Diff

Managed Agents now auto-evaluates each tool call individually

Managed Agents permission policies now include auto: instead of blanket rules (always run / always block a given tool type), the server evaluates each agent.tool_use and agent.mcp_tool_use call and decides to run, deny, or pause for human approval on a per-call basis. The evaluation decision surfaces in the event's evaluation field alongside evaluated_permission, so you can audit why each call was handled the way it was. Previously, granular per-call oversight meant either manual approval routing or complex static policy trees; auto makes the server the dynamic per-call gatekeeper. If Policy Assistant uses Managed Agents for protocol execution, this is the upgrade path from static permission policies to dynamic per-call evaluation — closer to the Sentinel pattern Meta built for Muse, but inside the Claude Platform.

via claude platform release notes · claude platform docs

shiftmodels

DeepSeek V4.1 Flash: 552B MoE, native vision, 60% cheaper cached reads

DeepSeek released V4.1 Flash on September 10 — a 552B-parameter mixture-of-experts model (8B active on input, 16B on output) with native vision, 1M token context, and a new Causal Encoder-Decoder architecture that cuts the KV cache to ~890 bytes per token (roughly one-quarter of V4-Flash). Pricing: cached input at $0.003/MTok (60% lower than V4-Flash), uncached at $0.15/MTok, output at $0.60/MTok. V4 Pro is being discontinued. The cached-read price is the headline: for cache-heavy agentic loops, V4.1 Flash is now competitive with Fable 5.1's $0.25/MTok cache reads on a cost-per-task basis, while being multimodal. If you track model economics and cost per agentic task, this is the new reference point for what a Flash-tier open-weight model costs at scale.

via deepseek (x announcement) · cellcog · intelligent living

techniquetooling

Fable 5.1 thinking blocks persist across turns — and bind to your prefix

Since the September 1 launch, Claude Fable 5.1 preserves thinking blocks in conversation history so the model can build on its own previous reasoning in later turns. Two constraints matter for production: (1) thinking blocks replay only to Fable 5.1 or newer — earlier models silently receive them empty; (2) on accounts created August 31 or later, if anything before a thinking block changes between turns (system prompt, tools, earlier messages), replaying that block returns a 400 error. The thinking-binding-controls-2026-08-01 beta header gives you control: prefix_mismatch_behavior chooses between rejecting or silently dropping mismatched blocks, and input_transformations in the response reports what was dropped. If Policy Assistant runs multi-turn sessions that inject per-turn context into the system prompt or swap tool specs mid-conversation, this binding behavior is the new constraint to test against — it can silently fail unless you opt in to the beta header.

via claude platform release notes · claude platform docs — preserved thinking

Meta Muse: dedicated VM + Sentinel gatekeeper between agent and internet

Meta launched Muse, a personal AI agent where each user gets an isolated VM ("Muse Secure VM") and a separate gatekeeper model ("Sentinel") that must approve every outbound network request before it leaves the environment — nothing reaches the internet without Sentinel's sign-off. The containment is architectural: one model acts, a second model controls egress. This is the first production consumer agent to publish its security architecture at this level of specificity, and the Sentinel pattern is the concrete engineering answer to how you actually sandbox an agent with real-world tool access. Agent safety and sandboxing as an engineering discipline — not a policy statement — is listed as an interest in PROFILE.md; this is the reference design.

Fable 5.1 cache reads dropped 75% — now $0.25/MTok

Cache reads on Claude Fable 5.1 cost $0.25/MTok (0.025× base input price), compared to $1/MTok (0.1×) on Fable 5 and all other models. For a heavily cached agentic loop with a large system prompt or tool spec, this cuts the cache-read cost by three-quarters automatically once you are on Fable 5.1. The cost improvement alone strengthens the Fable 5.1 migration case — though the breaking change on tool_choice: any/tool (those now return 400; switch to auto or the strict tool-use pattern) needs a check before migrating, if Policy Assistant forces tool selection via tool_choice.

`clear_at` lets per-turn reminders render once and cost nothing after

clear_at: "next_user_message" on a mid-conversation system message makes it render once (after a tool result, for example), then silently drop to zero tokens on every subsequent request — while staying in the array byte-for-byte so the prompt cache prefix stays intact. The pattern it solves: per-turn reminders like "verify your previous step" or "request independent reads in one turn" that would otherwise pile up in message history and inflate costs over a long agentic loop. The constraint worth knowing: the cleared message must be re-sent byte-for-byte unchanged on every subsequent call; editing or removing it breaks the cache and any bound thinking blocks. If Policy Assistant has per-turn instruction patterns in its protocol loops, this is the clean fix.