Resamples a grayscale image to remove radial and tangential lens distortion, using the camera matrix and five distortion coefficients you supply — typically from a planar calibration. Dimensions and camera matrix are preserved, so the corrected image drops into an existing pipeline without any change to geometry downstream. Because the output frame is unchanged, strongly distorted corners are pushed outside the image and are simply lost rather than being retained in a larger canvas.

Current content of up to 100 web pages in one call as Markdown, JSON-LD and tables; priced per page

Solves the hand-eye problem: given a series of gripper-to-base poses read from the robot and the matching target-to-camera poses seen by the camera, it recovers the rigid transform between camera and gripper that stays constant as the arm moves. This is what lets a robot act on what its camera sees. The poses must be genuinely synchronized — a pairing that is off by one sample produces a confident and entirely wrong transform, and nothing in the data reveals it.

Takes matched object and image points from multiple views of a planar calibration target and returns the camera matrix, five distortion coefficients and the RMS reprojection error that the fit actually achieved. The error figure is the point: it tells a caller whether the calibration is usable before anything is built on top of it, instead of handing back numbers that look plausible.

Given paired calibration observations and the intrinsics of both cameras, this computes the rotation and translation between them and derives everything a stereo pipeline needs downstream: rectification transforms for both views, both projection matrices, and the disparity-to-depth mapping. The RMS error in pixels says how well the rig geometry actually fits the data.

Runs semi-global block matching over a rectified grayscale stereo pair and returns per-pixel disparity together with the count of pixels that could actually be matched. Pixels with no reliable match stay unmatched rather than being interpolated, so occlusions and textureless regions are visible in the output instead of hidden by plausible-looking values. Disparities are positive and bounded at 16 pixels, which suits small-baseline rigs and close range.

Fits a 2D projective transform between two point sets with RANSAC at a caller-chosen reprojection threshold, and returns both the homography and the inlier mask. The mask is what makes this useful beyond a plain fit: bad correspondences are named rather than quietly averaged in, so a caller can re-run detection, drop a frame or raise an alarm. Suitable for document rectification, planar tracking and image stitching where the scene really is flat.

Which x402 API should my agent call? Live API finder for any task. Required: task. Returns up to 5 paid APIs that answered a fresh unpaid probe with a valid Base USDC 402, ranked by task fit, price and Bazaar-reported repeat buyers, each with live price, network and a ready-to-call request. Provider calls are paid separately. No live match, no charge. $0.005 USDC on Base. No account, no API key. Recommendations, not a safety guarantee. By Ontario Protocol.

Whale moves: every large taker order (default >= $100k) on Hyperliquid perps (top 40 markets by volume: BTC, ETH, SOL, HYPE...) in the last window, newest first, with side, USD size, avg price, taker address and explorer link, plus net whale buy/sell per coin (accumulating/distributing). Params min_usd (>=25000, default 100000), window (15m|1h|4h|24h|7d, default 1h), coin, side (buy|sell), limit. Captured live from the public Hyperliquid trade stream; built for polling

Web search API for AI agents, an Exa / Tavily alternative that costs less than both: Google-quality results re-ranked for authority (primary sources first), merged with papers (arXiv, Semantic Scholar), GitHub and news when the query needs them. Category (news, research paper, github, financial, company, pdf, docs), include/exclude domains, country, language, city, freshness, private mode. Each result: domain authority rank and type. ~1.7 s.

Ad cost benchmarks: CPM, CPC and CTR by platform (Meta, Google Search, Google Display, TikTok, LinkedIn) and industry, each figure with source report name, URL, as-of date and sample. Optional filters platform, industry, metric (cpm|cpc|ctr). Compiled from public benchmark reports (WordStream/LocaliQ, Triple Whale, DemandSense); not live auction prices

Monitor a topic: call with a query (web or news) and a watchId; each call returns only the results you have not seen before (newResults) plus the full list, so polling agents stop re-reading old links and spending tokens on them. No setup: the first call creates the watch. Hashed URLs only, kept 60 days; the query text is never stored.

DeFi intel bundle in one call: top 10 protocols by TVL, biggest 1d TVL gainers and losers (TVL over $100M), top 10 chains by TVL, top 10 stablecoins by supply with 1d/7d supply change, and top 10 risk-ranked yields. No parameters. Refreshed about every 30 min. Replaces the retired ds_crypto_defi_intel dataset

Google Scholar academic search. Returns papers with snippets, publication info, and citation counts.

Page to clean text. REQUIRED: url (example GET /v1/scrape/text?url=https://example.com). Any web page as clean, LLM-ready Markdown text with navigation, ads and boilerplate stripped; title, text and character count only. $0.003 per page. No charge if the page cannot be fetched

Price move alerts for agents that poll: ~155 Chainlink-priced coins whose price moved more than a threshold (default 5%) over 1h and/or 24h, biggest movers first, plus DeFi protocols whose TVL moved more than the threshold in 1d (TVL over $10M). Params threshold (0.5-100, default 5), window (1h|24h|both, default both), include_defi (default true), direction (up|down|both). Returns an empty list (still a valid answer) when nothing moved. Refreshed about every 30 min

Crypto market rows built from Chainlink USD price feeds read on-chain (Ethereum, Arbitrum, Base): coin name and symbol, rank, USD price, 1h/24h/7d/30d % change and 24h high/low from our own price history. Market cap, volume and FDV are not available on-chain and are null. Route id kept for compatibility. Filter with q=eth, up to 1,000 rows per call, JSON or CSV Formats: format=json (default) or format=csv (text/csv); optional q, limit, offset.

DeFi rows built from the lending markets and DEX pools we read on-chain: protocol TVL with 1d/7d change, category and chains, plus per-chain TVL rows (row_type protocol or chain). Coverage is the pools we track, not every contract; fees and revenue are null. Route id kept for compatibility. Filter with q=aave or q=chain, up to 1,000 rows per call, JSON or CSV Formats: format=json (default) or format=csv (text/csv); optional q, limit, offset.

Google Shopping product search. Returns products with prices, sellers, ratings, and links.

Fair Offer certification for an x402 API: every claim in your 402's fair-offer block checked against live behaviour. Your public receipts are reconciled against Base, your free sample is attested, the block is checked against the 0.2 schema and across two requests, and the listing price against the Bazaar. Pass or fail, the signed result is published at a public URL.