{"slug": "show-hn-ai-agents-for-osint-sigint", "title": "Show HN: AI Agents for Osint/Sigint", "summary": "Ritwik launched Makra, a memoization-driven web scraping engine for AI agents, claiming it cuts data retrieval costs to one-tenth of prior generations while eliminating hallucinations and prompt injection. The tool, presented as a Show HN post on Aug. 27, 2026, uses a read-only browser harness to extract structured data at scale, addressing the limitations of long-context LLMs in dynamic web environments.", "body_md": "# The State of AI Web Scraping in 2026\n\nLAUNCH · AUG.27.2026 · Ritwik\n\n75% of the web is dynamic. Basically database records presented in HTML. Built for humans.\n\nFor Agents, this problem was solved by:\n\n**RAG** was the first generation (2024)- And then,\n**long-context LLMs** took it to a completely new level (2025)\n\nFast-forward to 2026, and we're basically dumping anything and everything into the context windows; taking Nvidia to the moon.\n\nThis consists of claws, browser/computer-use, and agentic crawlers, and scrapers. These have made it easier to retrieve data, but the cost has stayed the same. Underneath, these still rely on long-context LLMs.\n\nAnd, here's where they fall short:\n\n- Cost at scale: for large datasets, cost grows linearly on a per-page basis.\n- Hallucinations: LLMs are prone to hallucinations\n- Context pollution and prompt injection\n\nLet me now show you what we have built:\n\n**Memoization-driven Knowledge and Data Retrieval**\n\nIt's an work-in-progress implementation for a continual learning harness focusing on **Browser-use.**\n\n### Large-request correctness versus cost\n\nScatter plot of correctness against average USD cost for Makra, Exa, and Firecrawl. Lower cost is toward the left.\n\nBuilding applications on top of a memoization-engine like [Makra](/) solves all 3 of these problems:\n\n- Lower cost at scale: as layouts are memoized, cost comes down to a vector query.\n- Makra extracts data like a traditional scraper. So, hallucinations? Theoretically, none.\n- Your agentic loop only has to deal with the DOM nodes it's concerned with: More signal, less noise.\n\nWe have built an in-house **browser harness** (read-only for now) that does the following for you:\n\n- Spins up a browser instance and sets up proxies.\n- Extracts any structured or tabular data from the web for you.\n- And you can do all this at 1/10th the cost of all the other previous generations.\n\nTry it out in: [/playground](/?playground=true)", "url": "https://wpnews.pro/news/show-hn-ai-agents-for-osint-sigint", "canonical_source": "https://www.makralabs.org/about", "published_at": "2026-08-29 21:39:05+00:00", "updated_at": "2026-08-29 22:18:32.366734+00:00", "lang": "en", "topics": ["ai-tools", "ai-infrastructure", "ai-agents"], "entities": ["Makra", "Ritwik", "Nvidia", "Exa", "Firecrawl"], "alternates": {"html": "https://wpnews.pro/news/show-hn-ai-agents-for-osint-sigint", "markdown": "https://wpnews.pro/news/show-hn-ai-agents-for-osint-sigint.md", "text": "https://wpnews.pro/news/show-hn-ai-agents-for-osint-sigint.txt", "jsonld": "https://wpnews.pro/news/show-hn-ai-agents-for-osint-sigint.jsonld"}}