llms.txt
+33 −13
https://tools.openkrill.app/llms.txt
− > Agent Tools by OpenKrill provides free Model Context Protocol (MCP) servers and streamable HTTP data APIs for AI coding assistants, autonomous agents, and developers worldwide: YouTube transcripts, open jobs from company careers pages, app store reviews, and X posts. Results are cached, so repeat requests are answered from the cache.− - [US Federal Regulations (CFR)](https://tools.openkrill.app/tools/rules): Read US federal regulations and new proposals in the Federal Register. Free. MCP: https://rules.openkrill.app/mcp− - [Remote Jobs by Country](https://tools.openkrill.app/tools/remote-jobs): Find remote jobs you can be hired for from your country, each with the line of the post that says so. Free. MCP: https://remotejobs.openkrill.app/mcp− - [Tu Vi Chart](https://tools.openkrill.app/tools/tuvi): Cast a Tu Vi (Zi Wei Dou Shu) birth chart from a birth date and time, with readings that cite the classic. Free. MCP: https://tuvi.openkrill.app/mcp− - [Hiring Signals](https://tools.openkrill.app/tools/hiring-signals): Brief a company's hiring and watch companies for new and closed roles. Free. MCP: https://hiring.openkrill.app/mcp− robots.txt allows every crawler on every page and names these in their own allow group: Googlebot (Google Search and Brave Search), Bingbot (Bing, and the Bing index behind Yahoo and DuckDuckGo results), DuckDuckBot (DuckDuckGo), Slurp (Yahoo), Applebot (Apple Spotlight and Siri), YandexBot (Yandex), OAI-SearchBot (ChatGPT search), PerplexityBot (Perplexity), ClaudeBot (Claude), Google-Extended (Gemini), Amazonbot (Alexa and Amazon answers).− - CVE Risk Check: https://cve.openkrill.app/mcp. Three tools: `check_cve` (a CVE id: priority exploited, likely, routine or unknown, CVSS score and severity, CISA Known Exploited Vulnerabilities status with due date, EPSS probability, a patch or advisory link and the NVD and EPSS as-of dates), `check_cves` (up to 10 CVE ids, most urgent first, with a count per priority) and `list_recent_kev` (CVEs CISA added to its exploited list in the last 1 to 90 days, with an optional product word). Sources − ## Etsy Listing Signals− Which new Etsy listings in a niche are gaining saves fastest, and which tags they use. Our collector reads the Etsy Open API every 4 hours and keeps one reading a day of each tracked listing's views and favorites ("saves"); nothing is estimated and sales are not shown. Each listing comes with views and saves, both daily changes, the daily save rate (saves gained / views gained x 100), a 7-day saves trend and one of five buckets: `super_breakout`, `graduated_winner`, `steady_grower`, `calibrating− - Page: https://tools.openkrill.app/etsy-signals− - MCP: `https://etsysignals.openkrill.app/mcp`, tool `find_etsy_breakouts` with `niche` (one to three plain words), optional `max_age_days` (7, 14 or 30), `sort` (`delta_saves`, `daily_save_rate`, `delta_views`, `newest`), `buckets` and `limit` (1 to 50).− - `GET https://tools.openkrill.app/etsy-signals/read?niche=cat%20mom&max_age_days=14&sort=delta_saves` answers the same JSON. A niche not tracked yet is queued and searched from the next run.− - Is my domain ready for cold email? Our own score before and after one DNS record: https://tools.openkrill.app/blog/cold-email-domain-check-before-after-dmarc (Markdown: https://tools.openkrill.app/blog/cold-email-domain-check-before-after-dmarc.md). A real before and after on our own domain: 80 (almost) with no DMARC record, 96 (ready) the day we added one TXT record, and the rules that still matter.+ > Agent Tools by OpenKrill provides free Model Context Protocol (MCP) servers and streamable HTTP data APIs for AI coding assistants, autonomous agents, and developers worldwide: open jobs from company careers pages, app store reviews, X posts, and the per-tool servers listed below. Results are cached, so repeat requests are answered from the cache.+ - [Regulation Reader](https://tools.openkrill.app/tools/rules): Read US federal regulations and new proposals in the Federal Register. Free. MCP: https://rules.openkrill.app/mcp+ - [Post Kit](https://tools.openkrill.app/tools/postkit): Draft a post for each social platform, checked against its rules. Free. MCP: https://postkit.openkrill.app/mcp+ - [Remote Jobs by Country](https://tools.openkrill.app/tools/remote-jobs): Find remote jobs you can be hired for from your country, each with the line of the post that says so. Free. [Browse open remote jobs](https://tools.openkrill.app/remote-jobs). MCP: https://remotejobs.openkrill.app/mcp+ - [Tu Vi Chart](https://tools.openkrill.app/tools/tuvi): Cast a Tu Vi (Zi Wei Dou Shu) birth chart from a birth date and time, with readings that cite the classic. Free. Chart page: https://tools.openkrill.app/vi/la-so-tu-vi (in Vietnamese); Tu Vi by birth date: https://tools.openkrill.app/vi/tu-vi-theo-ngay-sinh (in Vietnamese). MCP: https://tuvi.openkrill.app/mcp+ - [Hiring Signals](https://tools.openkrill.app/tools/hiring-signals): Brief a company's hiring and watch companies for new and closed roles. Free. [Fastest-hiring companies this week](https://hiring.openkrill.app/leaderboard). MCP: https://hiring.openkrill.app/mcp+ robots.txt allows every crawler on every page and names these in their own allow group: Googlebot (Google Search and Brave Search), Bingbot (Bing, and the Bing index behind Yahoo and DuckDuckGo results), DuckDuckBot (DuckDuckGo), Slurp (Yahoo), Applebot (Apple Spotlight and Siri), YandexBot (Yandex), OAI-SearchBot (ChatGPT search), PerplexityBot (Perplexity), ClaudeBot (Claude model training), Claude-SearchBot (Claude search), ChatGPT-User (ChatGPT, for a person's request), Claude-User (Claude, + - [KrillData](https://data.openkrill.app): an MCP server and REST API that gives AI agents web and social data.+ - CVE Risk Check: https://cve.openkrill.app/mcp. Four tools: `check_cve` (a CVE id: priority exploited, likely, routine or unknown, CVSS score and severity, CISA Known Exploited Vulnerabilities status with due date, EPSS probability, a patch or advisory link and the NVD and EPSS as-of dates), `check_cves` (up to 10 CVE ids, most urgent first, with a count per priority), `list_recent_kev` (CVEs CISA added to its exploited list in the last 1 to 90 days, with an optional product word) and `triage_d+ - How to make an Apple Live Photo from HTML: one render, checked byte by byte: https://tools.openkrill.app/blog/make-apple-live-photo-from-html (Markdown: https://tools.openkrill.app/blog/make-apple-live-photo-from-html.md). One real render of an HTML card: 4 files back, the JPEG and the MOV carrying the same content identifier, their sizes, and the limits on length, size and frame rate.+ - Is my domain ready for cold email? Our own score before and after one DNS record: https://tools.openkrill.app/blog/cold-email-domain-check-before-after-dmarc (Markdown: https://tools.openkrill.app/blog/cold-email-domain-check-before-after-dmarc.md). A real before and after on our own domain: 80 (almost) with no DMARC record, 100 (ready) on the latest run after we added one TXT record, and the rules that still matter.+ - Search the Job Title, Not the Posting: Vercel "engineer" Went From Precision 0.40 to 1.00: https://tools.openkrill.app/blog/searching-engineer-on-a-job-board-returned-sales-roles (Markdown: https://tools.openkrill.app/blog/searching-engineer-on-a-job-board-returned-sales-roles.md). Matching the job title, not the whole posting, took Vercel "engineer" from precision 0.40 to 1.00 and our truth cases from 0 to 10 of 10.+ - We Asked Claude 10 Tool Questions With Web Search: 1 Answer Named Us, and It Linked a Directory: https://tools.openkrill.app/blog/we-asked-claude-10-tool-questions-and-1-answer-named-us (Markdown: https://tools.openkrill.app/blog/we-asked-claude-10-tool-questions-and-1-answer-named-us.md). In total: 10 answers, 96 cited links on 47 distinct sites, and 1 answer named us (10%).+ - We Read 33 Open US Class Action Settlements From Their Official Sites: 13 Blocked Our Script: https://tools.openkrill.app/blog/we-read-33-open-us-class-action-settlements-from-their (Markdown: https://tools.openkrill.app/blog/we-read-33-open-us-class-action-settlements-from-their.md). 13 of the 33 official settlement pages blocked our script, so those rows are re-read by hand.+ - We Ran 48 Real Broken Product CSVs Through Our Catalog Checker: It Caught 6, and Where It Fell Short: https://tools.openkrill.app/blog/we-ran-48-real-broken-product-csvs-through-our-catalog (Markdown: https://tools.openkrill.app/blog/we-ran-48-real-broken-product-csvs-through-our-catalog.md). We built one audit CSV with 7 known errors. All 7 were found in 1 call, each with the row, the rule, the fix and a link to the Shopify doc.+ - We Checked 6 Product Names: 5 Failed on the .com, and npm Passed the One With a Live Product: https://tools.openkrill.app/blog/we-checked-6-names-for-a-new-tool (Markdown: https://tools.openkrill.app/blog/we-checked-6-names-for-a-new-tool.md). We ran no live trademark office search. The free USPTO search API answered 404 to our scripts, so we could not read it automatically.+ - Our Falco Sensor Went Silent for Hours While Still Running: 611 Connects in 85 Seconds From a Fresh Process: https://tools.openkrill.app/blog/our-falco-sensor-went-silent-for-hours-while-still-running (Markdown: https://tools.openkrill.app/blog/our-falco-sensor-went-silent-for-hours-while-still-running.md). Time to notice a silent sensor went from 3 hours, connects only, to about 1 hour, each sensor on its own.+ - A Node Job Hit Its 384 MB Cap on 11,933 Records: Keeping 4 Fields per Record Cut the Peak to 246 MB: https://tools.openkrill.app/blog/a-node-batch-job-hit-its-384-mb-memory-cap-on-11-933-records (Markdown: https://tools.openkrill.app/blog/a-node-batch-job-hit-its-384-mb-memory-cap-on-11-933-records.md). The judge step now keeps 4 fields per record: the key, the subject, the row and a due flag.+ - 9 Odd DNS Lookups Traced to Our Own Crawler: 8 CDN CNAME Targets and 1 Broken robots.txt Redirect: https://tools.openkrill.app/blog/9-odd-dns-lookups-traced-to-our-own-crawler (Markdown: https://tools.openkrill.app/blog/9-odd-dns-lookups-traced-to-our-own-crawler.md). A check that judges the name the program asked for, not the CNAME target, would not have fired on 8 of the 9.+ - Reading 8 App Store and Google Play Data URLs Against robots.txt by RFC 9309: 3 Allowed, 5 Disallowed: https://tools.openkrill.app/blog/reading-8-app-store-and-google-play-data-urls-against-robots (Markdown: https://tools.openkrill.app/blog/reading-8-app-store-and-google-play-data-urls-against-robots.md). Three rules did the work. Our tool sends no named product token, so the * group applies (section 2.2.1).+ - Kit API custom fields: send the key, not the label, or the value is dropped with only a warning: https://tools.openkrill.app/blog/kit-api-custom-fields-key-not-label (Markdown: https://tools.openkrill.app/blog/kit-api-custom-fields-key-not-label.md). Send custom fields to the Kit API by their key, such as signup_page, not their label, such as "Signup page". In our run the label returned 201 Created but saved null, with only the warning "Unknown custom field key". A PUT or a second POST with th+ - Make data store test: act only on new RSS items, and repeat runs cost 6 credits, not 11: https://tools.openkrill.app/blog/make-data-store-dedupe-test (Markdown: https://tools.openkrill.app/blog/make-data-store-dedupe-test.md). A Make data store can remember which items a scheduled job already handled. In our test, the first run saved 5 post addresses for 11 credits, a second run with nothing new cost 6, and a run with 1 new item cost 7. The check costs 1 credit per item on every run.+ - Apify schedule test: a daily crawl from 2 API calls, and one fixed URL for its latest results: https://tools.openkrill.app/blog/apify-schedule-task-daily-crawl-test (Markdown: https://tools.openkrill.app/blog/apify-schedule-task-daily-crawl-test.md). An Apify task plus a schedule turns a crawl into a daily job with 2 API calls. Our 3-page test run took 16.5 seconds and cost 0.0011 US dollars, about 0.03 US dollars a month, and the task's runs/last address returned the newest successful results+ - Make webhook test: 5 curl calls answered in under 1 second, 2 credits each: https://tools.openkrill.app/blog/make-custom-webhook-curl-test (Markdown: https://tools.openkrill.app/blog/make-custom-webhook-curl-test.md). A Make scenario with a Custom webhook and a Webhook response module answered each of our 5 curl calls in 0.81 to 0.90 seconds and used 2 credits on each call, so the free plan's 1,000 credits cover about 500 calls a month. With the scenario off, the call still got HTTP 200, but t+ - Kit API test: tag a new subscriber in 2 calls, but the tag's list took up to 6 minutes to show them: https://tools.openkrill.app/blog/kit-api-add-subscribers-with-a-tag (Markdown: https://tools.openkrill.app/blog/kit-api-add-subscribers-with-a-tag.md). Create the subscriber with POST /v4/subscribers, then tag them with POST /v4/tags/{id}/subscribers. In our run both answered 201 in 1.4 to 4.2 seconds, and a repeat tag returned 200 with no duplicate. The tag's subscriber list stayed empty for o+ - Apify Website Content Crawler from the API: 5 pages to Markdown in 16 seconds, for 0.12 cents: https://tools.openkrill.app/blog/apify-website-content-crawler-api-test (Markdown: https://tools.openkrill.app/blog/apify-website-content-crawler-api-test.md). One API call with waitForFinish ran Apify's Website Content Crawler on 5 pages of our blog and returned when it was done, and one more GET read the results: 15.9 seconds of crawl, 17.9 seconds end to end, 0.0012 US dollars, and 3,130 to 6,586 + - Pictory trial test: our blog post became a 44-second video in about 5 minutes, and 1 line said the opposite of the post: https://tools.openkrill.app/blog/pictory-trial-blog-post-to-video-test (Markdown: https://tools.openkrill.app/blog/pictory-trial-blog-post-to-video-test.md). Pictory's Turn URL into video made a 44-second 1080p video from our blog post in about 5 minutes for 13.90 of 300 trial credits. Every number was right, but one scene claimed a test the post says we did not run, and eve+ - Writesonic AI visibility trial: the first report said 16 percent (4 of 25 answers), the non-branded dashboard 2.7 percent (2 of 74): https://tools.openkrill.app/blog/writesonic-ai-visibility-trial-branded-prompts (Markdown: https://tools.openkrill.app/blog/writesonic-ai-visibility-trial-branded-prompts.md). Writesonic's first AI visibility report gave our site 16 percent, but some of the test prompts it used named our domain. The dashboard's non-branded view, a separate set of 74 answers, show+ - Synthesia free plan test: a 30-second avatar video in about 4 minutes, with garbled text in the first 5 seconds: https://tools.openkrill.app/blog/synthesia-free-plan-avatar-video-test (Markdown: https://tools.openkrill.app/blog/synthesia-free-plan-avatar-video-test.md). On Synthesia's free Basic plan the chat builder turned our 52-word script into a 30-second avatar video in about 4 minutes and let us download a watermarked 1080p file. The first 5 seconds showed garbled text that the builder's+ - Murf free plan test: 17 of 20 tech terms read clearly, and the free plan cannot download the audio: https://tools.openkrill.app/blog/murf-free-plan-tech-terms-voice-test (Markdown: https://tools.openkrill.app/blog/murf-free-plan-tech-terms-voice-test.md). Murf's free plan read 17 of 20 developer terms clearly in our 149-word script. It likely said OAuth as oath and clipped dot env. It used 56 seconds of its 10 free minutes, counted by audio length, and the free plan does not let you download t+ - Kapwing free plan test: auto subtitles got 140 of 141 words right in 12.7 seconds, then exported at 720p with a watermark: https://tools.openkrill.app/blog/kapwing-free-plan-auto-subtitles-test (Markdown: https://tools.openkrill.app/blog/kapwing-free-plan-auto-subtitles-test.md). Kapwing's free plan subtitled our 49-second clip in 12.7 seconds for 1 of 10 credits, with 140 of 141 words exact. The free export is capped at 720p and carries the Kapwing logo on every frame.+ - Our Pinterest affiliate experiment: 9 SaaS guides, 30 pins, 0 US dollars spent, and the numbers that decide if it worked: https://tools.openkrill.app/blog/pinterest-affiliate-experiment-9-saas-guides (Markdown: https://tools.openkrill.app/blog/pinterest-affiliate-experiment-9-saas-guides.md). OpenKrill tested 9 SaaS products on their free plans and trials and wrote one number-first guide for each. On 2026-10-21, before the first pin went up, OpenKrill set 3 targets for 30 Pinterest pins, posti+ - How to draft one announcement for X, Bluesky and LinkedIn and count weighted characters, graphemes and links: https://tools.openkrill.app/blog/draft-social-posts-for-x-bluesky-and-linkedin (Markdown: https://tools.openkrill.app/blog/draft-social-posts-for-x-bluesky-and-linkedin.md). On 2026-10-20, Post Kit turned a 231-character city library announcement with a 107-character booking URL into 3 verified social drafts in 12,755 ms: an X post at 162 of 280 weighted characters (counting the link a