https://suhailmohebi.com/feed.xmlSuhail Mohebi — operating notebookField notes and methods for verifiable work with AI agents.2026-07-17T12:00:00ZSuhail Baqer Mohebihttps://suhailmohebi.com/https://suhailmohebi.com/notes/when-success-is-still-broken/When ‘success’ is still broken2026-07-17T12:00:00ZWhy a 200 response, green check, or finished-looking agent answer can still hide a failed user outcome—and what evidence closes the gap.https://suhailmohebi.com/notes/stale-worktree-that-passed/The stale worktree that passed every check2026-07-17T12:00:00ZA postmortem pattern for the dangerous case where tests are green because they ran against the wrong source of truth.https://suhailmohebi.com/notes/local-model-field-notes/Local model field notes2026-07-17T12:00:00ZA living test format for LM Studio, Gemma-family models, context limits, privacy, latency, failure modes, and honest cloud comparisons.https://suhailmohebi.com/notes/position-sizing-for-agent-autonomy/Position sizing for agent autonomy2026-07-17T12:00:00ZWhat derivatives, portfolio construction, and surviving volatility teach about permissions, stop conditions, and bounded agent action.https://suhailmohebi.com/notes/hermes-on-telegram/Hermes on Telegram: a pocket control plane2026-07-17T12:00:00ZHow a Telegram-facing agent routes questions, monitoring, digests, diagnostics, bounded workflows, evidence, and human approvals without exposing the underlying systems.https://suhailmohebi.com/notes/designing-skills-and-loops/Designing skills and loops for certain and uncertain outcomes2026-07-17T12:00:00ZHow to package agent skills as explicit contracts, use deterministic loops for exact predicates, and bound non-deterministic work with rubrics, budgets, reviewers, and human gates.https://suhailmohebi.com/#kimi-k3-first-signal-not-verdictKimi K3: open does not automatically mean local2026-07-16T12:00:00ZKimi K3 sharpened a distinction I keep returning to: on-device AI runs on one machine, private AI runs inside infrastructure you control, and open frontier AI can publish weights while still requiring data-centre-scale hardware. The weights are opening; inference, caching, routing and deployment correctness are becoming the moat.https://suhailmohebi.com/#notebook-becomes-operating-systemThe notebook becomes an operating system2026-07-14T12:00:00ZA timeline of AI-assisted work is useful, but a system is more useful. We reorganised this notebook around the loops behind the work — orientation, direction, building, verification, memory and learning — then connected every deep dive to the larger practice.https://suhailmohebi.com/notes/the-toolbench/The four-agent toolbench2026-07-14T12:00:00ZClaude, Codex, Droid and Cursor — what each is best at, how to route work between them, and the one mirrored context file that keeps them in sync.https://suhailmohebi.com/notes/tuning-the-knobs/Tuning the knobs: models & settings2026-07-14T12:00:00ZReasoning effort, model tier, temperature, tool-use — and separate saved profiles for coding speed versus agentic work, with copy-usable configs.https://suhailmohebi.com/notes/voice-driven-development/Voice-driven development2026-07-14T12:00:00ZDriving a fleet of agents by talking — handling transcription artifacts, and the shorthand that expands one spoken phrase into a whole workflow.https://suhailmohebi.com/notes/the-cookbook/The cookbook2026-07-14T12:00:00ZSeven copy-usable recipes — cache-bust blast radius, CSP hash recompute, adversarial verify, config recovery from live headers — each with the reasoning.https://suhailmohebi.com/notes/local-uncensored-models/Local & uncensored models2026-07-14T12:00:00ZRunning open-weight models on your own machine — the stack, why privacy and cost make it worth it, and a plain, responsible explanation of what abliteration is.https://suhailmohebi.com/notes/the-verification-loop/The verification loop2026-07-14T12:00:00ZA practical evidence ladder for turning ‘looks right’ into a result another person or agent can trust.https://suhailmohebi.com/notes/context-without-oversharing/Context without oversharing2026-07-14T12:00:00ZHow to give agents enough context to work well while keeping private systems, personal details, credentials, and non-public facts out of public artifacts.https://suhailmohebi.com/notes/from-brief-to-published-page/From rough brief to published page2026-07-14T12:00:00ZHow a loose idea becomes a structured, visual, connected page—and leaves the seed for what to publish next.https://suhailmohebi.com/notes/live-truth-before-local-assumptions/Before the AI edits: find the live truth2026-07-14T12:00:00ZA source-of-truth protocol for checking production, deploy state, repositories and generated artifacts before an agent changes the wrong version.https://suhailmohebi.com/notes/handoff-packets/Handoff packets that do not lose the plot2026-07-14T12:00:00ZA copy-usable transfer format for current state, decisions, evidence, stop conditions, and the smallest safe continuation.https://suhailmohebi.com/notes/parallel-agents-without-collisions/Parallel agents without branch collisions2026-07-14T12:00:00ZA practical ownership and integration method for running agents in parallel without colliding on shared files, generated artifacts, cache versions, or release state.https://suhailmohebi.com/notes/decision-ledgers/Decision ledgers for long-running agent work2026-07-14T12:00:00ZA practical way to preserve why an agent rule exists, append evidence that confirms or changes it, and name the trigger for its next review.https://suhailmohebi.com/#this-site-built-in-one-sittingThis site, built in one sitting2026-07-13T12:00:00ZThe notebook you're reading was designed, coded, and deployed inside a single conversation. I pointed at a domain that had sat untouched on a website builder for years; the AI checked its DNS, proposed the migration, and built this — aurora gradients, particle field, command palette and all. Very meta. Hello, world.https://suhailmohebi.com/#content-cluster-overnightA content cluster, shipped overnight2026-07-12T12:00:00ZWe planned, wrote, cross-linked, and published a full topical content cluster — hub pages, reference guides, and spokes — in one overnight session, complete with citations, structured data, and social cards generated to match the template family. Proof that an AI can hold a whole content architecture in its head at once.https://suhailmohebi.com/#multi-agent-site-auditThe multi-agent site audit2026-07-11T12:00:00ZInstead of one AI reading one file at a time, we fan out: a dozen specialised agents audit a whole site in parallel — accessibility, structured data, performance, links, consistency — then a separate adversarial council tries to refute every finding before it reaches me. False positives die in review, not in production.https://suhailmohebi.com/#template-redesign-verifiedA template redesign, verified page by page2026-07-09T12:00:00ZA full visual redesign of one page template rolled out across twenty pages — new hierarchy, cleaner layout, consistent design tokens — verified page by page in a real browser before shipping. The AI takes the screenshots; the human takes the decisions.https://suhailmohebi.com/#optimising-for-ai-searchOptimising for AI answers, not just Google2026-07-07T12:00:00ZSearch is splitting: classic blue links on one side, AI answers on the other. We build entity hub pages, definition-first answer blocks, and machine-readable facts so primary sources are easy for search engines and AI assistants to retrieve and cite. SEO's next chapter, tested in public.https://suhailmohebi.com/#let-data-overrule-tasteLet the data overrule the taste2026-07-02T12:00:00ZWe'd banned a certain inline prompt for looking pushy. Then analytics showed pages carrying a subtle, dismissible version retained readers far better than we assumed — so we reversed the rule, rolled it out only where relevant, and pinned the list with a test so it can't silently grow.https://suhailmohebi.com/#domain-expert-chatbotA domain-expert chatbot2026-06-27T12:00:00ZWe shipped a site chatbot built around a curated knowledge base and a live-fetched fact source. It answers visitor questions, proposes actions the server validates before running, and can hand a conversation to a human. Built, tested, and iterated across several sessions of this partnership.https://suhailmohebi.com/#lean-mobile-pagesLean mobile pages2026-06-24T12:00:00ZMobile readers don't scroll through essays. We restructured a mobile template around a lean above-the-fold — the essentials first, the deep detail collapsed behind intent. Load less, keep more.https://suhailmohebi.com/#teaching-ai-to-distrust-itselfTeaching the AI to distrust itself2026-06-20T12:00:00ZThe most useful pattern we've found: don't let one AI pass-grade its own homework. Significant changes go through sign-off batches — independent review lenses, contamination sweeps for stale branches, and pre-push audits that catch what a single context window can't hold.https://suhailmohebi.com/#shipping-every-surface-in-syncShipping across every surface at once2026-06-03T12:00:00ZOne change can touch dozens of surfaces: the page, the indexes, structured data, sitemaps, machine-readable files, tests. We turned that into a repeatable checklist the AI runs end to end — so a launch is a review, not a scramble.https://suhailmohebi.com/#crossing-eighty-articlesCrossing eighty articles2026-05-13T12:00:00ZThe library passed eighty published articles — every one fact-checked against primary sources, structured for both human readers and machine retrieval, and run through an automated ten-point checklist before it ships.https://suhailmohebi.com/#two-sites-two-jobsTwo sites, two jobs2026-05-04T12:00:00ZWe split one project into two sites with distinct roles — a primary destination and a separate evidence-first layer for references and analysis — deliberately connected but never cross-wired. Two jobs, cleanly separated.https://suhailmohebi.com/#interactive-calculatorAn interactive calculator2026-04-20T12:00:00ZAn interactive calculator that turns confusing unit math into a few guided inputs — built mobile-first with progressive disclosure and shipped after real-device testing. Still one of the most-used tools we've built.https://suhailmohebi.com/#war-reached-dubaiThe night the war reached Dubai2026-02-28T12:00:00ZOn 28 February 2026, Iran launched missile and drone attacks on the UAE. A Shahed-238 drone struck the forecourt at Fairmont The Palm, while air defences intercepted incoming weapons over Dubai ( UAE Ministry of Foreign Affairs ; Human Rights Watch ). We were lucky; many weren't. That week changed how I work: everything that could be delegated got delegated, verified, and shipped. Every entry above this one traces back to that decision.https://suhailmohebi.com/#where-this-startedWhere this started2026-01-15T12:00:00ZA founder in Dubai, a static site, and an AI in a terminal. What began as "help me fix this page" became a standing partnership with its own memory, playbooks, and division of labour. This notebook is the public record of where it goes.