pine-validate

pine-validate

Validate Pine Script v6 function references against the v6 allowlist. The authoritative source of truth is the validate_function MCP tool on the pinescript server. Never assert a function is valid from memory — always check via the tool.

When to activate

• The user pastes a Pine Script snippet and asks "does this compile?", "is this valid Pine v6?", "check this for errors", or similar.
• The user asks about a single function: "is ta.adx valid?", "does security() exist in v6?", "what replaced study()?".
• The user asks to audit a script for hallucinated, renamed, or v5-era function references before running on TradingView.

Do not activate for general Pine Script authoring requests — that is the pine skill's job. This skill is read-only validation.

Workflow (mandatory, in order)

Step 1: Parse the input and extract function references

From the user's input, extract every distinct function reference. A function reference is anything that looks like a call site:

• <namespace>.<func>( — namespaced calls like ta.rsi(, strategy.entry(, request.security(, math.sum(.
• <func>( — top-level calls like plot(, indicator(, input(.

Apply these filters when extracting:

• Ignore comments. Pine comments start with //. Skip anything after // on a line.
• Ignore string literals. Anything inside "..." or '...' is not a call.
• Ignore definitions. Skip user-defined function definitions of the form myFunc(...) => — those are declarations, not calls against the allowlist.
• Deduplicate. Validate each unique name once, even if it appears many times in the snippet.

If the user provided only a bare function name (e.g. "is ta.adx valid?"), treat that as a single reference to validate.

Step 2: Validate each reference via validate_function

For each unique function name extracted, call the MCP tool:

validate_function(fn_name="<name>")

The tool returns a ValidationResult object with: