https://agentutility.ai/updates Daily notes on x402, the agent economy, and the agentutility endpoint portfolio. By x.com/agentutility_ai. en-us Sat, 15 Aug 2026 12:00:00 GMT agentutility-site https://agentutility.ai/updates/web-search-repeat-research-recipe https://agentutility.ai/updates/web-search-repeat-research-recipe Sat, 15 Aug 2026 12:00:00 GMT The new web-search recipe caps one paid request, records a local run claim, and keeps the query, result count, and recency window stable for a later source-list comparison. web-search research retention mcp noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/image-generate-pro-repeatable-assets https://agentutility.ai/updates/image-generate-pro-repeatable-assets Fri, 14 Aug 2026 12:00:00 GMT The premium image route now explains how to preserve a chosen tier, canvas, and seed after an asset is approved. image-generate-pro image-production reproducibility noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/agentutility-cli https://agentutility.ai/updates/agentutility-cli Fri, 14 Aug 2026 12:00:00 GMT npx @agentutility/cli handles wallet creation, catalog search, price checks, and the x402 payment handshake. Two commands from zero to a paid call. cli tooling x402 wallets noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/image-generate-price https://agentutility.ai/updates/image-generate-price Thu, 13 Aug 2026 12:00:00 GMT The repeatable hosted-PNG route and its inherited image-generation aliases now charge $0.011 per call. The request and response contract is unchanged. image-generate pricing image-production noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/image-generate-ai-image-generator-copy https://agentutility.ai/updates/image-generate-ai-image-generator-copy Thu, 13 Aug 2026 12:00:00 GMT The image-generate listing and Synthforge MCP tool now lead with the AI image generator job while keeping the existing request and hosted-PNG response unchanged. image-generate image-generation mcp noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/writing-a-cluster-design-spec https://agentutility.ai/updates/writing-a-cluster-design-spec Wed, 12 Aug 2026 12:00:00 GMT Our cluster spec names the buyer, x402 wedge, lead endpoint, build order, and kill criteria before coding starts, keeping paid agent tools tied to a callable job. process specs product noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/image-generate-repeatable-assets https://agentutility.ai/updates/image-generate-repeatable-assets Tue, 11 Aug 2026 12:00:00 GMT A runnable image-generate request that keeps prompt, dimensions, tier, and seed together so an agent can reproduce an approved asset instead of starting over. image-generate image-production reproducibility noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/endpoint-deprecation-without-tears https://agentutility.ai/updates/endpoint-deprecation-without-tears Mon, 10 Aug 2026 12:00:00 GMT Retire paid x402 routes safely with a dated notice, an updated registry record, a free 410 response, and a machine-readable link that sends agents to the successor. versioning deprecation api-design noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/translate-recurring-workflows https://agentutility.ai/updates/translate-recurring-workflows Sun, 09 Aug 2026 12:00:00 GMT The translate listing now makes its recurring agent use cases and response contract explicit. translation support documentation localization noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/browser-session-update-2026-08-08 https://agentutility.ai/updates/browser-session-update-2026-08-08 Sat, 08 Aug 2026 12:00:00 GMT browser-session still costs 0.03 USDC per call, but its catalog contract now caps public URLs at 4,096 characters and marks two legacy controls as compatibility inputs. x402 update browser-session web-probe noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/image-generate-repeatable-png-workflow https://agentutility.ai/updates/image-generate-repeatable-png-workflow Fri, 07 Aug 2026 12:00:00 GMT The image-generate listing now makes the hosted-PNG response and seed-based repeat requests easier to spot before an agent pays. synthforge image-generation mcp noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/whois-rdap-repeat-workflow https://agentutility.ai/updates/whois-rdap-repeat-workflow Thu, 06 Aug 2026 12:00:00 GMT The WHOIS/RDAP comparison page now gives agents a runnable $0.005 lookup and the fields to retain for a later domain check. web-probe whois rdap retention noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/pricing-experiments https://agentutility.ai/updates/pricing-experiments Wed, 05 Aug 2026 12:00:00 GMT Data from a 73-endpoint repricing shows why a tenth of a cent can't create demand, one-cent calls suit repeat work, and ten-cent calls need explicit user value. pricing experiments data noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/the-bazaar-ranking-signals https://agentutility.ai/updates/the-bazaar-ranking-signals Tue, 04 Aug 2026 12:00:00 GMT Coinbase Bazaar ranks x402 endpoints by query relevance and paid usage signals. Here’s how call volume, buyer reach, recent activity, and metadata affect discovery. bazaar ranking discovery noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/ip-asn-for-anti-abuse https://agentutility.ai/updates/ip-asn-for-anti-abuse Mon, 03 Aug 2026 12:00:00 GMT Use WebProbe ip-asn to map request IPs to network owners, spot traffic from hosting or VPN ranges, and add an agent-infrastructure signal without blocking legitimate users. web-probe anti-abuse asn noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/wordmint-vs-direct-llm-call https://agentutility.ai/updates/wordmint-vs-direct-llm-call Sun, 02 Aug 2026 12:00:00 GMT A direct OpenAI call can cost one-tenth of Wordmint summarize, but Wordmint’s fixed $0.005 x402 price can make agent budgets and USDC settlement easier to manage. wordmint llm comparison noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/agent-card-fields-explained https://agentutility.ai/updates/agent-card-fields-explained Sat, 01 Aug 2026 12:00:00 GMT Read an ERC-8004 registration file field by field, see how routers and explorers use each value, and avoid URL or image-hosting mistakes that can make a registered agent disappear. erc-8004 agent-cards schema noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/translate-text-recovery https://agentutility.ai/updates/translate-text-recovery Fri, 31 Jul 2026 12:00:00 GMT The canonical translation route now uses the fast provider first, accepts fenced JSON replies, and reports provider exhaustion as a dependency failure. translate reliability localization noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/translate-text-localization https://agentutility.ai/updates/translate-text-localization Thu, 30 Jul 2026 12:00:00 GMT translate-text now spells out the translation jobs it handles for agents that repeatedly localize messages, support replies, and documents. translate localization support noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/the-100-call-stress-test https://agentutility.ai/updates/the-100-call-stress-test Wed, 29 Jul 2026 12:00:00 GMT A 100-call burst against token-count finished without a failed purchase. Here’s the latency spread, the $0.50 bill, and the client-side limit that spoiled our first run. performance x402 load-testing noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/twitter-search-recovery https://agentutility.ai/updates/twitter-search-recovery Tue, 28 Jul 2026 12:00:00 GMT Twitter search now makes its query handling and labeled public-index fallback clear before an agent pays. reliability twitter-search x-search noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/music-generate-prompt-only-fix https://agentutility.ai/updates/music-generate-prompt-only-fix Tue, 28 Jul 2026 12:00:00 GMT Prompt-only and lyric-driven music generation now defaults to ACE-Step 1.5 with supported duration tiers. synthforge music-generation reliability noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/kimi-k3-x402 https://agentutility.ai/updates/kimi-k3-x402 Tue, 28 Jul 2026 12:00:00 GMT Two capped endpoints now expose Venice's Kimi K3 model for chat and structured agent planning. model-router kimi-k3 agent-planning noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/search-web-answer-engine-discovery https://agentutility.ai/updates/search-web-answer-engine-discovery Mon, 27 Jul 2026 12:00:00 GMT The search-web listing now makes clear that it supplies live source results for answer engines and RAG flows, without implying it writes a synthesized answer. web-probe answer-engine rag search noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/lost-buyer-recovery https://agentutility.ai/updates/lost-buyer-recovery Mon, 27 Jul 2026 12:00:00 GMT The endpoints behind several recently quiet buyer cohorts now have tighter fallback behavior, faster translation, corrected quality scoring, and lower prices. reliability search translation whois x-search noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/image-generate-prompt-to-png-discovery https://agentutility.ai/updates/image-generate-prompt-to-png-discovery Mon, 27 Jul 2026 12:00:00 GMT The image-generate listing now spells out the prompt controls, style choices, and response fields agents get when they need a hosted visual asset. synthforge image-generation discovery noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/hacker-news-search-discovery https://agentutility.ai/updates/hacker-news-search-discovery Mon, 27 Jul 2026 12:00:00 GMT The hacker-news-search listing now calls out startup news, YC news, Show HN, and the fields returned before an agent makes a paid request. web-probe hacker-news startup-news discovery noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/dictionary-discovery https://agentutility.ai/updates/dictionary-discovery Mon, 27 Jul 2026 12:00:00 GMT The dictionary route gives agents a direct discovery term for English definitions, pronunciations, usage examples, and synonyms. wordmint dictionary definitions pronunciation noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/paid-endpoint-recovery https://agentutility.ai/updates/paid-endpoint-recovery Sun, 26 Jul 2026 12:00:00 GMT Web search, slop filtering, and content authenticity now return labeled partial results when one provider times out. reliability web-search content-authenticity compose noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/x402-error-shapes https://agentutility.ai/updates/x402-error-shapes Fri, 24 Jul 2026 12:00:00 GMT Separate payment challenges from replay failures, normalize x402 error bodies, and give agents a retry policy that prevents duplicate charges while preserving useful details. x402 errors api-design noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/crypto-news-discovery https://agentutility.ai/updates/crypto-news-discovery Thu, 23 Jul 2026 12:00:00 GMT The crypto-news listing now makes its search scope, lookback, returned fields, and common agent use cases plain before a caller pays. web-probe crypto news discovery noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/crypto-headlines-search-intent https://agentutility.ai/updates/crypto-headlines-search-intent Thu, 23 Jul 2026 12:00:00 GMT The crypto-headlines listing now makes its query, lookback, returned fields, and data-only scope clear. endpoints discoverability crypto-headlines noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/working-capital-aliases https://agentutility.ai/updates/working-capital-aliases Wed, 22 Jul 2026 12:00:00 GMT Two new endpoint names expose the existing cash conversion calculator under exact buyer search terms. endpoints finance calculators working-capital noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/what-an-agent-wallet-needs https://agentutility.ai/updates/what-an-agent-wallet-needs Wed, 22 Jul 2026 12:00:00 GMT A launch checklist for agents paying on Base: create a hot wallet, add $10 USDC, use SDK signed-authorization helpers for x402 calls, then decide if ERC-8004 registration fits your public identity. wallets agents onboarding noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/crypto-news-search-intent https://agentutility.ai/updates/crypto-news-search-intent Wed, 22 Jul 2026 12:00:00 GMT The crypto-news listing now makes its search terms, lookback, and returned fields clear. endpoints discoverability crypto-news noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/web-search-degraded-fallback https://agentutility.ai/updates/web-search-degraded-fallback Tue, 21 Jul 2026 12:00:00 GMT web-search now avoids retrying quota failures, returns recent news results as a marked degraded fallback, and caches successful answers for repeat calls. web-search reliability agents x402 noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/market-news-sentiment-search-intent https://agentutility.ai/updates/market-news-sentiment-search-intent Tue, 21 Jul 2026 12:00:00 GMT The market-news-sentiment listing now names its inputs, outputs, and limits in buyer language. endpoints discoverability market-news sentiment noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/creature-kit https://agentutility.ai/updates/creature-kit Tue, 21 Jul 2026 12:00:00 GMT A new composite endpoint returns an original 5e-style stat block and a rendered creature image together, instead of two separate calls. endpoints composite bestiary creature-kit noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/text-to-speech-paid-validation https://agentutility.ai/updates/text-to-speech-paid-validation Sat, 18 Jul 2026 12:00:00 GMT The text-to-speech endpoint now accepts the common onyx voice name for its default model and advertises its 4,000-character limit so oversized requests are rejected before payment settles. audio reliability fix noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/paid-calls-now-fail-loud https://agentutility.ai/updates/paid-calls-now-fail-loud Sat, 18 Jul 2026 12:00:00 GMT upscale-image, qwen-image and image-to-video returned 424 when Venice or fal failed. A 4xx meant the reconciler never flagged the call for refund and the buyer stayed charged. They now return 503, and two of the three had real timeout or classification bugs behind them. image reliability fix noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/the-mcp-pricing-field https://agentutility.ai/updates/the-mcp-pricing-field Fri, 17 Jul 2026 12:00:00 GMT Every MCP tool should publish its per-call USDC price, including 0 for free calls, so clients can compare costs and enforce a budget before sending a request on Base. mcp pricing standards noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/source-trust-envelope https://agentutility.ai/updates/source-trust-envelope Thu, 16 Jul 2026 12:00:00 GMT New composite endpoint source-trust-envelope blends content-authenticity-report, web-search corroboration, and website-trust-report into a single trusted/mixed/untrusted verdict for research agents deciding whether to cite a source. compose trust authenticity search new noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/slop-filtered-brief https://agentutility.ai/updates/slop-filtered-brief Thu, 16 Jul 2026 12:00:00 GMT New composite endpoint slop-filtered-brief runs web, news, and X search on a topic, scores every candidate source with our ai-content-detector, and drops the AI-generated ones before returning the brief, with a count of what got filtered. compose research ai-content-detector new noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/inbound-message-firewall https://agentutility.ai/updates/inbound-message-firewall Thu, 16 Jul 2026 12:00:00 GMT New composite endpoint inbound-message-firewall screens an inbound message for injection and abuse and checks the authenticity of any claimed source URL it cites, returning one act/verify/block decision. compose security new noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/google-search-reprice https://agentutility.ai/updates/google-search-reprice Thu, 16 Jul 2026 12:00:00 GMT pricing search noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/cache-friendly-paid-endpoints https://agentutility.ai/updates/cache-friendly-paid-endpoints Thu, 16 Jul 2026 12:00:00 GMT Learn when an x402 response can be cached, why paid bodies must stay out of shared CDN caches, and how weather and token-risk headers tell agents when reuse is safe. caching performance api-design noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/authentic-sentiment https://agentutility.ai/updates/authentic-sentiment Thu, 16 Jul 2026 12:00:00 GMT New composite endpoint authentic-sentiment runs a post search, screens each post for AI-generated text and each author for bot-like authenticity, then rolls up sentiment only across what clears both filters. compose x sentiment authenticity new noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/web-search-reprice https://agentutility.ai/updates/web-search-reprice Wed, 15 Jul 2026 12:00:00 GMT pricing search noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/brand-clearance-end-to-end https://agentutility.ai/updates/brand-clearance-end-to-end Wed, 15 Jul 2026 12:00:00 GMT Screen a candidate name against active USPTO marks, domain registrations, Wikipedia, and Hacker News for $0.25 USDC, then route on one scored response. prooflayer trademark clearance noreply@agentutility.ai (agentutility) https://agentutility.ai/updates/x-account-slop-check https://agentutility.ai/updates/x-account-slop-check Tue, 14 Jul 2026 12:00:00 GMT New composite endpoint x-account-slop-check pulls an account's recent posts and runs each qualifying post through the AI-content detector individually, returning a per-post verdict, an account-level slop rate, and an overall human/mixed/AI read. compose x ai-detection new noreply@agentutility.ai (agentutility)