Free tool / Agent responses
Agent output
validator.
Check JSON syntax, top-level keys and tool-call structure before connecting an agent to a workflow.
Runs in your browser. Nothing you paste is stored or sent.
Paste the JSON itself, without Markdown fences. Maximum 200,000 characters.
Comma-separated, case-sensitive key names. Presence is checked; values may still be null.
Results
Your results will appear here.
How to use it
- Paste the response and optionally enter the keys your next step requires.
- Select Validate response. Syntax errors include a line, column and character position, all counted from 1.
- Correct the response or agent instructions and check again.
Top-level objects containing name or arguments are treated as tool calls. A tool_calls array supports direct calls and OpenAI-style function wrappers. Each call needs a non-empty string name and arguments that are an object, or a JSON string that parses to an object.
Warnings flag possible sk- keys, AWS access key IDs and password text, even when JSON does not parse. Detection is limited and can produce false positives. A passed check does not prove that a response is safe, matches a full schema or should be executed. This tool never executes tool calls.
For help testing one agreed agent workflow, see agent reliability.