Agents After Dark, by Prefactor

Agents After Dark, by Prefactor

Archives
Log in
Subscribe
11 November 2026

The Agent Brief — Nov 11

Prefactor
Get Ready for Agents
The Agent Brief
Wednesday 11 November · what moved in the agent world today
Focus areas
Heating up
ClaudeClaude codeAgentic AI production deploymentAI agents pharmaAi codingMCP
Cooling off
AnthropicAI agents production deploymentWebmcpAgent governanceCursorOpenai
Builders · 4
Anthropic releases Claude Fable 5.2 and Mythos 5.2 with 30% faster output and API price cuts
Anthropic released Claude Fable 5.2 and Claude Mythos 5.2, with Fable 5.2 scoring 65 on the Artificial Analysis Intelligence Index and priced at $8 per million input tokens and $40 per million output tokens. The models generate output up to 30% faster than prior versions and include an optional fast mode that speeds token generation by up to 2.5 times. During prerelease testing, the models designed three working electronic devices from scratch and solved two open problems in black hole physics.
Read articleRead article
Anthropic faces lawsuits over Claude Max subscription performance claims and service reliability
According to Reuters and Reddit reports, Anthropic faces multiple lawsuits challenging Claude Max subscription marketing claims at $200 per month, with users reporting model performance degradation without notice and concerns about autonomous execution without proper safeguards. The lawsuits allege the service fails to deliver advertised capabilities and does not maintain consistent performance levels for subscribers.
Read articleReddit thread
PSI uses unreleased Anthropic model to solve two-decade-old quantum computing problem
According to TechCrunch, Physical Sciences Inc. (PSI) used an unreleased Anthropic Claude model to answer a fundamental quantum computing research question that scientists had pursued for nearly 20 years. The achievement demonstrates Claude's capability on scientific discovery tasks, though the specific problem and methodology were not detailed in available reporting.
Read article
eBay blocks AI agents including Meta Muse from autonomous platform browsing
According to The Verge, eBay has blocked AI agents from accessing its platform, preventing tools like Meta's Muse from autonomously browsing products, curating digests, and generating shopping links for users. The move restricts agent capabilities for e-commerce use cases and represents a platform-level control on autonomous agent access.
Read article
Enterprise · 1
Mintigate releases control layer and audit trail for autonomous AI coding agents on repository tasks
According to GitHub documentation, Mintigate provides an audit trail and control layer for coding agents like Claude Code operating autonomously on repository tasks. The tool allows teams to trace agent actions and conditionally approve or deny agent create/modify/delete operations.
Read article
Read the full brief →
A daily intelligence digest by Prefactor — every weekday, traced to its sources.
Don't miss what's next. Subscribe to Agents After Dark, by Prefactor:
← Newer The Agent Brief — Nov 12 Older → The Agent Brief — Nov 10
LinkedIn
Twitter
youtube.com
www.prefactor.ai
Powered by Buttondown, the easiest way to start and grow your newsletter.