y and 3Y CAGR
Debt-to-equity
Why this peer was selected (1-line)

Sector-level context (separate from peer selection):

Sector-level context: current cycle position (early/mid/peak/late/trough), major tailwinds, major headwinds
Regulatory developments affecting the sector in the last 12 months
Interest-rate, inflation, currency, and policy-risk sensitivity for the sector

Primary: Screener.in peers page. Cross-ref: Tickertape peers. For sector context: web search {SECTOR} India outlook {CURRENT_YEAR} from ET, Livemint, Business Standard, or industry bodies. No NSE curl — Agent A handles NSE. See Source fetching guide for URLs.

Agent D — Ownership, governance, and news

Collect:

Latest shareholding pattern: promoter %, FII %, DII %, public %, government %
Promoter pledging % (current and trend over last 4 quarters)
Change in FII and DII holding over last 4 quarters (direction and magnitude)
Board composition: number of independent directors vs total, any recent board changes
Auditor name and any recent auditor changes
Related-party transactions flagged in the latest annual report (if material)
Management commentary: key quotes or themes from the latest 2 earnings calls or investor presentations
Top 5 recent news items (last 6 months) relevant to long-term fundamentals — not daily price noise
Any ongoing regulatory actions, SEBI orders, tax disputes, or litigation

Primary: Screener.in consolidated (shareholding trends, pledging). Secondary: Tickertape (shareholding, MF holdings, institutional depth). Cross-ref: Moneycontrol (detailed shareholding, board of directors). Web searches for: earnings call transcript (latest quarter), investor presentation (current year), SEBI/regulatory/litigation news, general news from ET/Livemint/Business Standard. See Source fetching guide for all URLs.

Agent failure protocol

When agents return, the main agent must triage before proceeding to Phase 2:

Agent result	Action
Agent returns with data	Proceed normally to Phase 2.
Agent returns with >50% of key metrics as `DATA UNAVAILABLE`	Dispatch a replacement agent targeting the same metrics but using different sources (e.g., if Screener.in failed, try Tickertape + Moneycontrol). One retry only.
Agent crashes, times out, or returns an error	Log the failure. Proceed with remaining agents. The missing agent's metrics become explicit `DATA UNAVAILABLE` gaps in the report. Lower data confidence accordingly.
NSE session expired (Agent A returns empty/HTML)	Re-establish session (Step 1 curl) and dispatch Agent A again as a single retry.

Rules:

Maximum 1 retry per failed agent. Do not retry indefinitely.
Never block on a retry — if a replacement agent is dispatched, proceed with Phase 2 using data from the agents that succeeded. Merge the replacement's data when it returns.
If 2+ agents fail with no recovery, set data confidence to LOW or VERY LOW and note the blind spots prominently in the report header and View section.

Phase 2 — Review, cross-check, and merge (main agent)

After agents return (including any replacement agents), the main agent must:

Cross-reference overlapping metrics. Agents A, B, and C will have some overlapping data points (e.g., P/E, market cap, ROE). Compare values across agents. If they match (within 5% tolerance for ratios, 2% for absolute values), use the value from the highest-priority source. If they diverge beyond tolerance, flag it in the report as a data note.
Resolve conflicts using the source hierarchy. When two sources disagree:
- NSE/BSE filing > Screener.in > Tickertape > Moneycontrol > news articles
- More recent data > older data
- State which source was chosen and why if the divergence is material (>10%). Reporting basis mismatch (consolidated vs standalone): This is not a data conflict — it is a basis mismatch. Before applying the source hierarchy, check whether both values use the same reporting basis (consolidated or standalone) as locked in Phase 0. Discard any value that uses the wrong basis, regardless of source rank. If Screener.in shows consolidated and Tickertape shows standalone, use the one matching Phase 0's locked basis — do not average or blend them. If a source's reporting basis cannot be determined, mark it [basis unconfirmed] and deprioritize it.
Tag data freshness. For each major section, note: Data as of: [date or quarter]. If different metrics are from different periods, note this.
Identify gaps. List any metrics that came back as DATA UNAVAILABLE from all agents. These become explicit blind spots in the report.
Validate internal consistency. Check that:
- Market cap ≈ CMP × shares outstanding
- P/E ≈ CMP / EPS (TTM)
- D/E direction is consistent with debt and equity trends
- Margin trends are consistent with revenue and profit trends
- If any check fails, investigate and note the discrepancy.
Validate split/bonus adjustment. Using the corporate action list from Phase 0:
- Check that EPS, DPS, and book value series do not show sudden Nx jumps around known split/bonus dates.
- If a discontinuity is found, do NOT use the unadjusted figures for CAGR calculations or scenario analysis. Either adjust manually (divide pre-action values by the ratio) or mark the affected CAGRs as DATA UNAVAILABLE — unadjusted historical data.
Compile the merged dataset. Only after cross-checking, proceed to analysis and report writing.

Do not begin writing the report until Phase 2 is fully complete.

Multi-segment company rules

This section activates ONLY when Phase 0 classifies the company as multi-segment (2+ segments each >15% of revenue). For single-segment companies, skip this entirely.

Segment contribution table

Include this table in section 1.1 (Company and business):

Segment	Revenue (₹ Cr)	% of total	EBIT/EBITDA margin	3Y Rev CAGR	Primary sector

How segments modify core analysis sections

Valuation (§2): Add subsection "Segment-level valuation context". For each segment, note the typical pure-play valuation range and state whether the consolidated multiple is dragged down or inflated by segment weighting. Do NOT compute sum-of-the-parts or implied per-segment value — explain the direction of blending only.

Growth (§3): After the consolidated growth table, add a segment growth table:

Segment	Rev 3Y CAGR	Margin trend (expanding/stable/compressing)	Growth driver (1-line)

Identify the primary growth engine vs drag. Note if segment mix is shifting (e.g., "High-margin segment's share grew from 30% to 42% over 3Y").

Health (§4): If segment-level debt/capital employed is disclosed, note which segments carry most leverage. If not: Segment-level debt attribution not disclosed. Consolidated D/E reflects the blended group.

Returns (§5): If segment-level capital employed is available, note which segments earn highest/lowest ROCE and whether consolidated ROCE is pulled up by a capital-light segment or dragged down by a capital-heavy one.

Scenario analysis (§6): Build scenarios at the consolidated level but add one bullet per scenario naming which segment drives the outcome.

What NOT to do for multi-segment companies

Do not treat each segment as a separate stock. The analysis is still about the consolidated entity.
Do not compute segment-level P/E, P/B, or EV/EBITDA (these require segment-level market cap or EV, which don't exist for a single listed entity).
Do not speculate on demerger probability or value unlocking.

Qualitative red flags

Assess only material, source-based red flags such as:

Management over-promising vs actual delivery
Frequent auditor changes or accounting policy shifts without clear justification
Unusual related-party transactions
Weak board independence or regulatory-compliance concerns
Persistent high receivables or inventory days vs peers
Repeated capex write-downs, project delays, restructurings, or strategy drift
One-time items, other-income spikes, or non-cash gains inflating profits
Regulatory, tax, labour, environmental, or governance controversies

Do not add unsupported "gut-feel" warnings.

Sector-cycle and macro awareness

Assess: sector position (early/peak/late/trough), interest-rate sensitivity, inflation and input-cost exposure, currency and import/export impact, policy-risk dependence. Do not forecast interest rates, GDP, or future stock prices. Explain how these factors affect business quality, margins, balance-sheet risk, or valuation perception.

Liquidity analysis

Include a short liquidity and market-structure subsection.

Liquidity classification

Classification	Criteria
HIGHLY LIQUID	Avg daily traded value >₹50 Cr AND in Nifty 500 AND held by 30+ MF schemes
MODERATELY LIQUID	Avg daily traded value ₹5-50 Cr OR in a broad index but below ₹50 Cr daily value
POORLY LIQUID	Avg daily traded value <₹5 Cr OR not in any major index OR held by <5 MF schemes

If data is insufficient to classify, default to MODERATELY LIQUID and note: Liquidity classification based on limited data.

Institutional depth proxy

Instead of bid-ask spread (unavailable), use these as proxies for institutional depth:

FII + DII combined holding >30% → strong institutional depth
Number of MF schemes >50 → high retail-institutional overlap
F&O eligibility → exchange considers it liquid enough for derivatives
State whether the stock is F&O eligible.

Classification framework

Use these labels where appropriate, always with a one-line explanation:

Valuation: CHEAP / FAIR / EXPENSIVE / MIXED
Growth: ACCELERATING / STEADY / SLOWING / DECLINING
Health: SAFE / WATCH / RISK / LEVERAGED
Returns: GOOD / AVERAGE / WEAK
Peers: LEADING / MID-PACK / LAGGING
Liquidity: HIGHLY LIQUID / MODERATELY LIQUID / POORLY LIQUID
Overall fundamental quality: STRONG / MODERATE / WEAK

Scenario analysis

Build 3 illustrative scenarios — Bear, Base, Bull. The same anchoring applies to both the 5Y and 10Y horizons; the difference is the compounding implication.

What to model

Each scenario covers exactly 3 metrics: Revenue CAGR, EBITDA margin (average), EPS CAGR.

How to anchor each scenario

Do not invent growth rates. Derive from the company's own history:

Scenario	Revenue CAGR anchor	Margin anchor	EPS CAGR anchor
Bear	Worst rolling 3Y CAGR from the last 10 years (or available history)	Lowest annual EBITDA margin in the last 5 years	Worst rolling 3Y EPS CAGR from available history
Base	5Y trailing CAGR (most recent)	5Y average EBITDA margin	5Y trailing EPS CAGR
Bull	Best rolling 3Y CAGR from the last 10 years	Highest annual EBITDA margin in the last 5 years	Best rolling 3Y EPS CAGR from available history

If 10 years of data is not available, use whatever history exists and note the limitation.

How to compute rolling 3Y CAGRs

Agent B collects annual revenue, net profit, and EPS for the last 5–10 years. The main agent computes rolling CAGRs during report writing (not during data collection).

Given annual data points [Y0, Y1, Y2, ..., Yn] where Y0 is the oldest year:

A 3Y CAGR starting at Yi = (Y(i+3) / Yi)^(1/3) - 1
Slide the window: compute for every possible 3Y span in the dataset
From the resulting list, pick the worst (bear) and best (bull)

Example with 6 years of revenue (FY20–FY25):

Windows:  FY20→FY23, FY21→FY24, FY22→FY25
CAGRs:    12.1%,      8.3%,       18.7%
Bear anchor: 8.3% (FY21→FY24)
Bull anchor: 18.7% (FY22→FY25)

The Base case uses the most recent 5Y CAGR: (FY25 / FY20)^(1/5) - 1.

Important: Use split-adjusted per-share values for EPS CAGR (see Phase 2 step 6). Revenue and EBITDA are not affected by splits.

Output format

Scenario	Rev CAGR	EBITDA margin	EPS CAGR	Historical basis
Bear	X%	Y%	Z%	[which period this mirrors, e.g., "mirrors FY19-22 slowdown"]
Base	X%	Y%	Z%	[e.g., "trailing 5Y average"]
Bull	X%	Y%	Z%	[e.g., "mirrors FY16-19 expansion phase"]

After the table, add 6 bullets — two per scenario (one for 5Y, one for 10Y):

5Y implication: What this scenario means if sustained for 5 years — near-term compounding quality, margin risk, relevance for medium-term investors.
10Y implication: What this scenario means if sustained for 10 years — long-term structural compounding, whether the business model endures, relevance for patient capital.

The same CAGR compounding at different horizons tells very different stories. A bear-case 5% EPS CAGR is tolerable over 5Y but devastating over 10Y for a stock trading at 40x P/E.

Hard prohibitions

Do NOT compute a future stock price, implied market cap, or target valuation under any scenario.
Do NOT multiply projected EPS by any P/E multiple.
Do NOT use words like "likely", "expected", "should reach", "target", or "guaranteed".
Do NOT extrapolate beyond 10 years.
If the company has fewer than 3 years of history (recently listed), skip scenario analysis entirely and state: Scenario analysis omitted — insufficient operating history.

Insight rules

Insight quality

Every section must go beyond raw numbers and explain what those metrics mean for a long-term investor.
Use plain-English insight bullets to show what is strong, what is weak, and why it matters.
Avoid generic statements unless backed by metrics, comparisons, and explanation.
If valuation looks unusually cheap or expensive, explain what the market may be pricing in.

Insight in every section

After every major metric table, add 1-2 short bullets explaining: how the point affects long-term risk, return potential, and why it matters relative to surrounding metrics.

Pros and cons discipline

In the final View section: 3-5 key strengths, 3-5 key risks/watch points, 1 concise balanced summary sentence. Keep the tone neutral and analytical.

Writing style

No hard word limit. Every section must be present in full. Quality is not compromised for brevity. But density is mandatory — every sentence must carry data or insight. Apply these rules:

Tables first, prose second

Lead every section with the standardized table. The table is the primary content.
Insight bullets follow the table. They explain why the data matters, not what the data says — the reader can read the table.
Do not restate table values in prose. "ROE is 22.4%" after a table row showing ROE = 22.4% is waste.

One bullet, one point

Each insight bullet makes exactly one claim backed by one data point or comparison.
Bad: "The company has strong margins, good returns, and solid growth." — three claims, zero data.
Good: "EBITDA margin (28.3%) is 6pp above the 5Y sector median (22.1%) — suggests pricing power." — one claim, two data points, one inference.

No filler

Do not write:

"It is worth noting that..." — just state it.
"As mentioned earlier..." — don't reference other sections.
"Overall, the company..." — the View section summarizes; other sections don't need mini-summaries.
"This is an important metric because investors often look at..." — the reader knows; skip the preamble.
"Let us now look at..." / "Moving on to..." — section headings handle transitions.

Plain English

Write for a reader who understands finance but prefers clarity over jargon.
Prefer "debt is 1.8x equity" over "the leverage ratio stands at 1.8x."
Prefer "FII holdings fell 3.2pp in 2 quarters" over "There has been a notable decline in foreign institutional investor participation."

Data density

Every insight bullet must contain at least one number, percentage, ratio, or comparison.
Bullets without data are not insights — they are opinions. Remove them.
If a section has nothing quantitative to say, mark it DATA UNAVAILABLE rather than filling it with qualitative padding.

Output contract

You MUST write the final report to a file. This is not optional.

Build the full Markdown report in memory.
Save using the Write tool to {TICKER}_{YYYY-MM-DD}.md in the current working directory. NSE ticker uppercase, today's date.
Confirm the file path to the user.

Do not output the full report to chat, wrap in code fences, output HTML, or ask whether to save — just save it. If the Write tool fails, output the report to chat with: ⚠ FILE SAVE FAILED — copy this output manually.

Fixed heading template

If multi-segment (per Phase 0): add segment tables in §1.1, §2, §3, §4, §5 as specified in Multi-segment company rules.

Use this exact structure:

[Stock Name] – Fundamental Analysis

1. Snapshot

1.1 Company and business

1.2 Key price and size metrics

1.3 Liquidity and market structure

2. Valuation

2.1 Multiples vs sector and history

2.2 Valuation signal

2.3 Why the market may be pricing this way

3. Growth

3.1 Revenue, profit, EPS, margin trends

3.2 EPS – last 8 quarters

3.3 Growth insight (pros and cons)

4. Health

4.1 Balance-sheet and liquidity

4.2 Free cash flow and coverage

4.3 Health insight (pros and cons)

5. Returns

5.1 ROE, ROCE, dividend metrics

5.2 Return quality signal

5.3 Return-quality insight (pros and cons)

6. Scenario analysis

6.1 Bear / Base / Bull table

6.2 Scenario insight (5Y and 10Y implications per scenario)

7. Peers

7.1 Peer comparison table

7.2 Peer standing and edge

7.3 Peer insight (where this stock wins and loses vs peers)

8. Ownership

8.1 Promoter, FII, DII, pledging

8.2 Management commentary and governance notes

8.3 Ownership and governance insight (pros and cons)

9. View

9.1 Overall fundamental quality

9.2 Key strengths (3-5 pros)

9.3 Key risks / watch points (3-5 cons)

9.4 Red-flag and governance highlights

9.5 Sector-cycle and macro sensitivity insight

9.6 1-sentence summary (balanced)

10. Source Appendix

Standardized tables

Use exactly these column layouts:

Type column values: peer (true business-model peer) or adjacent (closest available, different business model). If the target is a monopoly/niche with no true peers, add a note row: True listed peers: [N]. Adjacent peers used: [N].

Data confidence and summary snapshot

At the top of the report include, in this order:

A confidence note: Data confidence: [HIGH / MODERATE / LOW / VERY LOW] — based on number of metrics backed by live sources.
The freshness block (see "Data freshness — Level 1" for exact format).
A summary snapshot table:

Aspect	Verdict (Signal)	Why it matters (1-line)
Business quality	STRONG / MODERATE / WEAK	[1-line]
Growth trend	ACCELERATING / STEADY / SLOWING	[1-line]
Financial health	SAFE / MODERATE / RISK	[1-line]
Valuation	CHEAP / FAIR / EXPENSIVE	[1-line]
Liquidity	HIGH / MODERATE / POOR	[1-line]
Overall quality	STRONG / MODERATE / WEAK	[1-line]

Failure handling

If data is incomplete or weak: mark missing items DATA UNAVAILABLE, do not estimate or backfill, lower the data confidence rating, and mention material blind spots in the relevant section and the View section.

If the stock is too obscure, recently listed, thinly traded, or poorly covered: say so clearly, focus on available filings/ownership/liquidity/risk notes, and avoid false precision.

Data freshness

Financial data comes from different reporting periods. The report must make this visible to the reader at three levels:

Level 1 — Report header

At the top of the report, immediately after the data confidence note, add:

Report generated: [today's date]
Price data as of: [date/time CMP was fetched]
Latest quarterly results: [quarter, e.g., Q3 FY26 (Dec 2025)]
Latest annual report: [year, e.g., FY25]
Shareholding pattern as of: [quarter end date]

Level 2 — Section-level tags

Each major section (Valuation, Growth, Health, Returns, Peers, Ownership) must open with: Data as of: [reporting period or date]. Source: [primary source]. If mixing periods: Data as of: CMP [date], financials [quarter/year]. Source: [sources].

Level 3 — Inline flags for stale or mixed-vintage data

When a metric is materially older than surrounding data, append: [stale: Q2 FY26] (2+ quarters behind), [latest] (most recent filing), or [intraday] (live price). Example: | Debt-to-equity | 0.42 [stale: FY25 AR] | Declining | SAFE | Latest balance sheet is 9 months old; verify with Q3 filing |

Cross-referencing with freshness

During Phase 2 merge: if values differ from different reporting periods, it is a vintage mismatch — use the more recent value and note the discrepancy. If values differ from the same period, it is a true conflict — resolve using the source hierarchy.

Source appendix

Use this exact table format, grouped by source:

Source	URL / endpoint fetched	Sections supported	Data as of
NSE quote-equity API	`/api/quote-equity?symbol=TCS`	§1.2, §2.1	04-Apr-2026
Screener.in	`screener.in/company/TCS/consolidated/`	§3, §4, §5, §7, §8	FY25 AR
...	...	...	...

Keep it to actual sources fetched — not sources listed in the skill. If a source was attempted but returned DATA UNAVAILABLE, do not include it.

Completion checklist

Before you consider this task done, verify every item:

File output:

The .md file was saved using the Write tool (not just printed to chat)
The file name matches {TICKER}_{YYYY-MM-DD}.md
The file path was confirmed to the user

Report structure:

All 10 sections (§1–§10) are present in the report
Summary snapshot table is at the top of the report, immediately after the data confidence note
Scenario analysis table exists — or is explicitly skipped with Scenario analysis omitted — insufficient operating history
Peer table has exactly 3 peers — or has a documented exception (True listed peers: [N]. Adjacent peers used: [N].)
No DATA UNAVAILABLE appears in the summary snapshot table without an explanation in the corresponding "Why it matters" column

Data quality:

Every section has a Data as of tag
The report header includes the freshness block
If corporate actions (splits/bonuses) occurred, all per-share CAGRs use adjusted data or are marked DATA UNAVAILABLE — unadjusted

If any box is unchecked, you are not done. Go back and complete it.

Final note

This is a view based on fundamentals only, not investment advice.

About this skill

Operating objective

Non-negotiable rules

Trigger behavior

Input contract

Research workflow

Preferred sources

Source fetching guide

NSE India (primary — official data via curl)

Screener.in (secondary — fills gaps NSE doesn't cover)

Tickertape (tertiary — cross-reference)

Moneycontrol (for qualitative data)

General fetching rules

Phase 0 — Entity resolution (main agent, mandatory first step)

Phase 1 — Parallel data collection (multi-agent)

Agent A — Price, size, and valuation

Agent B — Financials and health

Agent C — Peers and sector

Agent D — Ownership, governance, and news

Agent failure protocol

Phase 2 — Review, cross-check, and merge (main agent)

Multi-segment company rules

Segment contribution table

How segments modify core analysis sections

What NOT to do for multi-segment companies

Qualitative red flags

Sector-cycle and macro awareness

Liquidity analysis

Liquidity classification

Institutional depth proxy

Classification framework

Scenario analysis

What to model

How to anchor each scenario

How to compute rolling 3Y CAGRs

Output format

Hard prohibitions

Insight rules

Insight quality

Insight in every section

Pros and cons discipline

Writing style

Tables first, prose second

One bullet, one point

No filler

Plain English

Data density

Output contract

Fixed heading template

1. Snapshot

1.1 Company and business

1.2 Key price and size metrics

1.3 Liquidity and market structure

2. Valuation

2.1 Multiples vs sector and history

2.2 Valuation signal

2.3 Why the market may be pricing this way

3. Growth

3.1 Revenue, profit, EPS, margin trends

3.2 EPS – last 8 quarters

3.3 Growth insight (pros and cons)

4. Health

4.1 Balance-sheet and liquidity

4.2 Free cash flow and coverage

4.3 Health insight (pros and cons)

5. Returns

5.1 ROE, ROCE, dividend metrics

5.2 Return quality signal

5.3 Return-quality insight (pros and cons)

6. Scenario analysis

6.1 Bear / Base / Bull table

6.2 Scenario insight (5Y and 10Y implications per scenario)

7. Peers

7.1 Peer comparison table

7.2 Peer standing and edge

7.3 Peer insight (where this stock wins and loses vs peers)

8. Ownership

8.1 Promoter, FII, DII, pledging

8.2 Management commentary and governance notes