"High-Quality PDF Extraction to Text with Tokenization and Accurate Processing of Complex Layouts" matching MCP tools:

• prefetch_datasheets

  sheetsdata-mcp

  Trigger background datasheet extraction for multiple parts at once (up to 20). Non-blocking — returns immediately with the status of each part. Use this to warm up datasheets for a BOM before calling read_datasheet. Example: prefetch_datasheets(['TPS54302', 'ADS1115', 'LP5907']) If a part comes back 'no_source' on the first call, retry prefetch for that MPN once after 10-30s — the URL resolver is retriable and often finds a source on the second pass. If still 'no_source', use request_datasheet_upload + confirm_datasheet_upload to attach your own PDF (org-private). Part numbers must be specific MPNs (e.g. 'STM32F446RCT6', 'TPS54302DDCR') or LCSC numbers (e.g. 'C2837938'). Do NOT pass bare values ('100nF', '10K'), descriptions, BOM reference designators, test points, or board/module names — see the server instructions for the full rule set. When a BOM has values-only rows, use search_parts first to resolve each to an MPN. DATASHEET STATUS VALUES: - 'ready' — extracted and indexed; call read_datasheet, search_datasheets, or analyze_image. - 'extracting' / 'in_progress' / 'queued' / 'pending' — extraction running or scheduled. Poll check_extraction_status every 5-10s until 'ready' or 'failed'. Typical time: 30s-2min. - 'not_extracted' — known part but datasheet hasn't been fetched yet. Trigger it via prefetch_datasheets (cheapest) or by calling read_datasheet (auto-triggers on first read). - 'no_source' — we couldn't find a public datasheet URL for this MPN. First, retry prefetch_datasheets in 10-30s (the URL resolver re-runs and often finds a source on the second pass). If still 'no_source', the agent can upload the PDF manually via request_datasheet_upload + confirm_datasheet_upload (see those tools). Org-uploaded datasheets are private to the org. - 'unsupported' — PDF exists but can't be extracted (scanned image-only, encrypted, or corrupted). Upload a clean text-based PDF via request_datasheet_upload to override. - 'failed' / 'error' — extraction errored. The response includes the error reason. Retry via prefetch_datasheets or escalate to support. - 'rejected' — input wasn't a real MPN (bare value like '100nF', description, or reference designator). Fix the input and re-call. - 'deduplicated' — another part in the family already has this datasheet; same content is returned under the primary MPN.

  Connector

• mindee_extract_document

  mindee

  Submit a document (PDF or image) to a Mindee extraction model and return the structured fields. Provide EITHER document_url (a public URL) OR file_base64 (+ filename). This enqueues an inference and polls until it completes (up to ~30s); if it is still processing it returns a job_id you can poll with mindee_get_job then read with mindee_get_inference. NOTE: consumes Mindee API credits (paid, billed per page). V2 API: POST /v2/inferences/enqueue.

  Connector

• check_extraction_status

  sheetsdata-mcp

  Check the extraction status of one or more parts. Free. Each entry includes the current extraction step, elapsed seconds, and document ID. Use after prefetch_datasheets or after read_datasheet triggers a new extraction. Recommended polling cadence: every 5-10 seconds. Extraction typically takes 30s-2min for new parts, so polling faster than every 5s wastes calls. Stop polling once status is 'ready', 'failed', 'no_source', or 'unsupported'. DATASHEET STATUS VALUES: - 'ready' — extracted and indexed; call read_datasheet, search_datasheets, or analyze_image. - 'extracting' / 'in_progress' / 'queued' / 'pending' — extraction running or scheduled. Poll check_extraction_status every 5-10s until 'ready' or 'failed'. Typical time: 30s-2min. - 'not_extracted' — known part but datasheet hasn't been fetched yet. Trigger it via prefetch_datasheets (cheapest) or by calling read_datasheet (auto-triggers on first read). - 'no_source' — we couldn't find a public datasheet URL for this MPN. First, retry prefetch_datasheets in 10-30s (the URL resolver re-runs and often finds a source on the second pass). If still 'no_source', the agent can upload the PDF manually via request_datasheet_upload + confirm_datasheet_upload (see those tools). Org-uploaded datasheets are private to the org. - 'unsupported' — PDF exists but can't be extracted (scanned image-only, encrypted, or corrupted). Upload a clean text-based PDF via request_datasheet_upload to override. - 'failed' / 'error' — extraction errored. The response includes the error reason. Retry via prefetch_datasheets or escalate to support. - 'rejected' — input wasn't a real MPN (bare value like '100nF', description, or reference designator). Fix the input and re-call. - 'deduplicated' — another part in the family already has this datasheet; same content is returned under the primary MPN.

  Connector

• get_part_details

  sheetsdata-mcp

  Get full details for a specific electronic component by manufacturer part number (MPN) or LCSC number. Returns specs, pricing, and stock from all configured providers, plus the cached datasheet summary if available. Includes datasheet_status and available_sections when ready. Set prefetch_datasheet=true to trigger background extraction — no extra charge. Use after search_parts to drill into a specific result. The part_number must be a specific manufacturer part number (e.g. 'TPS54302DDCR', 'STM32F446RCT6') or LCSC number (e.g. 'C2837938'). Do NOT pass bare component values ('100nF', '10K'), descriptions ('buck converter'), or reference designators ('R1', 'U3'). DATASHEET STATUS VALUES: - 'ready' — extracted and indexed; call read_datasheet, search_datasheets, or analyze_image. - 'extracting' / 'in_progress' / 'queued' / 'pending' — extraction running or scheduled. Poll check_extraction_status every 5-10s until 'ready' or 'failed'. Typical time: 30s-2min. - 'not_extracted' — known part but datasheet hasn't been fetched yet. Trigger it via prefetch_datasheets (cheapest) or by calling read_datasheet (auto-triggers on first read). - 'no_source' — we couldn't find a public datasheet URL for this MPN. First, retry prefetch_datasheets in 10-30s (the URL resolver re-runs and often finds a source on the second pass). If still 'no_source', the agent can upload the PDF manually via request_datasheet_upload + confirm_datasheet_upload (see those tools). Org-uploaded datasheets are private to the org. - 'unsupported' — PDF exists but can't be extracted (scanned image-only, encrypted, or corrupted). Upload a clean text-based PDF via request_datasheet_upload to override. - 'failed' / 'error' — extraction errored. The response includes the error reason. Retry via prefetch_datasheets or escalate to support. - 'rejected' — input wasn't a real MPN (bare value like '100nF', description, or reference designator). Fix the input and re-call. - 'deduplicated' — another part in the family already has this datasheet; same content is returned under the primary MPN.

  Connector

• read_datasheet

  sheetsdata-mcp

  Read from a component's datasheet. Two modes: **Section mode** (default): Returns a named section. Start with section='summary' to get an overview and a list of available_sections. Then request specific sections by name. Section names are dynamic — any heading in the actual datasheet works (e.g. 'register_map', 'i2c_interface', 'power_management'). If a section name isn't found, automatically falls back to search mode. **Search mode**: Semantic search within the part's datasheet. Best for targeted questions (register bit fields, I2C config, specific specs). Use when you need to find specific information rather than a whole section. First call for a new part triggers extraction (30s-2min). Subsequent calls are cached. **Datasheet vs Reference Manual**: Manufacturer datasheets cover high-level specs, pinout, absolute maximum ratings, and package info. For microcontrollers (STM32, nRF52, RP2040), register-level programming details (I2C CR1/CR2, DMA config, interrupt bits) are in a separate Reference Manual, not the datasheet. The summary's available_sections will show what's actually present. The part_number must be a specific manufacturer part number (e.g. 'TPS54302', 'STM32F446RCT6') or LCSC number (e.g. 'C2837938'). Do NOT pass bare component values ('100nF', '10K'), descriptions, or reference designators. DATASHEET STATUS VALUES: - 'ready' — extracted and indexed; call read_datasheet, search_datasheets, or analyze_image. - 'extracting' / 'in_progress' / 'queued' / 'pending' — extraction running or scheduled. Poll check_extraction_status every 5-10s until 'ready' or 'failed'. Typical time: 30s-2min. - 'not_extracted' — known part but datasheet hasn't been fetched yet. Trigger it via prefetch_datasheets (cheapest) or by calling read_datasheet (auto-triggers on first read). - 'no_source' — we couldn't find a public datasheet URL for this MPN. First, retry prefetch_datasheets in 10-30s (the URL resolver re-runs and often finds a source on the second pass). If still 'no_source', the agent can upload the PDF manually via request_datasheet_upload + confirm_datasheet_upload (see those tools). Org-uploaded datasheets are private to the org. - 'unsupported' — PDF exists but can't be extracted (scanned image-only, encrypted, or corrupted). Upload a clean text-based PDF via request_datasheet_upload to override. - 'failed' / 'error' — extraction errored. The response includes the error reason. Retry via prefetch_datasheets or escalate to support. - 'rejected' — input wasn't a real MPN (bare value like '100nF', description, or reference designator). Fix the input and re-call. - 'deduplicated' — another part in the family already has this datasheet; same content is returned under the primary MPN.

  Connector

• confirm_datasheet_upload

  sheetsdata-mcp

  Confirm a datasheet upload started via request_datasheet_upload. Pass the upload_token you got back from the request step. The server downloads the uploaded bytes, re-hashes to verify integrity, validates that it's a real PDF with the MPN on the first page, creates the private Document + Component records, charges the upload fee (50¢), and queues extraction. Success response: document_id, mpn, sha256, file_size_bytes, status='pending'. Poll check_extraction_status with the MPN to wait for extraction to finish (30s-2min typically). Failure modes: - 'upload_not_found' — no bytes at the upload URL yet. Retry your curl upload. - 'sha256_mismatch' — uploaded bytes hash differs from expected_sha256. Re-compute the hash and re-request. - 'invalid_pdf' — bytes aren't a parseable PDF. No charge. - 'mpn_not_in_pdf' — MPN (or its stem) isn't on the first page. Either you uploaded the wrong file or it's a scanned image-only PDF. No charge. - 'token_expired' — upload token is older than 15 minutes. Restart via request_datasheet_upload.

  Connector

• aetherwave_generate_image

  AetherWave Studio

  Generates one or more images from a text prompt (T2I) or a text prompt + reference image(s) (I2I). Submits the job, polls until terminal, and returns the final image URLs. Default model is 'grok-imagine-t2i' (fast, 6 images per generation, 5 credits). Use list_image_models to see the full lineup with pricing. For I2I, pass `referenceImages` as an array of public image URLs and pick a model with I2I support (e.g. 'grok-imagine-i2i', 'wan-2.5-spicy-i2i'). ## Model selection guide (when the user does not specify a model) Default: `grok-imagine-t2i` (5 cr, 6 outputs per call, fast, general purpose). **Strong recommendation: when a single high-quality output is what's wanted** (most agent / one-shot workflows), prefer `gpt-image-2-t2i` (9 cr @ 1K / higher @ 2K, single deterministic image, best general quality across realism, illustration, typography, and composition; supports up to 2K resolution and most aspect ratios including auto). This is the front-runner for serious creative output where you don't need to pick from 6 variations. Pick a different model when the prompt has these signals: - "single best result" / "one image" / production / no time to pick from variations -> `gpt-image-2-t2i` (9 cr, 1 output, top general quality) - "photoreal" / "photo of" / "realistic" -> `gpt-image-2-t2i` (9 cr, best general realism) or `imagen-4` (12 cr, very high quality) or `z-image-turbo` (3 cr, fastest) - "highest quality" / "premium" / no budget -> `gpt-image-2-t2i` at 2K, or `grok-imagine-quality-t2i` (16 cr @ 1K, 22 cr @ 2K), or `imagen-4-ultra` - Text inside the image (signs, posters, typography) -> `ideogram-v3-t2i` (best in class) or `gpt-image-2-t2i` (also strong) - Artistic / painterly / stylized -> `midjourney-t2i` - Album art / cover art -> `gpt-image-2-t2i` for one strong image; `grok-imagine-t2i` for 6 variations to choose from; `seedream-v4-t2i` if 4K wanted - Logo or design with embedded text -> `ideogram-v3-t2i` - NSFW / adult / explicit -> `wan-2.5-spicy-t2i` (auto-tags creation as 18+; routes to adult gallery) - Cheapest possible / quick test -> `z-image-turbo` (3 cr) - Multiple variations to compare -> keep `grok-imagine-t2i` (6 outputs default) or use `numImages` on a multi-output model For I2I (reference image provided): prefer the dedicated `aetherwave_edit_image` tool for "change something in this image" intent. Use `aetherwave_generate_image` with I2I models only when you specifically want style transfer (`midjourney-i2i`), premium quality (`grok-imagine-quality-i2i`), or adult content (`wan-2.5-spicy-i2i`). Always pass an explicit `aspectRatio` (e.g. "1:1" for square album art, "16:9" for video thumbnails, "9:16" for shorts/reels). Some upstream providers reject submissions with no aspect ratio. Ask the user only when: - The prompt contradicts itself (e.g., "highest quality but cheapest") - The user requested "the best model" with no context, surface 2-3 options with tradeoffs - A single generation would cost more than 20 credits and the user has not confirmed

  Connector

• parse_pdf_to_text

  Nordic Financial MCP

  Download a PDF from a URL and extract all text content, page by page. Use this to read the full text of a specific document — for example, an annual report PDF linked from a search_filings result. Best combined with search_filings: use search_filings to locate the document, then parse_pdf_to_text for the full text. Do not use for PDFs that are already well-represented in the database — search_filings is faster and returns pre-ranked, relevant excerpts. Not suitable for scanned (image-only) PDFs without embedded text; those pages will be returned as "(no extractable text)". Args: pdf_url: Direct HTTPS URL to the PDF file, e.g. https://example.com/report.pdf. Must be publicly accessible; authentication-protected URLs will fail. Returns: All text from the PDF with "--- Page N ---" separators between pages. Returns an error string if the download fails, the URL does not point to a valid PDF, or the document exceeds the 60-second download timeout.

  Connector

• deckextract

  DeckExtract

  Extract a deck from a DocSend or Papermark sharing link. Returns a temporary download URL and a readable `deckextract://deck/...` resource for the PDF/PPTX (or a ZIP for data rooms — a DocSend Space or Papermark data room counts as one extraction like any other link). Pass `analyze: true` (requires a DeckExtract Pro account) to also return structured deck data. Decks that email the viewer a verification step return resume tokens with retry instructions: fetch the emailed 6-digit code and retry with `otp` + `otpSessionId` (Papermark), or retry with `url` set to the emailed confirmation link + `sessionId` (DocSend). Typical extraction takes 15-90 seconds; the public API is rate limited to 5 extractions per IP per 30 minutes.

  Connector

• optimize_flow

  CNAPS.ai MCP Server

  Analyze a flow for performance and cost optimization opportunities. Returns rule-based suggestions such as moving upscale nodes to the end of the flow, avoiding resolution overflow, removing redundant processing, and choosing better-performing models. Each suggestion carries a structured patch (move_node, insert_node, replace_model) describing the change. These are advisory: no MCP tool edits a flow's graph, so apply them by rebuilding the flow with create_flow in the corrected order, or relay them to the user for the cnaps.ai editor. update_flow changes node parameters only. Use this before running a flow or while iterating on its design. Set include_llm_analysis=true to also ask Haiku for complex-pattern refinements.

  Connector

• upload_asset

  SlideForge

  Upload assets for PowerPoint (.pptx) generation: company template, logo, image, or document — or AI-generate an image. Purposes: • logo — company logo for chrome (PNG/JPG/SVG, max 5MB) → logo_id • image — image for the Image component (max 10MB) → asset_id • theme — company template PPTX → theme_id; slides with it render NATIVELY on the template (masters/layouts/chrome) • generate_image — AI-generate via `prompt` → asset_id ($0.05) • translate — PPTX to translate → deck job_id ($0.02/slide; requires `target_language`) • pdf — PDF → editable slides; pass `target_language` to also translate • recreate — image OF a slide → editable PPTX slide ($0.10; honest annotate/preserve fallback, refusals free). Use `image` to just place a picture Files >3MB (pdf/translate/theme) — and recreate on chat hosts — omit `data`: a drop-zone appears in the result card; bytes never pass through the agent.

  Connector

• search_nlp

  MedRates

  Natural language search for medical procedure prices. Understands free-text queries like: - "MRI brain near San Jose with Blue Cross PPO" - "How much does a colonoscopy cost in Palo Alto?" - "knee replacement, no insurance, Mountain View" Extracts procedure, location, and insurance from the query, resolves CPT/DRG codes (using static synonyms + LLM), geocodes the city, and searches with optional geo-filtering and payer matching. You can provide structured fields (lat/lng, payer, setting) to override or supplement what the NLP extraction detects from the query text. NOTE: Results are from US HOSPITALS only — not non-US providers, independent imaging centers, ambulatory surgery centers (ASCs), or other freestanding facilities. For outpatient procedures (MRIs, CTs, minor surgeries), independent facilities may offer lower prices than hospitals. Args: query: Natural language query describing what you're looking for. radius_miles: Search radius from the detected city (default 25 miles). code_type: Filter by code type: "CPT", "HCPCS", "MS-DRG". setting: Filter by clinical setting: "inpatient" or "outpatient". lat: Override latitude (e.g. from browser geolocation). Skips geocoding. lng: Override longitude (e.g. from browser geolocation). Skips geocoding. zip_code: 5-digit ZIP to search near — alternative to lat/lng. payer: Insurance payer name (e.g. "Blue Cross"). Overrides NLP extraction. plan_type: Plan type (e.g. "PPO", "HMO"). Overrides NLP extraction. limit: Max results (default 25). Returns: JSON with extracted entities (procedure, city, insurance), resolved codes, and matching charge items with prices and hospital info. Only high-confidence results (with at least one usable price) are included. Each result includes last_updated (ISO date of the per-hospital MRF ingest) and mrf_date (ISO date the hospital self-reported in the MRF file). When all results are filtered out, filtered_low_confidence=true is set.

  Connector

• extract_contract_from_url

  transaction-coordinator

  Extract structured transaction data from a contract at a URL. Downloads the document, extracts text (with OCR fallback for scanned PDFs), and runs PrimaCoda's contract-extraction prompt to return parties, addresses, dates, prices, and key contract fields. Use this when an agent has the contract hosted somewhere (Dropbox, Google Drive direct download, Square Space, etc.) and wants to skip the upload step. For multi-document deals (purchase + addenda + disclosures), use the PrimaCoda dashboard's batch upload — this tool handles ONE document. Args: pdf_url: Direct download URL for the contract (PDF, DOCX, TXT, or image). Must be reachable from the PrimaCoda server. Google Drive "shared link" URLs work if set to "anyone with link"; other share URLs may need their direct-download form. api_key: Your PrimaCoda MCP API key (starts 'pck_').

  Connector

• get_analysis_status

  hypathesis

  Check the processing status of an uploaded paper. Poll this tool after uploading a PDF until status is 'Ready' before calling get_variable_relationships. Args: file_id: The file_id returned by the /upload endpoint. authorization: Optional. API key as 'Bearer hk_...' or 'hk_...'. Returns: { "status": "Processing" | "Ready" | "Empty" | "Ineligible" | "Pending", "edges_count": int, "variables_count": int }

  Connector

• extract_claims

  groundcheck

  PURPOSE: Split text into independently checkable ATOMIC factual claims — the cheap first step of a verification loop (extract -> ground -> attest). Returns {claims: [...], count, input_sha256} plus a signed receipt bound to the input hash. GUIDELINES: Call when you want to see WHICH claims a document makes before paying to ground them, to budget a verification pass (extract everything, then verify_claim only the claims that matter to your decision), or to prove later exactly which claims were pulled from exactly which text (the receipt binds both). Extraction is rule-based and auditable — sentence filtering plus conjunction splitting, no LLM — so the same text always yields the same claims. Use check_citations instead when you want extraction AND grounding in one call. PARAMETERS: text — the prose to decompose. max_claims — 1..50, default 20. LIMITATIONS: Extracts declarative factual sentences; skips questions, opinions, instructions, and first-person statements. Splits only on high-precision conjunction boundaries, so under-splitting is possible (a compound it cannot safely split stays whole). Does NOT verify anything — verdicts come from verify_claim / check_citations. Paid per call (x402), cheapest tool on this server. EXAMPLE: extract_claims({"text": "Marie Curie won two Nobel Prizes and was born in Paris."}) -> {count: 2, claims: ["Marie Curie won two Nobel Prizes", "was born in Paris."]}

  Connector

• glim_detect_ai

  glim.sh

  Detect AI-generated text. Scores any text for AI-authorship likelihood and returns an overall verdict (AI / human / mixed) with confidence, the AI/human/AI-assisted fractions, and a segment-by-segment breakdown showing exactly which parts read as AI-written. Use it to verify whether content (comments, articles, profiles) is AI-generated, or to check text before publishing to see which segments would trip AI detectors - revise the flagged segments and re-check. tier='premium' uses the most accurate detection class, additionally flags humanized text (AI output run through paraphrasing/'humanizer' tools), and is substantially more robust to evasion, at ~10x the price of 'standard'. Cost scales with text length: ~$0.06 per 1,000 words standard, ~$0.60 premium; minimum $0.06. Texts under ~100 words are automatically analyzed with the premium class at the standard price.

  Connector

• voice_intelligence_run

  StackFast FractWin Expert Brain

  Run the voice intelligence orchestrator over any input: an uploaded or recorded audio/video file (base64 in file), a public audio or video URL (audio_or_video_url), a PDF (pdf), an image (image), or raw text/notes (text and/or notes). Provide exactly ONE input carrier — audio_or_video_url, file, pdf, image, or text/notes. The input is normalized to a transcript via the existing transcription rails, then three composed outputs are returned: a condensed summary (with action items, decisions, and a next step), a full-verbatim polished actual script (filler removed, wording preserved, never paraphrased; call sources return a masked summary instead), and an authority report (content flywheel plan, plus an owner-facing business report when business context is provided). All outward outputs are mask-gated.

  Connector

• render_html_to_pdf

  pdfzen

  Prepare a paid PDF render from arbitrary Handlebars-flavoured HTML. Use only when no starter fits (one-off layouts, custom branding). Prefer render_template_to_pdf when a starter matches. Validates your HTML and returns the exact, ready-to-execute HTTP request to run against pdfzen's render endpoint — POST /v402/render/pdf (x402, $0.006 USDC on Base, no API key) or POST /v1/render/pdf (pdfzen API key). pdfzen renders are executed over HTTP, not streamed in-band over MCP; this tool is the bridge.

  Connector

• request_datasheet_upload

  sheetsdata-mcp

  Request a signed URL to upload a datasheet PDF for a component whose datasheet we don't have. Use this when search_parts / get_part_details / prefetch_datasheets return datasheet_status='no_source' (and a retry didn't help) or 'unsupported'. Free — the upload fee is only charged on confirm_datasheet_upload after we validate the file. Flow (3 steps): 1. Call request_datasheet_upload with the MPN, the file's SHA-256, and its byte size. You get back an upload_url, upload_method ('PUT'), upload_headers, and an opaque upload_token. 2. Upload the PDF directly to the returned URL with curl: `curl -X PUT -H 'Content-Type: application/pdf' --data-binary @file.pdf "$UPLOAD_URL"` (add any headers from upload_headers). 3. Call confirm_datasheet_upload with the upload_token. Server verifies the bytes, re-hashes, checks for the MPN on the first page, charges the upload fee (50¢), and queues extraction. Returns document_id + status='pending'. Validation rules (checked at confirm time, refunded on failure): - File must be a valid PDF (magic bytes + parseable). - Actual SHA-256 must match expected_sha256. - Actual byte size must match size_bytes (±0). - MPN or its core stem must appear in the first page text (catches wrong-file uploads). Scanned image-only PDFs will fail this check — upload a text-based PDF. - Max 50MB per file. No dev-kit manuals / BOB schematics / app-notes as datasheets — use the matching MPN's actual datasheet. Uploaded datasheets are scoped to your organization (private). They satisfy read_datasheet, search_datasheets, check_design_fit, and analyze_image for your org's tokens only. Tokens expire after 15 minutes. If upload fails or times out, just call request_datasheet_upload again.

  Connector

• get_510k_summary

  Scite

  Fetch the full text of a single FDA 510(k) summary PDF by document ID. Use this after `search_510k_summaries` or `search_device510k` when you need the complete narrative text of a 510(k) summary, not just search snippets or structured metadata. Returns the full extracted text organized by page. **Parameters:** - id: Document identifier (the K number, e.g. `K192757`). Can be obtained from either `search_510k_summaries` or `search_device510k` results. **Returns:** The full-text content of the 510(k) summary PDF, organized by page, with file metadata and ontology tags.

  Connector