More screenshots
Works with the AI Agents you already use
Earnings Quality Forensics: real profit vs engineered
Red-flag table for accruals, cash conversion, SBC, one-offs, goodwill and dilution, plus a script that checks SEC filings.
$7
Earnings Quality Forensics: real profit vs engineered
Example session with this skill installed
Is Apple's profit backed by cash? Check accruals, SBC, dilution and buybacks over the last three fiscal years.
- Read your context and instructions
- Compiled the earnings quality forensics
Run 2026-11-16 (real SEC data): Apple Inc. (CIK 0000320193, SIC 3571 Electronic Computers) lens: standard USD millions (shares in millions), oldest to newest; first row is the base year for growth; source: SEC EDGAR companyfacts FY end revenue net income cfo fcf sbc owner fcf assets goodwill shares 2023-09-30 383,285 96,995 110,543 99,584 10,833 88,751 352,583 n/a 15,813 2024-09-28 391,035 93,736 118,254 108,807 11,688 97,119 364,980 n/a 15,408 2025-09-27 416,161 112,010 111,482 98,767 12,863 85,904 359,241 n/a 15,005 2026-09-26 478,834 132,175 145,097 136,056 13,789 122,267 404,548 n/a 14,720 FCF = CFO - capex; owner FCF = FCF - SBC. n/a = tag absent for that year. Check Value Rule Verdict Note accruals 2026 -3.2% > 10% ok SBC 2026 2.9% rev / 10% FCF >= 10% rev or 50% FCF ok accruals 2025 0.1% > 10% ok SBC 2025 3.1% rev / 13% FCF >= 10% rev or 50% FCF ok accruals 2024 -6.7% > 10% ok SBC 2024 3.0% rev / 11% FCF >= 10% rev or 50% FCF ok accruals 2023 -3.8% > 10% ok SBC 2023 2.8% rev / 11% FCF >= 10% rev or 50% FCF ok CFO / NI, 4y sum 1.12 < 0.70 ok goodwill, latest n/a assets / n/a equity > 50% assets or > 100% equity ok share count CAGR -2.4% >= 3% named, >= 5% heavy ok buybacks shrink it buyback minus SBC 2026 72,601 M < 0 means buybacks only offset pay ok receivables vs revenue growth 10.2% vs 15.1% gap > 10% ok inventory vs revenue growth 154.3% vs 15.1% gap > 10% FLAG non-operating share of pretax 0.7% |x| > 10% ok 1 flag(s). A flag is a question for the filing, not a verdict. Not investment advice.
Connects securely to your tools. The creator never sees your data.
About this skill
A cheap P/E on profits the cash never shows is not cheap. This skill gives your agent the checks a forensic analyst runs before trusting an earnings number, with the thresholds that flag them and the order to run them in. ## What it does Your agent classifies the business (standard, bank or insurer, REIT, utility, pre-profit), then tests accruals, cash backing over several years, share-based pay, owner free cash flow, dilution and buybacks that only offset issuance, goodwill load, tax and non-operating one-offs, and receivables and inventory against sales. Each check has a formula, a flag level, what it usually means and what filing line to read next. It also computes the P/E without the one-offs and shows both. ## What you get | Path | Holds | |---|---| | SKILL.md | eight rules, a 14-row red-flag table, an ordered method | | references/formulas-and-xbrl-tags.md | formulas, SEC tag fallbacks, EDGAR traps, threshold origins | | references/one-offs.md | tax, stake-gain and drop tests; the adjusted multiple | | references/sector-exceptions.md | banks, insurers, REITs, utilities, pre-profit names | | scripts/quality_check.py | SEC EDGAR facts (or your CSV) to a flag table, stdlib Python | | templates/ | report write-up and an example CSV | ## Who it's for Self-directed investors, analysts and developers who want an agent to stress-test a bull case, write a bear case from the filings, or check whether free cash flow survives paying staff in cash. ## Scope Covered: US filers' annual 10-K figures, the method for quarterly one-off tests, sector exceptions. Thresholds are screens from working practice, not research results, and the references say which are tuned and which are rules of thumb. Not covered: IFRS filers (the script exits with a clear message), forensic review of footnotes, fraud detection. A flag is a question for the filing; no flag does not prove clean books. SEC requests need your contact in a User-Agent. This is research tooling, 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.
Annual 10-K figures for US filers only; the script does not run the quarterly one-off tests, EBITDA, ROIC or the FCF-float test (the method does them by hand). XBRL tags can be missing or wrong: a missing tag prints n/a. Share counts are as reported, so a split reads as a jump and is flagged for checking.
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
- Accounts
Sec
- Software
Python 3.9+ (standard library only)Internet access for EDGAR mode
SEC_USER_AGENT env var with your name and email.
Permissions
Allowed Hosts
The script calls only the two SEC hosts, writes the ticker map to the system temp dir (or EDGAR_CACHE_DIR), reads one CSV you name in offline mode. No credentials stored.