---
title: 'In the News: November 3, 2026 (Morning)'
description: "Anthropic's message threads send only new turns to the Messages API; it reports median time to first token down about 35% at 500k to 1M tokens."
canonical_url: 'https://darkfactory.dev/news/2026-11-03-morning'
markdown_url: 'https://darkfactory.dev/news/2026-11-03-morning.md'
collection: news
date_published: '2026-11-03T07:10:00-05:00'
date_modified: '2026-11-03T07:10:00-05:00'
---

# In the News: November 3, 2026 (Morning)


Anthropic added a way for long agent sessions to stop resending their whole history on every request, and says the change cuts the wait before the first token. Separately, Kiro's applied science team reports that it checked 26 functions in its own agent harness with Lean and reproduced 18 bugs.

## 1. Anthropic says message threads cut median time to first token by about 35% for requests of 500k to 1M tokens

**[Make long-running agents faster with message threads](https://claude.com/resources/articles/message-threads)** · Anthropic · Claude Platform announcement, November 2, 2026

Message threads is in public beta on the Claude Platform. Today a coding agent resends every earlier message and tool result with each request. With a thread, the first request carries the full conversation and starts the thread. Every later request carries only the new messages and the ID of the previous response. Anthropic says each thread is held in memory, encrypted and not written to disk, and expires 5 minutes after its last request, after which the client resends the full conversation. The post says the feature is on by default for most Claude Code users and that requests of 500k to 1M tokens see "median TTFT drop by about 35%." It also lists availability on Claude Platform on AWS, Claude in Microsoft Foundry, and Claude Models on Google Cloud (Anthropic managed).

The 35% figure is Anthropic's own, and we did not test it. The post states that Claude still reads the full conversation on every turn, so token usage stays the same, and that standard token rates and rate limits apply.

**Why it matters:** The gain is latency, not cost. For a harness that runs many tool-call turns against a long context, the saving belongs to active loops. A pause longer than the 5-minute expiry, such as a person reviewing a diff, means the next request goes back to a full resend.

## 2. Kiro's team says it modeled 26 functions of its own agent harness in Lean, proved 70 properties and reproduced 18 bugs on the real code

**[Kiro proves properties of its own harness in Lean](https://kiro.dev/blog/harness-properties-in-lean/)** · Yaroslav Kharkov, Ferhat Erata, Behrooz Omidvar-Tehrani and Rajdeep Mukherjee, Kiro Applied Science · Kiro blog, November 2, 2026

The authors had the Kiro CLI agent, running Claude Opus 5.5 on Amazon Bedrock, write Lean 4 models of 26 functions across three codebases: the Kiro agent harness, the CLI client and Kiro Crew. The agent then wrote proofs and differential tests that run each model against the real TypeScript, Rust or Python code. Of 96 candidate properties, Lean proved 70, two got bounded checks, 18 were falsified in the model and reproduced on the analyzed code, and six were partial or observations. The final differential runs covered 97,938 cases with no mismatches.

Two of the proved properties: a sub-agent's combined permission decision is never more permissive than its parent's, and truncated tool output never exceeds its byte cap. Among the bugs, the agent server's web fetch tool budgets in bytes but cuts by JavaScript string length, so a page of Chinese or Japanese text can come back up to 1.5 times the cap. In Kiro Crew, cancelling a one-shot job in the moment before its work begins leaves the job switched on, so the scheduler runs it again (PR 25130). The authors say none of the bugs were security findings.

These are the team's internal results. The agent server and CLI repositories are private, and only Kiro Crew, at commit 909e0fc, is public. The authors state that a proof covers the model, not the code, and that their differential tests are sampled evidence of a match. The code listings and Figure 1 did not load in our read of the page.

**Why it matters:** The pairing is the part that transfers. A proof about a model tells you little until a differential test ties it to the code, and most of the bugs here sat where two representations meet, such as UTF-16 against UTF-8 or a file path against a glob. If your harness enforces byte caps, permission inheritance or a stable cache prefix, those are the short functions this method can check.

---

## Also this cycle

- **[Claude Code v2.1.317](https://github.com/anthropics/claude-code/releases/tag/v2.1.317)** · Anthropic, GitHub release, November 3, 2026 (01:25 UTC) · Adds `onFailure: "block"` for Agent SDK hook callbacks, so a callback that fails blocks the action instead of letting it run. Also adds a prompt before a web or shell tool call sends your account email address to the network, and a cost report endpoint on the Claude apps gateway admin API, `GET /v1/organizations/analytics/cost`.