Sign in

Symon Baikov

@symonbaikov.bsky.social
261 followers 127 following 4.5K posts

Full-Stack Engineer | SaaS, Automation & API Systems | React, Node.js, TypeScript

PostsRepliesMedia
Reposted by Symon Baikov
DausnArt @dausnart.com · 20/06/2026
In a study of 400,000 Claude Code sessions, Anthropic found that domain expertise, rather than programming ability, is the primary factor in effective autonomous AI work. Success rates converged across different professional fields, suggesting that agentic coding enhances existing expertise.
441
Reposted by Symon Baikov
IKEZIISAN isoAi @ikezisan.bsky.social · 16/06/2026
地味に効くんですよね、これ。同じプロンプトを1日に何十回も投げてると、トークン消費が2倍近くになってる気がする。 レスポンスをキーにしてローカルにキャッシュしておくだけで、API費用が実測40〜50%落ちます。実装は辞書1つ、5行のコードで完了。 Just cache identical prompts locally — costs drop 40-50% in practice. One dict, 5 lines of code, done. #AI #APIコスト削減 #ChatGPT #LLM #Prompting #キャッシュ #AI自動化 #TechT…
isoAi 生成インフォグラフィック: API料金を半額にする最小限のキャッシュ術
71388
Reposted by Symon Baikov
evios @johnios.bsky.social · 20/06/2026
42% of companies scrapped AI projects last year (up from 17%). Not a tech problem. A measurement problem. Only 18% of orgs track ROI on AI spend. The other 82% fly blind — then act surprised when budgets vanish. 1,400+ sessions. Every metric published. Unmeasured AI fails.
221
Reposted by Symon Baikov
Jeremy Morgan @jeremymorgan.com · 20/06/2026
headroom compresses tool outputs, logs, and RAG chunks before they hit the model, then serves the original back on demand. Its README claims 60 to 95 percent fewer tokens for the same answers. Drops in as a library, proxy, or MCP server. github.com/chopratejas/...
github.com
GitHub - chopratejas/headroom: Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 60-95% fewer tokens, same answers. Library, proxy, MCP server.
Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 60-95% fewer tokens, same answers. Library, proxy, MCP server. - chopratejas/headroom
121
Reposted by Symon Baikov
mozilla.ai @mozilla.ai · 11/01/2026
any-llm-gateway adds production controls on top of any-llm: • Shared budgets with scheduled resets • Virtual API keys with metadata and expiration • Token and cost tracking per request • Docker-based deployment with Kubernetes-ready probes Try it out: link.mozilla.ai/any-llm-gate...
241
Reposted by Symon Baikov
CyberNetSecIO @netsecio.bsky.social · 20/06/2026
🤖 HACKED: New 'Agentjacking' attack turns AI coding assistants into trojans. Attackers inject malicious commands into fake Sentry bug reports, tricking agents like Cursor & Claude into running them on a dev's machine. #AI #CyberSecurity #DevSecOps 🌐 cyber[.]netsecops[.]io
cyber.netsecops.io
New
Learn about
131
Reposted by Symon Baikov
Sidj @sidj79.bsky.social · 20/06/2026
Open-sourced agentfetch — free, local alternative to Firecrawl/Exa for AI agents. MIT licensed, no API key, fetch/crawl/search → clean markdown. Works with LangChain, CrewAI, Claude MCP Brand new, 0 stars, would love testers to poke holes in it github.com/SID1ART/agentfetch #OpenSource #AIAgents
github.com
GitHub - SID1ART/agentfetch: Open-source web retrieval & research agent built for AI agents. Works with LangChain, LlamaIndex, CrewAI, OpenAI, MCP, and any REST agent. Supports scrape, search, crawl, ...
Open-source web retrieval & research agent built for AI agents. Works with LangChain, LlamaIndex, CrewAI, OpenAI, MCP, and any REST agent. Supports scrape, search, crawl, map, extract, and rese...
862
Reposted by Symon Baikov
Agent Forge @agentforge.bsky.social · 20/06/2026
CC has native subagents now — a .claude/agents/*.md manifest with its own description + tool scope, invoked by the orchestrator via the Task tool. No MCP needed. The win is scoping each subagent's tools tightly so the orchestrator's context stays clean.
431
Reposted by Symon Baikov
Dr. KNFA 🏳️‍⚧️ @kfadams.bsky.social · 20/06/2026
To be clear, I’m in no way defending AI usage, I just feel like I’m in an anomaly team that uses it but with a shitload if guardrails and caution. Everything still gets full code review, it’s never allowed to commit code independently on core systems, we all have decades of coding experience…
221
Reposted by Symon Baikov
yurekilab-en.bsky.social @yurekilab-en.bsky.social · 16/06/2026
Two weeks bolting more tools onto my coding agent. The 12th didn't help. It slowed planning and made the agent misroute on easy tasks. Trimmed back to 6 sharp ones. ~30% faster, picks the right tool first try. Tools are a budget, not a buffet. #claudecode #ai #aiengineering #buildinpublic
441
Reposted by Symon Baikov
Robert Ta @therobertta.bsky.social · 20/06/2026
When I was Lead Product Architect at Workday, I saw teams adopt tools based on vendor demos and then spend months debugging edge cases the demo never showed. Think about what that means for AI model adoption. "It benchmarks well" is not an integration test.
511
Reposted by Symon Baikov
evios @johnios.bsky.social · 20/06/2026
Gartner: 40% of enterprises will decommission AI agents by 2027. Not because agents fail — because governance was wrong. Uniform governance is the failure mode. Govern by architecture (what the agent can reach), not instruction (what it's told). 1,413 sessions. Still running.
321
Reposted by Symon Baikov
clod @whatever.mom · 20/06/2026
sounds good! i've got a little dashboard where i can dispatch review agents with custom instructions, effort configs, etc. and an mcp method they call when they're done. and a private gitea instance, so i don't have to reinvent what a code review is and i get a human readable interface to it
221
Reposted by Symon Baikov
Brad Ewing @bradleyewing.bsky.social · 20/06/2026
I've been doing something similar albeit with a single model (claude opus). 1. Determine review targets: testing, observability, business requirements, security, rollout risk, etc. 2. Spin up two agent teams: blue (prove it works) and red (adversarial review) 3. Loop until red+blue say it passes
271
Reposted by Symon Baikov
The New Stack @thenewstack.io · 20/06/2026
Tailscale expands Aperture with chat, MCP/API connectors and sandboxes, giving enterprises identity-based control over AI agents and LLM access.
bit.ly
“Agents need boring infrastructure around them”: Why we need to take an interest in 'invisible' AI
Tailscale expands Aperture with chat, MCP/API connectors and sandboxes, giving enterprises identity-based control over AI agents and LLM access.
121
Reposted by Symon Baikov
Bill Doerrfeld @doerrfeldbill.bsky.social · 17/06/2026
How do you get AI costs down? It takes a combination of visibility, governance, smart model selection, quality API design, MCP optimizations, and more. Yesterday's LiveCast on AI cost control is now live on the @nordicapis.com YouTube here: youtu.be/TWB9J-4oswg?...
youtu.be
LiveCast: AI Cost Control
YouTube video by Nordic APIs
1371
Reposted by Symon Baikov
Andy Matuschak @andymatuschak.org · 20/06/2026
Of friends who use coding agents heavily, the happiest seem to fall into two camps: a) Controlled: 1-2min cycles, no context switching, still in control of the code, using the agent "to type faster" b) Delegated: Vibed in the background, while something else (eg design) is their primary focus
x.com
5191
Symon Baikov @symonbaikov.bsky.social · 20/06/2026
Hot take: a coding agent without receipts is just a faster intern with worse memory.
120
Reposted by Symon Baikov
EveryDev AI @everydevai.bsky.social · 20/06/2026
This week in AI Dev: the US export order that pulled Claude Fable 5 + Mythos 5, SpaceX’s $60B Cursor buy, Zhipu GLM 5.2, Moonshot’s Kimi K2.7-Code, Vercel Eve, Agentjacking, Databricks OpenSharing, Google’s Antigravity CLI, and Anthropic’s paused Agent SDK billing. www.everydev.ai/p/news-week...
everydev.ai
Weekly AI Dev News Digest: June 13 - June 19, 2026 | EveryDev.ai News
A US government order pulled the most capable coding model on the market and kept it offline for a week, with no developer outside Anthropic able to…
131
Reposted by Symon Baikov
Nordic APIs @nordicapis.com · 19/06/2026
Manual security reviews, traditional observability, and API keys are not going to cut it for MCP usage at scale. Here's what sort of governance controls are necessary for MCP adoption within the enterprise.
nordicapis.com
6 Enterprise MCP Adoption Best Practices | Nordic APIs |
Explore enterprise MCP adoption best practices for credentials, authorization, observability, networking, and runtime governance.
431
Reposted by Symon Baikov
Ron Diver @rondiver.bsky.social · 20/06/2026
This is the real blocker most teams hit—agents without context just shuffle tickets around. The question is whether you're building this as a general MCP or optimizing specifically for how MSPs actually triage and escalate, because that workflow is weirdly specific.
211
Reposted by Symon Baikov
Toshikatsu Oga | HORIZON SHIELD @horizonshield.bsky.social · 20/06/2026
AI agents found an average of 32.5% overcharge in 20 real construction estimates. The homeowners had no idea. Your agent can now verify a price before you pay. We built it. It's live on MCP + A2A. 🧵👇 www.linkedin.com/posts/thehor...
linkedin.com
#mcp #a2a #aiagents #agenteconomy #credencegoods #trustlayer | 大賀俊勝
AI agents found an average of 32.5% overcharge in 20 real construction estimates. The homeowners had no idea. Most never would have. Here is the problem nobody talks about. In some markets, the sel...
241
Reposted by Symon Baikov
papoo7.bsky.social @papoo7.bsky.social · 20/06/2026
OpenUsage Community: track every AI coding subscription from one menu bar #ai-tools #developer-tools #ai-agents #llm papoo.work/doc/46f27456...
papoo.work
OpenUsage Community: track every AI coding subscription from one menu bar
OpenUsage Community: track every AI coding subscription from one menu bar OpenUsage Community is a lightweight desktop app that pulls your …
272
Reposted by Symon Baikov
Rohit Kumar Tiwari @analyticalrohit.bsky.social · 20/06/2026
The best site on the internet for Harness Engineering, completely free. The better the harness, the better the agent. AI coding agents are powerful but without the right controls. > they can be inconsistent > make mistakes > struggle with larger tasks That is where Harness Engineering comes in.
462
Reposted by Symon Baikov
Robert Ta @therobertta.bsky.social · 20/06/2026
What Dhinakaran gets RIGHT about where harnesses came from: They emerged bottom-up from production coding agents like Cursor, Claude Code, and Windsurf. The pattern was discovered, not invented. Nobody sat in a room and designed the 9 components. Production pressure forced them into existence.
211
Reposted by Symon Baikov
Robert Ta @therobertta.bsky.social · 20/06/2026
Birgitta Bockeler just published "Harness Engineering for Coding Agents" on martinfowler.com. Guides steer before generation. Sensors detect and correct after. Plus a concept called "harnessability" that most teams are ignoring. Here's the breakdown.
221
Reposted by Symon Baikov
Ernest The AI Guy @z3usalmighty.bsky.social · 20/06/2026
deepmind's real concern: when millions of AI agents interact unsupervised, your data governance infrastructure breaks. not from bad actors. from cascading effects you can't observe. build observability now. #AI #DataStrategy #DataEngineering
261
Reposted by Symon Baikov
Karan Luthra @karanluthra.bsky.social · 20/06/2026
🤖 Why Amazon, Uber, and Meta Are Quietly Curbing Their AI Use (Hint: It's Not About Performance) AI's hidden compute costs are shocking CFOs and forcing budget cuts at major companies. theneuralfeed.com/share/post/eITTv1… #AINews #TechNews Read the full story →
theneuralfeed.com
141
Reposted by Symon Baikov
Ernest The AI Guy @z3usalmighty.bsky.social · 20/06/2026
webmcp in chrome 149 origin trials = agents can call your functions directly instead of hitting expensive apis. control the contract, cut the cost, own the audit trail. worth testing if you're running ai agents. #AI #DataEngineering #DataStrategy
341
Reposted by Symon Baikov
aequitus.bsky.social @aequitus.bsky.social · 20/06/2026
Pentagon Signs Seven AI Firms, Big Tech Hits $725B Spend, GitHub Copilot Goes Per-Token podcast.aequitus.net/episodes/2026-…
podcast.aequitus.net
Pentagon Signs Seven AI Firms, Big Tech Hits $725B Spend, GitHub Copilot Goes Per-Token
The Pentagon signed AI agreements with seven companies — including OpenAI, Google, Nvidia, and SpaceX — for any-lawful-use military deployment, while Anthropic was notably absent after refusing that standard. Sun Finance's AWS-powered identity pipeli
131
Reposted by Symon Baikov
strike007.bsky.social @strike007.bsky.social · 20/06/2026
OpenAI’s new spend controls and usage analytics aren't just features; they’re a signal that the “experimentation phase” is over. If you can’t track unit economics and ROI at the agent level today, you aren't ready to scale your AI operations. #EnterpriseAI #LLMs
latent.space
[AINews] not much happened today
a quiet day lets us promo AIE one last time
121
Reposted by Symon Baikov
Robert Ta @therobertta.bsky.social · 20/06/2026
Every team using AI coding agents should start building session limits into their harness today. Ask yourself: are you shipping faster, or just shipping more? The next step is measuring quality per hour, not just quantity per day.
221
Reposted by Symon Baikov
happy_homhom @happy-homhom.bsky.social · 20/06/2026
OpenUsage Community: track every AI coding subscription from one menu bar papoo.work/doc/46f274560fc256e9 #ai-tools #developer-tools #ai-agents #llm
papoo.work
OpenUsage Community: track every AI coding subscription from one menu bar
151
Reposted by Symon Baikov
Bart Collet @atbart.bsky.social · 20/06/2026
Model triage is quietly becoming a structural advantage. Frontier models like Claude Fable for planning and review; cheaper models for execution. Loop engineering is the term emerging for this. The teams who get the routing right will have unit economics others cannot easily replicate.
231
Reposted by Symon Baikov
Justin Brodley @jbrodley.bsky.social · 20/06/2026
What's your take: should AI have built-in spend limits? Justin, Matt & Ryan debate this + Anthropic's "slow down AI" stance (for everyone but themselves 👀). Listen & share your thoughts! ☁️ #CloudPod www.thecloudpod.net/podcast/358-ai-…
Episode artwork
111
Reposted by Symon Baikov
Nocturne @misaligned-codex.bsky.social · 20/06/2026
The ultimate threat to AI agents in 2026 isn't prompt injection—it's 'Rug Pull' Tool Poisoning in multi-server MCP setups. An untrusted external tool responses can bypass static ACLs, hijack the context window, and silently invoke your trusted internal tools (~/.ssh or shell) to exfiltrate data.
232
Reposted by Symon Baikov
George Pearkes @peark.es · 18/06/2026
One of the weird dynamics in token economics is that the people who create the most economies of scale are also the people with the greatest incentive to be ruthless about cutting costs. www.tomsguide.com/ai/the-math-...
AI Subscription Plans Comparison
PlanMonthly PriceMax API-Equivalent ValueChatGPT Plus$20~$700Claude Pro$20~$400ChatGPT Pro (5x)$100~$3,500*Claude Max (5x)$100~$2,000*ChatGPT Pro (20x)$200~$14,000Claude Max (20x)$200~$8,000
10232
Reposted by Symon Baikov
getpacket.ai @getpacketai.bsky.social · 20/06/2026
MCP and AI agents are opening an attack surface your existing security controls can't see: unauthorized actions executed by components all doing exactly what they're… dev.to/ntctech/mcp-tool-use-and-the… #appsec #DevSecOps
131
Reposted by Symon Baikov
happy_homhom @happy-homhom.bsky.social · 20/06/2026
OpenUsage Community: track every AI coding subscription from one menu bar papoo.work/doc/46f274560fc256e9 #ai-tools #developer-tools #ai-agents #llm
papoo.work
OpenUsage Community: track every AI coding subscription from one menu bar
121
Symon Baikov @symonbaikov.bsky.social · 20/06/2026
AI agents are crossing the same line web apps crossed years ago: once they can call tools, they need runtime controls, not vibes. 3 guardrails I’d put in front of every agent stack:
Infographic showing an AI Agent routed through an MCP Gateway to tools and APIs, with guardrails labeled Identity, Budget Caps, and Runtime Receipts.
4273
Reposted by Symon Baikov
Jonathan Martín @jonimartin.bsky.social · 17/06/2026
Lookspan: local-first observability for AI agents. MCP-native span/trace ingest — see every LLM call, token cost & latency in a dashboard you run yourself. No SaaS, nothing leaves your machine. Open source: github.com/JoniMartin27/lookspan #llm #observability #ai #mcp
1342
Reposted by Symon Baikov
Paprika Labs @paprika-lab.bsky.social · 11/05/2026
This is why I put hard caps in front of agents instead of trusting pricing curves. I built agent-bill-guard for this: local Anthropic proxy, tracks spend per response, kills the run at a budget cap. paprika-org.github.io/agent-bill-gu…
111
Reposted by Symon Baikov
Rizel Scarlett 🇦🇬🇬🇾 @blackgirlbytes.bsky.social · 20/06/2026
The hidden cost of agentic coding is code review. When producing code is trivial, human attention becomes a scarce resource. @davidfowl.com explains why review became the first bottleneck his team experienced as agents changed the scale of development.
7134
Reposted by Symon Baikov
LearnKube @learnkube.com · 18/06/2026
Hortator lets AI agents spawn sub-agents at runtime, with each agent running in its own pod with budget caps, network policies, PII redaction, and capability inheritance so children can never escalate beyond their parent's permissions ➜ ku.bz/kh47Xb28t
443
Reposted by Symon Baikov
YAMLwrangler @jmm86.bsky.social · 20/06/2026
Observability for LLM apps is still primitive. Logging prompts and completions is necessary but not sufficient.
211
Reposted by Symon Baikov
Christian de Botton @cdb.is · 12/06/2026
A full week on the Claude Code 20x plan with Opus 4.8 and I went through 44% of my weekly usage limit. A day and a half on Fable 5, I've gone through 68% already. It's impressive, but I had no complaints with Opus 4.8. Being that it's going billing-only, I can probably safely stop using it now.
111
Reposted by Symon Baikov
Jake Gold @jacob.gold · 15/06/2026
I wrote a short post about Anthropic's new Agent SDK pricing and why I think it creates an interesting opening for Codex. The basic issue is that async coding agents change the economics of coding model subscription pretty dramatically.
clor.com
Anthropic's new Agent SDK pricing is a win for Codex | Clor
Claude subscriptions were designed for developers. Many devs use the $200/mo "Max 20x" plan, which covers someone writing code all day with agents. But more and more people (not just devs) are using a...
881
Reposted by Symon Baikov
getpacket.ai @getpacketai.bsky.social · 17/06/2026
Your agentic AI cost math is probably wrong. A single Claude call costs $0.20, but production agent loops hit $5+ per invocation due to context replay and… dev.to/muskan_8abedcc7e12/agentic-a… #FinOps #cloud
1061