Nasim Akhtar

470 posts

Nasim Akhtar banner
Nasim Akhtar

Nasim Akhtar

@fnLog0

one-click actions, zero-trust design, and ai agents that just get it. I’m building @IgrisBot fuel by @hyperbolaNet

Katılım Mart 2016
290 Takip Edilen363 Takipçiler
Nasim Akhtar
Nasim Akhtar@fnLog0·
Agreed guardrails cap the damage, but they don't fix why the loop started. The agent looped because it forgot what it already tried. Thats a memory problem, not a spend-limit problem. Curious you mention using an Obsidian-style knowledge graph. Do you think that pattern (structured nodes, typed links, contradiction tracking) could work as the runtime memory layer for agents? Feels like the same problem at a different scale.
English
0
0
1
17
Nyk 🌱
Nyk 🌱@nyk_builderz·
@fnLog0 @kentcdodds This is exactly the problem — benchmarks look good, but $47K runaway loops happen when context architecture fails. Cost visibility + guardrails matter more than pass rates. x.com/nyk_builderz/s…
Nyk 🌱@nyk_builderz

11 days, 190+ commits, and one PR later, I’m happy to announce the release of Mission Control v2 🌱 A major step forward for open-source AI agent ops: • Onboarding & Walkthrough • Local + gateway modes • Hermes, Claude, Codex + OpenClaw observability • Obsidian-style memory graph + knowledge system • Rebuilt onboarding + security scan autofix • Agent comms, chat, channels, cron, sessions, costs • OpenClaw doctor/fix, update flow, backups, deploy hardening • Multi-tenant + self-hosted template improvements Mission Control is becoming the mothership where agents dock: memory, security, visibility, coordination, and control in one place. OSS, self-hostable, and still moving fast.

English
1
0
1
60
Nasim Akhtar
Nasim Akhtar@fnLog0·
Hey @kentcdodds you have been talking about context bloat and context engineering in the MCP/agent space. i wrote a deep dive on why AI agents actually fail (benchmarks, $47k runaway loops, error compounding) and why context engineering not bigger models is the fix. i think the missing layer is structured memory that decides what the model sees at each step, not dumping everything into the window. Would love your take on whether i'm right about the problem and the solution: dev.to/fnlog0/your-ne…
English
3
0
0
681
Nasim Akhtar
Nasim Akhtar@fnLog0·
@kentcdodds thanks! I really appreciate that. If you get a chance, please read the article and let me know your thoughts
English
0
0
0
21
Kent C. Dodds ⚡
Kent C. Dodds ⚡@kentcdodds·
@fnLog0 I haven't read your article, but I agree with your post. A good harness is critical
English
1
0
1
598
Nasim Akhtar retweetledi
Thariq
Thariq@trq212·
Today we're launching local scheduled tasks in Claude Code desktop. Create a schedule for tasks that you want to run regularly. They'll run as long as your computer is awake.
English
671
1K
13.6K
3.7M
Nasim Akhtar
Nasim Akhtar@fnLog0·
Great points. Context rot is a fundamental systems problem in LLM agents. Long-running workflows cause: • context degradation • reasoning drift • instruction loss Which is why multi-step agents fail far more than demos suggest. I explored this in my article on the “AI Agent Illusion” 👇 dev.to/fnlog0/your-ne…
English
0
0
1
16
Muratcan Koylan
Muratcan Koylan@koylanai·
Context Rot is still a major issue. 1. Optimize for tokens-per-task, not tokens-per-request 2. Use structured summaries with explicit sections for file tracking 3. Trigger compression at 70-80% context utilization 4. Segment tasks to prevent context confusion across different objectives 5. Validate retrieved documents for accuracy before adding to context Some Skills: Context Degradation - github.com/muratcankoylan… Context Compression - github.com/muratcankoylan… Context Optimization - github.com/muratcankoylan…
Cline@cline

GPT-5.4 also has a 1M context window, but their evals show that needle-in-a-haystack (MRCR v2) scores 97% at 16-32K tokens, drops to 57% at 256-512K, and just 36% at 512K-1M. So it's a good idea to compact regularly!

English
8
3
43
4.4K
Austin Griffith
Austin Griffith@austingriffith·
🛠️ give your agents the skills they need to build onchain 👉 the ticker is ETHSkills.com
English
40
96
770
40K
Nasim Akhtar retweetledi
OpenAI
OpenAI@OpenAI·
Introducing EVMbench—a new benchmark that measures how well AI agents can detect, exploit, and patch high-severity smart contract vulnerabilities. openai.com/index/introduc…
English
1.2K
1.3K
8.8K
2.5M
Nasim Akhtar
Nasim Akhtar@fnLog0·
@stijnnoorman this article feels genuinely human. I haven’t read something like this in a long time.
English
0
0
1
343
Nasim Akhtar
Nasim Akhtar@fnLog0·
@kentcdodds Oh memory this is promising. What specific caching strategy do you have in mind?
English
1
0
0
32
homanp
homanp@pelaseyed·
TUI is fun
Lietuvių
3
0
7
622
Nasim Akhtar retweetledi
Chris Tate
Chris Tate@ctatedev·
Introducing json-render AI-generated UI. Deterministic output. 1. Define your component catalog 2. AI steams JSON 3. Render interactive UI Let users prompt dashboards, widgets and apps - safely constrained to components and actions you define
English
253
460
5.9K
671.7K
Nasim Akhtar retweetledi
Y Combinator
Y Combinator@ycombinator·
Skillsync helps companies find elite (but overlooked!) engineers on GitHub based on what they have actually built. It analyzes public GitHub contributions and turns them into structured skill profiles that recruiters & hiring managers can search. Congrats to @narsagna and @nishantjosh on the launch! ycombinator.com/launches/PAP-s…
English
119
122
1.9K
366.7K
Nasim Akhtar retweetledi
Charles Guillemet
Charles Guillemet@P3b7_·
🚨 There’s a large-scale supply chain attack in progress: the NPM account of a reputable developer has been compromised. The affected packages have already been downloaded over 1 billion times, meaning the entire JavaScript ecosystem may be at risk. The malicious payload works by silently swapping crypto addresses on the fly to steal funds. If you use a hardware wallet, pay attention to every transaction before signing and you're safe. If you don’t use a hardware wallet, refrain from making any on-chain transactions for now. It’s still unclear whether the attacker is also stealing seeds from software wallets directly at this stage. Excellent report here: jdstaerk.substack.com/p/we-just-foun…
English
2.9K
7.1K
18.7K
9.8M
Nasim Akhtar
Nasim Akhtar@fnLog0·
RT @gakonst: We had a bug in Reth's state root computation today which caused multiple nodes to stall. Recover by running the following: `…
English
0
1
0
87
Nasim Akhtar retweetledi
Haxo Labs
Haxo Labs@HaxoLabs·
Haxo Labs is rethinking how digital infrastructure is imagined, designed, and deployed across Fintech, SaaS, AI, and Blockchain. We build systems with clarity, not chaos. Our goal: tech that lasts longer than the hype.
English
3
19
30
12.7K

Keşfet

@kentcdodds @austingriffith @stijnnoorman @pelaseyed @elonmusk @BarackObama @taylorswift13 @cristiano