More screenshots
Works with the AI Agents you already use
Sector Lens Pack: the right metrics for each sector
One valuation lens per sector: what to trust, what lies, thresholds and traps, plus an offline metric checker.
$12
Sector Lens Pack: the right metrics for each sector
Example session with this skill installed
Is JPM cheap at 12x earnings? ROTCE is 20%, P/TBV 2.4x (illustrative inputs). And a chip maker at 9x trailing P/E on a 30% margin, median 15%?
- Read your context and instructions
- Compiled the sector lens
Run 2026-11-16 (illustrative inputs):
Lens check: Banks and lenders -> financial lens, mature stage (industry: Banks - Diversified)
TRUST rotce, ptbv, roe, pb, pe_fwd, pe_ttm, dividend_yield, payout_ratio, share_dilution
IGNORE
- net_debt_ebitda: debt is the raw material
- roic: leverage on customers' debts, not return on capital
READINGS (rule-of-thumb bands, a cue not a verdict)
pb 2.00 fair
pe_fwd 11.50 cheap
pe_ttm 12.00 cheap
ptbv 2.40 fair
roe 16.0% decent
rotce 20.0% strong
share_dilution -2.0% buybacks
DERIVED
justified_multiple 2.43
price_over_justified 0.99
FLAGS
[INFO] justified_multiple: P/TBV 2.40x vs justified 2.43x = (ROTCE 20.0% - 3%) / (10% - 3%); price/justified 0.99, at or below [engine]
[INFO] high_roe: ROE 16.0% >= 15% [engine]
BLOCK = a reason not to act on a cheap-looking reading; WARN = check before relying; INFO = context. Not investment advice.
Lens check: Semiconductors -> standard lens, mature stage
TRUST pe_fwd, roic, op_margin, net_debt_ebitda, revenue_growth
IGNORE
- pe_ttm: trailing P/E on peak margins misleads: see the normal P/E
READINGS (rule-of-thumb bands, a cue not a verdict)
pe_fwd 12.00 cheap
roic 18.0% strong
net_debt_ebitda 0.50 low (heavy from 3x for this lens)
DERIVED
normal_pe 18.0
FLAGS
[BLOCK] peak_margin: operating margin 30.0% is 1.5x+ its median 15.0% after a down-sales year: trailing P/E 9.0 is 18.0 on a normal margin [engine]
[BLOCK] earnings_falling: forward P/E 12.0 is 10%+ above trailing 9.0: the low multiple sits on a profit expected to shrink [engine]
[WARN] book_to_bill: book-to-bill 0.90 < 1: orders trail sales [guide]
NOT SUPPLIED (would sharpen this lens): fcf_yield
BLOCK = a reason not to act on a cheap-looking reading; WARN = check before relying; INFO = context. Not investment advice.
Connects securely to your tools. The creator never sees your data.
About this skill
A bank on 8x earnings, a REIT on a P/E of 40 and a chip maker on 9x peak-year profit all look like something they are not. The same ratio means different things by sector, and agents that open with P/E get all three wrong.
Sector Lens Pack gives your agent one lens per sector: which metrics to trust, which ones lie, the thresholds that matter and the traps that are not in a number. It carries 23 sectors in a router table, four scoring lenses (financial, real estate, utility, standard) with exact constants, and an offline script that checks a few metrics against the right lens.
What it does
- Routes a company to a lens, including mixed industries decided by the balance sheet (loans 50%+, equity 15% or less, property 50%+).
- Prints metrics to trust and to ignore, bands, derived figures such as a bank's justified price to tangible book, and flags tagged BLOCK, WARN or INFO.
- Tests peak margin and the normal P/E, debt lines (3x, 5x, 8x), stage (early, growth, mature), runway, dilution and dividend cover.
What you get
| File | Holds |
|---|---|
| SKILL.md | 8 rules, 23-row router, method |
| references/engine-thresholds.md | lenses, debt lines, scales, cautions, bands |
| references/financials.md | banks, insurers, asset managers, payments |
| references/real-assets.md | REITs, utilities, telecom |
| references/cyclicals.md | autos, aerospace, industrials, hardware, retail, staples |
| references/tech.md | software, platforms, semiconductors |
| references/health.md | pharma, biotech, medtech, healthcare |
| references/resources.md | energy, mining, nuclear and uranium |
| scripts/lens_check.py | offline lens checker (stdlib, exits 0, 1, 2) |
| scripts/test_lens_check.py | 25 unit tests, happy and refusal paths |
| templates/ | 4 illustrative metric files |
Who it's for
Self-directed investors, analysts and developers who let an agent screen or compare stocks and want the sector logic written down once.
Scope
Offline and rule-based: no network, no data feed, no API keys. Thresholds are starting points. Constants marked engine come from a working scoring engine; guide numbers are sector rules of thumb. Not covered: regulatory capital by jurisdiction, unlisted assets, crypto assets.
Not investment advice.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse All Claude Skills.
Offline and rule-based: no data fetching. Thresholds are starting points, not laws. The script tests numbers; qualitative traps (patent cliffs, contract books, captive finance arms) are in the references for the agent to apply. Template metrics are illustrative, not market data. No regulatory capital ratios, unlisted assets or crypto assets.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 1 day ago
- Software
Python 3.9+Your own sector metrics
Standard library only, no API keys. Give percentages as fractions, so 12% is 0.12.
Permissions
No network and no hosts. The script runs locally, reads only the metrics file you pass (or stdin) and writes to stdout.