Are agent skills safe to install? A 1.7M-install trojan
Trojanized skills hit 1.7M displayed installs before takedown. See how the attack worked, a 6-check audit script, and what it found on a real dev machine.
Wed Oct 28 2026 · 9 min read · 1 views
AI
Gemini 3.7 Flash now runs as 3.8 Flash: what to check
Google quietly routes gemini-3.7-flash calls to 3.8 Flash. Check your logs, the minimal thinking error, and the Jan 1, 2027 price doubling. Scanner and test included.
Wed Oct 28 2026 · 9 min read · 2 views
Claude Sonnet 4.5 retires Nov 30: what breaks on 5.5
Sonnet 4.5 retires Nov 30, 2026. Five settings now return 400 errors on Sonnet 5.5, and thinking costs can erase the price cut. Scanner and cost table inside.
Wed Oct 28 2026 · 10 min read · 1 views
AI coding tools hid 15 GB on my Mac. unclog found it
unclog is a free CLI that finds the caches, logs and orphaned processes AI coding tools leave behind. Here is how to install and use it.
Wed Oct 28 2026 · 11 min read · 3 views
Is your MCP server exposed? 3 of 5 bind every interface
We ran five MCP server setups with default settings. Three listened on every network interface and answered a request from the LAN address. Check yours in 10 seconds.
Wed Oct 28 2026 · 9 min read · 1 views
Outage drill: a fallback saved 2,452 failed AI requests
When a model provider slowed down for 60 seconds in a simulation of 12,000 requests, a client with plain retries failed 18.5 percent of them and sent 1.62 provider calls for every user request. Adding
Fri Oct 23 2026 · 11 min read · 1 views
pgvector plus a tenant filter found 4% of the right results
On 100,000 vectors, a pgvector HNSW index with a WHERE tenant = ... filter returned 3.6 percent of the true nearest neighbors, and still reported success. The same query on Qdrant found 99.95 percent.
Fri Oct 23 2026 · 10 min read · 1 views
Checkpointed agent jobs wasted 5x fewer model calls
Retrying a whole ten-step AI agent job after any failure wasted 4,143 model calls on 2,000 jobs, lost 75 jobs for good, and ran the email step twice for 273 jobs. Saving a checkpoint after each step
Fri Oct 23 2026 · 10 min read · 2 views
Duplicate usage events overbilled an AI SaaS bill by $115
A plain usage counter overbilled $3,266.75 of AI usage by $114.86, a 3.5 percent error, because the queue delivered 6,984 events twice. All 500 tenants were billed wrongly. Recording each event once
Fri Oct 23 2026 · 9 min read · 1 views
A noisy tenant made everyone wait 9.6 seconds: four fixes
In a shared FIFO queue, one tenant's burst of 2,000 AI jobs made every other tenant wait 9.6 seconds at the 95th percentile. A fair queue with no hard cap cut that to 27 milliseconds, and the noisy
Fri Oct 23 2026 · 10 min read · 1 views
What an AI chat request really costs: $0.017 or $0.0035
One request to an AI support chat costs $0.0168 on Claude Sonnet 5.5 when every call resends the full prompt. Two changes cut that to $0.0035: prompt caching and sending easy questions to a cheaper
Fri Oct 23 2026 · 10 min read · 2 views
AI SDK 7 upgrade: the codemod compiles, but a tool breaks
We upgraded a 55-line AI SDK 6 chat app to version 7.0.148 and ran the official codemod. TypeScript went from 4 errors to 1, then to 0 after one manual fix, but two things still changed at runtime
Thu Oct 22 2026 · 10 min read · 0 views
What an AI-resolved support ticket really costs in 2026
AI customer support agent pricing per resolution runs from $0.50 to about $4.00 for the same solved ticket. HubSpot charges $0.50 per resolved conversation, Fin charges $0.99, and Zendesk charges
Mon Oct 19 2026 · 10 min read · 3 views
What Claude's new OpenTelemetry traces actually send out
Claude API OpenTelemetry traces are now in public beta, and the default setting sends no prompts at all. We ran the Anthropic Python SDK against a mock server and counted what leaves your app.
Mon Oct 19 2026 · 10 min read · 2 views
Porting AutoGen to Microsoft Agent Framework: what breaks
An AutoGen to Microsoft Agent Framework migration is mostly a rename, except for four defaults that change silently. We ported a two-agent writer and reviewer team on Python 3.15.0, from AutoGen 0.7.5
Mon Oct 19 2026 · 10 min read · 1 views
Gemini 4 Argon price doubles December 1: what you will pay
Gemini 4 Argon API pricing doubles on December 1, 2026. Standard input goes from $2.00 to $4.00 per million tokens, and output goes from $10.00 to $20.00, according to Google's own pricing page. We
Mon Oct 19 2026 · 10 min read · 2 views
MCP went stateless: what breaks when versions mix
The MCP 2026-07-28 migration is less frightening than the headlines suggest. The Model Context Protocol removed sessions and its connection handshake, and the official MCP blog says the release
Mon Oct 19 2026 · 10 min read · 4 views
Copilot Pro at $10: how far do 1,500 AI credits go?
Copilot Pro costs $10 a month and includes 1,500 GitHub AI Credits. How far that goes depends almost entirely on the model you pick. On GitHub's published rates, the same agent session costs 3.4
Sun Oct 18 2026 · 10 min read · 3 views
Is your Go MCP server open to DNS rebinding? Check v1.4.0
CVE-2026-34742: Go MCP SDK before 1.4.0 accepts any Host header on localhost HTTP servers. We tested v1.3.1, v1.4.0 and v1.8.0, and show the fix and a proxy trap.
Sat Oct 17 2026 · 9 min read · 1 views
OpenAI shuts down GPT-4 and o1 on October 23: what to do
OpenAI retires gpt-4, gpt-4-turbo, o1, o3-mini, o4-mini and fine-tunes on October 23, 2026. A migration map with real costs: some replacements cost 4 to 5 times more.
Sat Oct 17 2026 · 9 min read · 7 views