cilipep/ai-agent-with-tslab/tslab-indicator-blocks

TSLab: Using Indicators in Block Scripts (API + Cubes)

Use this skill when the user asks to use an indicator in a script graph (cubes/blocks) and wants it wired and run via Web API.

If the user asks to implement a new indicator in C#, use $tslab-indicator-authoring.

Hard Rules

• Always use /api/handlers/{typeName} as the ground truth for inputs/outputs/parameters.
• If you are unsure which Execute(...) signature family the server supports (stream vs values vs precalc), use GET /api/sdk/contracts and copy a shipped template.
• Always read GET /api/scripts/{name}/explain before editing a graph; never guess block IDs or parameter names.
• After ops edits: POST /load -> POST /validate -> POST /build -> POST /run and treat warnings as actionable.
• Keep block names readable: meaningful ASCII CodeName; for new scripts prefer blockId = CodeName = VisualName.
• In formulas, i is built-in; do not connect it. UnknownExpressionReference for i is expected on some builds and is non-blocking.
• For period/window indicators in formulas, use startup guards (i >= period ? Expr : Fallback).

Choose the indicator source

1. Function indicator (custom indicator script):

• Use $tslab-api-indicators.
• In scripts: add DynamicItem and set HandlerName = <FunctionName> (not HandlerTypeName).

1. DLL indicator library:

• Disk: HandlersFolder hot reload (drop .dll + optional .pdb).
• DB: upload/replace/delete via $tslab-api-indicator-dlls (/api/indicator-dlls).

Read only when needed

• Add/connect blocks via ops: use $tslab-api-script-edit (primary).
• Indicator-specific wiring pitfalls: references/troubleshooting-stream-value-precalc.md
• Minimal "how to wire a handler block": references/using-handler-in-graph.md