feat: add Resemble Detect + Intelligence tool

[devshahofficial]

Resemble Detect + Intelligence tool

Adds a Resemble tool node + Resemble API credential for Resemble AI
media safety: deepfake detection, media intelligence, and watermarking. Scope is
Detect + Intelligence only (no TTS / voice cloning).

What it adds

• packages/components/nodes/tools/Resemble/ — node exposing a toolkit of
  DynamicStructuredTools to an agent:
  • resemble_detect — deepfake detection (audio/image/video) with toggles for
    intelligence, source tracing, visualize, reverse-search, OOD, zero-retention, model type
  • resemble_intelligence — transcription, translation, speaker info, emotion, misinformation
  • resemble_watermark_detect / resemble_watermark_apply
• packages/components/credentials/ResembleApi.credential.ts — API-key credential.

Implementation

• HTTP logic is in a dependency-free client.ts (global fetch), wrapped by core.ts
  into LangChain tools (@langchain/core/tools + zod/v3).
• Bearer auth; async detection polls to completion; watermark uses Prefer: wait.
• Large inline base64 artifacts (e.g. heatmaps) are stripped from tool output.

Testing

The client was run live against the Resemble API: media intelligence and watermark
detect/apply confirmed green; detection uses the same submit→poll path.

[gemini-code-assist]

Code Review

This pull request introduces a new integration for Resemble AI Detect + Intelligence, including credential configuration, a tool node, an HTTP client, and LangChain tool definitions. A critical issue was identified in the HTTP client (client.ts) where attempting to read the response as text after a failed JSON parsing attempt will throw a TypeError because the response stream is already consumed. It is recommended to read the response as text first and then parse it.

Important

The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.

[gemini-code-assist]

In the Fetch API, once a response body has been read (even if parsing fails), the stream is fully consumed and locked. Attempting to call res.text() in the catch block after res.json() fails will throw a TypeError: body stream already read instead of retrieving the raw text.

To fix this, read the response as text first, and then attempt to parse it as JSON.

    const text = await res.text()
    let data: any
    try {
        data = JSON.parse(text)
    } catch {
        data = { raw: text }
    }

[devshahofficial]

Feedback addressed: response bodies are now read once via text() then JSON.parse (the res.text()-after-failed-res.json() TypeError), plus polling hardening: timeouts raise instead of returning a non-terminal payload, the watermark-apply result is detected via watermarked_media presence, and intelligence polls the documented GET /intelligences/{uuid} route. All four operations re-tested against the live API. Synced with main; tsc and prettier clean.

[devshahofficial]

@HenryHengZJ gentle ping for a review when you have a moment — tsc/lint clean and review feedback addressed. We'd also be open to a Resemble ↔ Flowise partnership: dev.shah@resemble.ai. Thanks!