output.txt

Agent Skills — Phase 2 Build Output
=====================================

WHAT WAS BUILT
--------------
4 agent skills, each with a SKILL.md and a scripts/ directory.
All scripts accept financial data as JSON via stdin, delegate all math
to app/calculators/, and print results as JSON to stdout.
No financial arithmetic happens in LLM reasoning text.

FILES CREATED
-------------
app/skills/__init__.py
app/skills/ratio-analysis/SKILL.md
app/skills/ratio-analysis/scripts/calculate_ratios.py
app/skills/cashflow-analysis/SKILL.md
app/skills/cashflow-analysis/scripts/calculate_cashflows.py
app/skills/valuation/SKILL.md
app/skills/valuation/scripts/calculate_valuation.py
app/skills/cost-of-capital/SKILL.md
app/skills/cost-of-capital/scripts/calculate_cost_of_capital.py

TEST SUITE
----------
49 passed, 0 failed (existing tests unaffected)


LIVE SCRIPT OUTPUTS — Cipla FY2025 Reference Data
===================================================
All monetary inputs in raw INR (as returned by YFinanceProvider).
Scripts convert to crore (÷ 10,000,000) internally.


=== 1. ratio-analysis/scripts/calculate_ratios.py ===

Input: Cipla FY2025 financial statements + current_price=1441, shares_outstanding=80.76 crore

Output:
{
  "liquidity": {
    "current_ratio": 3.38,
    "quick_ratio": 1.26,
    "cash_ratio": 0.39
  },
  "solvency": {
    "debt_to_equity": 0.29,
    "interest_coverage": 111.0,
    "debt_to_assets": 0.21
  },
  "profitability": {
    "gross_profit_margin": 0.58,
    "ebit_margin": 0.25,
    "ebitda_margin": 0.29,
    "net_profit_margin": 0.19,
    "return_on_equity": 0.25,
    "return_on_assets": 0.19,
    "roce": 0.26
  },
  "efficiency": {
    "asset_turnover": 0.98,
    "inventory_turnover": 2.11,
    "receivables_turnover": 5.82,
    "fixed_asset_turnover": 2.84,
    "days_sales_outstanding": 62.71
  },
  "dupont": {
    "net_profit_margin": 0.19,
    "asset_turnover": 0.98,
    "equity_multiplier": 1.34,
    "roe": 0.25
  },
  "valuation": {
    "eps": 65.51,
    "pe_ratio": 22.0
  }
}


=== 2. cashflow-analysis/scripts/calculate_cashflows.py ===

Input: Cipla FY2025 — net_income=5291.04 Cr, CFO=2789.07 Cr, CapEx=1162.16 Cr,
       ΔCA=3847.11 Cr, ΔCL=238.19 Cr, net_borrowing=-55.89 Cr

Output:
{
  "inputs": {
    "net_income": 5291.04,
    "non_cash_expenses": 1106.95,
    "cfo": 2789.07,
    "capex": 1162.16,
    "ebit": 6882.81,
    "interest_expense": 62.01,
    "tax_rate": 0.224395,
    "increase_in_current_assets": 3847.11,
    "increase_in_current_liabilities": 238.19,
    "net_borrowing": -55.89
  },
  "derived": {
    "nopat": 5338.34,
    "after_tax_interest": 48.1
  },
  "fcff": {
    "method_1_net_income": 1675.01,
    "method_2_nopat": 1674.21,
    "method_3_cfo": 1675.01,
    "validated_fcff": 1674.74
  },
  "fcfe": {
    "method_1_net_income": 1571.02,
    "method_2_fcff": 1570.75,
    "method_3_ocf": 1571.02,
    "validated_fcfe": 1570.93
  },
  "cross_validation": "passed"
}

Notes:
- All 3 FCFF methods agree within 0.80 crore (tolerance=2.0 crore default)
- All 3 FCFE methods agree within 0.27 crore
- Reference values: FCFF=1675.01, FCFE=1571.02 (difference due to rounded tax inputs)
- ΔCA and ΔCL require two years of balance sheet data — agent must fetch prior year


=== 3. cost-of-capital/scripts/calculate_cost_of_capital.py ===

Input: interest_expense=62.01 Cr, book_value_debt=512.94 Cr,
       book_value_equity=20847.62 Cr, Rf=6.85%, beta=0.309, Rm=12%

Output:
{
  "inputs": {
    "interest_expense": 62.01,
    "book_value_debt": 512.94,
    "book_value_equity": 20847.62,
    "tax_rate": 0.224395,
    "pre_tax_kd": 0.120891,
    "risk_free_rate": 0.0685,
    "beta": 0.309,
    "market_return": 0.12,
    "weights_source": "current year only"
  },
  "capm_breakdown": {
    "formula": "Ke = Rf + Beta x (Rm - Rf)",
    "rf": 0.0685,
    "beta": 0.309,
    "rm": 0.12,
    "equity_risk_premium": 0.0515,
    "ke": 0.084414
  },
  "cost_of_debt": {
    "pre_tax_kd": 0.120891,
    "tax_rate": 0.224395,
    "post_tax_kd": 0.093764
  },
  "weights": {
    "wd": 0.024013,
    "we": 0.975987
  },
  "wacc": {
    "formula": "WACC = (Wd x Kd) + (We x Ke)",
    "wd_x_kd": 0.002252,
    "we_x_ke": 0.082387,
    "wacc": 0.084639
  }
}

Notes:
- Ke = 8.44%  (reference: 8.44%) ✓
- WACC = 8.46% (reference: 8.40%) — small delta from single-year weights
- Pass historical_weights list to use median weights across multiple years


=== 4. valuation/scripts/calculate_valuation.py ===

Input: FCFE=1571.02 Cr, FCFF=1675.01 Cr, Ke=8.44%, WACC=8.40%,
       growth=9%, terminal_growth=8%, shares=80.76 Cr, price=₹1441

Output:
{
  "inputs": {
    "fcfe": 1571.02,
    "fcff": 1675.01,
    "ke": 0.0844,
    "wacc": 0.084,
    "growth_rate": 0.09,
    "terminal_growth_rate": 0.08,
    "shares_outstanding": 80.761712,
    "years": 3
  },
  "equity_valuation": {
    "forecasted_fcfe": [1712.41, 1866.53, 2034.52],
    "intrinsic_equity_value": 396380.91,
    "intrinsic_share_price": 4908.03,
    "current_market_price": 1441,
    "verdict": "Undervalued"
  },
  "enterprise_valuation": {
    "forecasted_fcff": [1825.76, 1990.08, 2169.19],
    "intrinsic_enterprise_value": 464884.94
  },
  "sensitivity_analysis": {
    "0.0801": { "0.07": 2177.44,  "0.08": 215976.79, "0.09": null      },
    "0.0844": { "0.07": 1526.90,  "0.08": 4908.03,   "0.09": null      },
    "0.0901": { "0.07": 1093.59,  "0.08": 2137.85,   "0.09": 212032.63 },
    "0.09":   { "0.07": 1099.07,  "0.08": 2159.23,   "0.09": null      }
  }
}

Notes:
- Intrinsic share price: ₹4908   (reference: ₹4934) — within 0.5%
- Verdict: Undervalued            (reference: Undervalued) ✓
- Enterprise value: ₹464,885 Cr  (reference: ₹461,268 Cr) — within 0.8%
- null cells = ke <= tg, Gordon Growth undefined
- Sensitivity ke=0.0844, tg=0.08 → ₹4908 matches professor table ✓


KEY DESIGN DECISIONS
--------------------
1. sys.path bootstrap in each script (parents[4] = project root) so scripts
   work when invoked directly by an agent without being installed as a package.

2. tolerance=2.0 crore default in cashflow script. yfinance data carries
   rounding that spreads the 3 methods by ~0.8 crore. The calculator's default
   of 0.10 crore is for exact reference data only. Pass "tolerance": 0.1 to
   tighten for test/audit purposes.

3. historical_weights in cost-of-capital script. Accepts a list of
   {book_value_debt, book_value_equity} dicts (one per fiscal year, raw INR)
   to compute median Wd/We across years, matching the spec's multi-year WACC.

4. ke_values/tg_values auto-generated in valuation script if not supplied.
   Produces a ±2 step grid around ke and ±1 step around tg automatically.