TradingView Auto-Publish Toolkit and Original Script Batches

Publishing a TradingView script properly involves a dozen fiddly steps: create a fresh script without touching the open draft, load source, save, add to chart, verify it actually plots numbers, write a moderation-safe description, pick categories, tick house rules and publish. The toolkit automates all of it against TradingView Desktop over the DevTools Protocol, pinned to a scratch layout so my working chart is never touched.

Its optimiser drives TradingView's own engine through the study input API, splitting in-sample and out-of-sample by trade exit time at about two seconds per combination; a numba stop-and-reverse optimiser handles signal indicators on exported bars and matched on-chart counts exactly. Platform limits were discovered and encoded: five public scripts per 24 hours and fifteen per rolling thirty days.

Three batches shipped: six trend-following scripts, six data-driven originals (session auction profile, delta exhaustion, Markov regime matrix, relative-strength rotation, multi-timeframe liquidity pools, compression-expansion ladder) and five famous-trader method indicators (Minervini, Williams, Raschke, Dennis, Darvas) with a shared trade-journal layer of entry, stop and target lines and a results table. A companion skill reverse-engineered the design language of 134 AlgoAlpha open-source scripts into a style DNA, an idea matrix and a signal probe with a shuffled baseline, with the rule that no accuracy is claimed without an edge flag.

How it works

1. Idea drafted and compile-checked offline
2. stage.sh creates a new script, loads source, saves and adds to chart
3. Optimiser grids strategy inputs through TradingView's engine
4. Verifier reads plots and table cells to confirm output
5. pub.sh fills description, categories and house rules, then publishes
6. Scheduler spaces publishes to respect platform limits