Kanchi Dividend SOP turns the Kanchi five-step dividend method into a repeatable workflow for US stocks. It starts from a mandate and a ticker list, checks the regular forward yield, payout safety by sector, valuation and one-off events, and ends with a verdict tier, a one-page memo and a limit-order plan.
Python scripts use Financial Modeling Prep data and a web search step covers pending corporate events. The output is a research record with explicit assumptions and gives no investment advice.
What You Get
- A screening table with a verdict tier per ticker, from CLEAN-PASS to FAIL
- One-page stock memos with a provenance block for sources and open blockers
- A limit-order plan with split tranches and invalidation conditions
- A run context so results from different yield floors are never mixed
- Scripts that build the plan and the entry-signal files in a reports folder
Common Use Cases
- Screening a list of dividend stocks against a yield floor and payout safety checks
- Writing a one-page underwriting memo for one ticker
- Planning pullback entries as split limit orders with blockers noted
- Preparing a handoff to the review monitor and the tax accounting skill
What Makes It Unique
It is cautious by design. The yield check uses the regular forward yield, never a trailing figure that includes specials. A missing adjusted-earnings figure or a failed event scan caps the verdict at HOLD-REVIEW and blocks the first tranche. Safety checks differ for banks, regulated utilities and insurers, and the notes name a current limit where the five-year average yield cannot yet produce buy targets.
Under The Hood
Claude follows the written procedure and runs Python scripts from the skill folder through the shell, with an FMP API key from the FMP_API_KEY environment variable. The scripts call the Financial Modeling Prep web API for quotes, profiles and financials and write plan and entry-signal files into a reports folder. For the forward event scan, the instructions tell Claude to run web searches and check issuer investor-relations pages and SEC filings, then record findings in a JSON file the script reads. Reference files hold thresholds and templates.