feat(extend): add Extend AI document processing integration

[waleedlatif1]

Summary

• Add Extend AI integration for document processing (parse/extract)
• Supports PDF, images, and Office documents via file upload or URL
• V1 (legacy) and V2 block variants following Reducto/Pulse patterns
• Internal API route with DNS validation, secure fetch, file resolution
• Configurable output format, chunking strategy, and engine selection

Type of Change

• New feature

Testing

Tested manually

Checklist

• Code follows project style guidelines
• Self-reviewed my changes
• Tests added/updated and passing
• No new warnings introduced
• I confirm that I have read and agree to the terms outlined in the Contributor License Agreement (CLA)

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
docs	Ready	Preview, Comment	Mar 31, 2026 11:17pm

[cursor]

PR Summary

Medium Risk
Medium risk because it introduces a new authenticated API route that resolves user-provided file inputs/URLs and calls an external service, which can impact security (SSRF/file access) and reliability despite added DNS/IP pinning validations.

Overview
Adds a new Extend AI integration for document parsing, including a new ExtendIcon, docs page, and inclusion in docs/landing icon mappings and integrations metadata.

Introduces new extend and extend_v2 blocks plus extend_parser/extend_parser_v2 tools, wiring them into the block/tool registries and defining inputs for document upload/URL (v1) vs file/reference (v2) along with output format, chunking, engine, and API key.

Adds an internal Next.js API route POST /api/tools/extend/parse that authenticates requests, resolves the provided file input to a URL, validates the Extend endpoint via DNS, and proxies the request via pinned-IP fetch while normalizing the response into id/status/chunks/blocks/pageCount/creditsUsed.

^{Written by Cursor Bugbot for commit eb90d96. Configure here.}

[greptile-apps]

Greptile Summary

This PR adds a new Extend AI document processing integration, following the established Reducto/Pulse patterns in the codebase. It introduces a V1 (hidden, legacy) block supporting both file upload and URL input, and a V2 (user-facing) block that supports file upload in basic mode and chained block file references in advanced mode. The implementation includes a secure internal API route with DNS validation and pinned-IP fetch, Zod request validation, structured error surfacing, and proper tool/block registry entries.

Key changes:

• apps/sim/app/api/tools/extend/parse/route.ts: New internal route with auth, DNS validation, and error surfacing — mirrors the Reducto route cleanly.
• apps/sim/blocks/blocks/extend.ts: V1 (hidden) + V2 (visible) block configs; V2 derives its subBlocks from V1 by replacing the URL input with a file-reference input.
• apps/sim/tools/extend/parser.ts: V1 and V2 tool configs with correct request body construction and transformResponse wiring.
• apps/docs/content/docs/en/tools/extend.mdx: Documentation page — the Output section incorrectly states "This tool does not produce any outputs" while the tool returns id, status, chunks, blocks, pageCount, and creditsUsed.

Confidence Score: 5/5

Safe to merge — all functional code is correct; the only finding is a documentation output section that needs updating.

All prior P1 concerns (error surfacing, icon consistency, V2 placeholder clarity) were addressed in 4830a4c. The remaining finding is a P2 documentation inaccuracy in the output section of extend.mdx, which does not affect runtime behavior.

apps/docs/content/docs/en/tools/extend.mdx — output section needs to list actual tool outputs.

Important Files Changed

Filename	Overview
apps/sim/app/api/tools/extend/parse/route.ts	Internal API route for Extend parse; mirrors Reducto pattern with DNS validation, pinned-IP fetch, Zod schema validation, and proper error surfacing.
apps/sim/tools/extend/parser.ts	Defines V1 and V2 tool configs; V2 spreads V1 and overrides params/request; UserFile shape is compatible with RawFileInputSchema validation in the route.
apps/sim/blocks/blocks/extend.ts	V1 block (hidden) supports file upload and URL; V2 block (visible) derives subBlocks from V1 removing the URL sub-block and adding a file-reference input for chaining.
apps/sim/tools/extend/types.ts	Well-typed interfaces for V1/V2 input and parser output; UserFile used for V2 file param which is compatible with the route's RawFileInputSchema.
apps/docs/content/docs/en/tools/extend.mdx	Documentation references the correct block type (extend_v2) and inputs, but the Output section incorrectly states the tool produces no outputs.

Sequence Diagram

sequenceDiagram
    participant Block as ExtendV2Block
    participant Tool as extendParserV2Tool
    participant Route as /api/tools/extend/parse
    participant FileUtils as resolveFileInputToUrl
    participant DNS as validateUrlWithDNS
    participant Extend as api.extend.ai/parse

    Block->>Tool: params (file, apiKey, outputFormat, chunking, engine)
    Tool->>Route: POST {apiKey, file, outputFormat?, chunking?, engine?}
    Route->>Route: checkInternalAuth
    Route->>FileUtils: resolveFileInputToUrl(file | filePath, userId)
    FileUtils-->>Route: fileUrl (signed/resolved URL)
    Route->>DNS: validateUrlWithDNS("https://api.extend.ai/parse")
    DNS-->>Route: resolvedIP
    Route->>Extend: POST {file:{fileUrl}, config?} Bearer apiKey (pinned IP)
    Extend-->>Route: {id, status, chunks, blocks, pageCount, creditsUsed}
    Route-->>Tool: {success: true, output: {...}}
    Tool-->>Block: transformResponse → ExtendParserOutput

Loading

_{Reviews (3): Last reviewed commit: "lint" | Re-trigger Greptile}

[waleedlatif1]

@greptile

[waleedlatif1]

@cursor review

[waleedlatif1]

@greptile

[waleedlatif1]

@cursor review

[cursor]

✅ Bugbot reviewed your changes and found no new issues!

Comment @cursor review or bugbot run to trigger another review on this PR