Skip to main content

​
prompts_decode

Hydrate a previously encoded prompt (inline ktof string or stored payload ID) into structured JSON.

​
Parameters

NameTypeRequiredDescription
ktofstr✅Encoded payload returned by prompts_encode/optimize_request (or a payload ID supported by your account tier).
optionsDecodeOptions❌Toggle indentation or strict mode.
replay_metadict[str, Any]❌Supplementary data for audit logs when replaying historical prompts.
metadatadict[str, Any]❌Additional context to attach to the decoded response (merged with stored metadata).

​
Code example

encoded = await client.prompts_encode({"prompt": {"messages": [{"role": "user", "content": "Hi"}]}})

decoded = await client.prompts_decode({
    "ktof": encoded["result"],
    "metadata": {"decoded_by": "docs-demo"}
})

print(decoded["result"]["messages"][0]["content"])

​
Response example

{
  "operation": "prompts.decode",
  "status": "ok",
  "result": {
    "messages": [
      {"role": "user", "content": "Hi"}
    ],
    "metadata": {"workflow": "support-reply"}
  },
  "metadata": {"decoded_by": "docs-demo"}
}

​
Errors

  • 404 → Payload expired or ID not found—ensure you decode before TTL expiry.
  • 410 → Attempted to replay a payload whose retention window has passed.

​
Notes

  • prompts_decode keeps your original metadata intact. Merge in new metadata sparingly to avoid overwriting audit fields.
  • Use this method in tandem with prompts_encode anytime you store compressed prompts (databases, queues, caches).