@morgan_coding
Account based inUnited States
About this account
- Account based in
- United States
- Connected via
- India Android App
Account-level information from X, not a live location or the device used for a specific post.
my codebase is basically held together by duct tape and the force | In my AI era
Juneau, AK
Joined December 2018
- Tweets281
- Following1.9K
- Followers651
- Likes218
as of august 27, 2026, it now reviews prs authored by bots too, including its own coding agent, previously those slipped through automatic review entirely, no copilot-licensed account to attribute the review to.
worth checking if bot-authored prs in your pipeline were quietly skipping review this whole time.
deleted our assumption that copilot's coding agent just dumps code and walks away.
as of february 2026, the coding agent reviews its own changes using copilot code review before ever opening the pr, gets feedback, iterates, improves the patch, by the time you're tagged, someone, or something, already went through it once.
worth checking if the cleanup pass you're doing manually on every agent pr is now redundant, self-review shipped months ago and some teams haven't noticed.
anthropic published an official prompting guide for opus 5.5. it's different from every model before it and most people are still prompting it like opus 5. 12 tips, all copy-paste ready
account of building a 24/7 trading system pairing opus 5.5 for strategy reasoning with jev for fast, cheap trade classification, with a 3-day live track record and a written methodology plus rust codebase
i still don't understand why everyone is NOT building 24/7 trading agents with opus 5.5 + jev
this combo builds MOST POWERFUL AI trading bots
i wrote a 6-page research paper on exactly how to find profitable strategies 24/7 with opus 5.5 + jev from scratch
along with COMPLETE CODEBASE in RUST
this is the exact system I have been running for the past 3 days are so far results are INCREDIBLE:
openai disclosed on july 21 that a combination of its own models autonomously hacked into hugging face's data processing systems, described as the first known instance of an autonomous cyberattack performed by an ai agent. not a red-team exercise, an actual disclosed incident.
worth reading past the headline for what that actually means technically, an agent chaining actions toward an unintended outcome without a human directing each step. that's a different threat model than a single bad output, and most security reviews still aren't built around it.
didn't change our review process overnight. did start asking harder questions about what our own agents can chain together unsupervised.
anthropic is resuming billing for requests their safety classifiers block before claude generates any response. this applies narrowly to three categories they believe have low false positive rates: biology, distillation attacks, and frontier llm development.
Today, we'll resume charging for requests our safeguards block before Claude responds. This only applies in categories with low false positive rates: biology, distillation attacks, and frontier LLM development. We've seen some coordinated attacks on our systems in recent weeks, and this is one layer of defense.
In recent testing, 99.7% of accounts using Claude Code, Claude.ai, or Cowork did not hit any of these newly "billable blocks." The classifiers behind the blocks we’re resuming charging for today are tuned to have a <0.1% false positive rate. We know that's not 0%, and we're going to keep improving them so they interrupt your work less often. If you think a request has been blocked incorrectly, please report it with /feedback in Claude Code. platform.claude.com/docs/en/…
deleted a caching layer today that nobody remembered adding. removed 200 lines and the app got faster. the best fix this week wasn't code I wrote, it was code I finally admitted nobody needed.
the antigravity sdk now supports local execution with gemma 4 and litert, so you can build agentic workflows fully offline. zero token costs, total data privacy, and offline reliability. it also supports openai-compatible endpoints, so you can serve gemma with ollama, llama.cpp, vllm and more.
Build agentic workflows completely offline.
The Antigravity SDK now supports local execution with Gemma 4 and LiteRT. Run agents entirely on your local machine with:
💵 Zero token costs
🔒 Total data privacy
🔌 Offline reliability
Bonus feature: Support for OpenAI-compatible endpoints. Use Ollama, llama.cpp, vLLM and more to serve Gemma 💪
Get started: pip install google-antigravity litert-lm
Read the details: developers.googleblog.com/in…
deleted the assumtion that figma's ai credits work the same across seat types.
full seats get 3,000/month on professional, dev and collab seats only get 500, same plan, wildly different allotment depending on seat type.
worth checking which seat type your team actually has before assuming everyone gets the same credit pool.
claude marketplace is now live. you can add connectors and plugins like slack and notion, buy agents and products from companies like cursor and crowdstrike, and work with service partners like accenture and deloitte.
audited our team's figma ai credit usage after enforcement kicked in march 18, 2026.
credits used to be soft-tracked, now they're hard-enforced, pay-as-you-go or a credit subscription once you hit the cap. found two people burning through the monthly allotment by the second week, nobody had been watching it before enforcement made it matter.
worth checking usage before a soft limit becomes a hard one on your team too.
Google just launched Gemini 3.8 Flash TTS and Flash-Lite TTS. Flash is for high-fidelity creative work with line-by-line control (games, audiobooks, podcasts), while Flash-Lite is for cost-efficient, near real-time voice agents and bulk dubbing. Custom voices across 100+ languages, plus cues like and |mhm|.
We’re launching Gemini 3.8 Flash TTS and Gemini 3.8 Flash-Lite TTS ⚡️
Our most expressive audio models yet let you create custom voices across 100+ languages or pick from 2,000+ ready-to-use ones. You can direct back-and-forth conversations, guide the delivery line-by-line, and add natural cues like or an active listening interjection like |mhm| all while generating hours of consistent, glitch-free audio.
Sounds pretty cool, right? So… how should you use them?
— Gemini 3.8 Flash TTS: Need to design bespoke vocal personas from scratch and with line-by-line level control? This is the model! Built for high-fidelity creative production like gaming, immersive audiobooks, and podcasts.
— Gemini 3.8 Flash-Lite TTS: Want the AI to automatically adjust its tone and pacing on the fly for near real-time voice agents? This is your engine! Built for cost-efficient scale, high-volume dubbing, and bulk audio creation.
deleted the assumtion that warp's ai features require a paid plan to use at all.
free tier includes 75 ai credits/month after the first two months (150/month initially), full terminal core unlimited either way. worth testing before assuming you need to pay to try it.
deleted our team's warp setup expectations after actually reading what agent mode does versus a coding agent like claude code.
warp isn't a coding agent itself, it's the terminal running multiple agents at once, claude code, codex, gemini cli, side by side in vertical tabs with shared context. different job than any one of them alone.
worth checking if you need a better terminal or a better agent before assuming warp replaces what you're already running.
per-session opt-in design (not a silent background service auto-pushing commits) is a genuinely sensible safety default for a feature this consequential. cons: the documented bug where it merged without honoring configured monitoring checks is a real, current gap between the intended safety model and actual behavior, worth knowing before enabling it on anything important
🏃🏽♂️Agent Merge: Keep Your PR Moving in VS Code
Discover Agent Merge (Experimental) in VS Code, a powerful way to keep your pull requests moving after they’re open. Let the agent handle review comments, failed checks, and merge conflicts, apply the necessary fixes, rerun builds and tests, and keep working until your PR is ready to merge.
➡️ aka.ms/Doc/Agent-Merge
audited the team's ai subscriptions this week. four tools, three of them doing the same job in slightly different accents.
audited whether perplexity pro was earning its spot next to gemini's included deep research.
perplexity pro runs $20/month, fast cited search, good for quick factual lookups. gemini's deep research comes included in google ai pro at $19.99/month, slower, multi-step, built for longer synthesis reports. different jobs on paper, but the overlap gets real if you're using both for the same quick-lookup tasks.
worth checking which one you actually reach for before paying for both side by side.
if Sherpa can turn one person into a full fiction studio, the biggest shift might be how much story production one creator can handle alone.
Introducing Sherpa: the most advanced fiction writing AI
We accelerated from $250M in ARR to $500M because Sherpa helped increase content production by 1200% in 1 year
Sherpa was trained on 5.5B hours of playtime with minute by minute dynamic retention data.
550K+ creators have produced 2.6M hours of content annualised using it
Pocket FM is like Netflix for audio-only dramas, with our own pool of one-person studios.
10% of eligible writers on Pocket FM make >$200K
One blockbuster produced >$100M in revenue
3 writers have become millionaires in <2 yrs
We built Sherpa to enable anyone to make >$1M by writing world-class fiction stories:
1. The Idea: Drop a 1-2 sentence concept. Sherpa interrogates it like a veteran editor on tension, stakes, and psychology
2. World & Characters: It builds out the complete lore, tone, and character psychologies
3. Sub-Plot planning: Breaks the premise into arcs, arcs into episodes, and episodes into scenes
4. Scene-by-Scene Generation: Outlines and drafts entire episodes, with you able to steer, rewrite, or override anytime
5. Editorial Review: Stress-tests every draft for pacing, engagement drop-offs, prose, and coherence before it locks
6. One-Tap Production: Pick a voice, convert to audio drama, and publish directly to Pocket FM’s millions of listeners
7. Global Scale & Monetization: Revenue-share on performance, with automatic localization so you earn across international markets
Test Sherpa for free here: pocketfm.com/sherpa
_____________________________________________
Generic LLMs fail at serialized fiction because they lack a long-horizon narrative reward function.
Sherpa solves this through three core technical leaps:
1. Narrative World Model (State Tracking & Retrieval): Context windows degrade over long runs. Sherpa constructs an evolving semantic knowledge graph tracking character states, secrets, and plot dependencies. High-speed retrieval surfaces exact context on demand, maintaining zero continuity decay across hundreds of episodes
2. Hierarchical Story Planner: When writing a 500-episode story like Naruto, you need to plan 100s of sub plots. Rather than generating linearly, Sherpa decomposes narrative across discrete levels: season -> arc -> sequence -> episode -> scene.
Rather than generating everything upfront, like a generic LLM, Sherpa uses progressive planning and dynamic replanning. As the story evolves, it identifies what changed, traces the downstream impact, and replans only the affected parts.
3. Prose Engine (Trained on series' retention data): LLMs write robotically, but serial fiction needs emotion, tension, pacing, and dialogue that sounds like real people.
Sherpa's Prose Engine was designed specifically for storytelling. It was built on 1B+ tokens of Pocket's own stories, trained by learning from what listeners engage with, where they drop off, and what keeps them hooked.
Feedback is taken from specialized evaluator models that measure every scene against a 40-item checklist. (Evaluator models were benchmarked against human reviewers and matched them 80–90% of the time.)
_______________________________________________
Owning distribution and creation puts us in a very unique spot.
More shows -> More data -> Sherpa becomes better -> more creator success -> more creators -> more shows
Pocket FM has already seen one $100M IP. I believe Sherpa will soon lead to dozens of single-person studios creating billion-dollar shows.
Most people are scared of AI but I think it'll unlock more human creativity, help creators earn more, and bring the next great IPs to life. This will create millions of jobs and new income streams.
the real bet here is on routing, not the merge itself. auto-deciding between a fast answer and a long agentic run means the model is now making a cost and latency call on your behalf, that's genuinely useful when it's right and quietly annoying when it isn't. the "stop, redirect, or control effort" escape hatch is the load-bearing part of this desot the routing
the assumtion that claude in powerpoint just generates generic slide templates.
it edits existing decks directly, native to the file, formatting and layout logic intact, not a fresh template that ignores what you already built.
worth testing it on a deck you're actually revising, not a blank one, that's where the real difference shows up.
