---
collection: frontier-ai-labs
collection_name: Frontier AI Labs
week_start: 2026-09-14
title: Claude rebuilds Projects as a live conversation while Codex and OpenAI's SDKs race ahead
release_count: 37
generated: 2026-09-21
canonical: https://releases.sh/collections/frontier-ai-labs/digest/2026-09-14
---

# Claude rebuilds Projects as a live conversation while Codex and OpenAI's SDKs race ahead

Anthropic turned Claude Projects from a folder of chats into a single multi-threaded conversation that runs in the cloud, while Claude Code, Codex CLI, and OpenAI's SDKs all landed substantial feature releases the same week.

### Projects become a workspace, not a folder

The week's biggest structural change came from Anthropic, which [rebuilt Projects as a single multi-threaded conversation](/release/rel_PCagz-yiYqhvuxtsoCQts-claude-redesigns-projects-as-a-single-multi-threaded-conversation): instead of nesting chats inside a folder, a project is now one thread where each request spawns work Claude runs in the cloud and reports back on, distinguishing what's ready for review from what's waiting on you. It's in beta for select Pro and Max users and rolling out across Claude Code, chat, and Cowork — existing projects are untouched for now, which softens the migration but also means two mental models coexist for a while.

That's part of a broader consolidation. Anthropic [merged chat and Cowork into one unified Claude](/release/rel_lu24lHJ2YPLlDCTvhxAJY-anthropic-merges-claude-cowork-and-chat-into-one-unified-claude), so a quick question and a long-running task live in the same conversation, and [launched Claude Slides, Design, and Docs in beta](/release/rel_FpkOB2NeJrh0gN4cdMUtb-anthropic-launches-claude-slides-claude-design-and-claude-docs-in-beta) so a launch brief can turn into a deck, social images, and a one-pager without leaving the chat. Both push toward Claude as the place work happens rather than a place you ask questions. On the enterprise side, a [Salesforce plugin with 37 pre-built sales skills](/release/rel_BULe5EjELbqY3icbx4dUR-claude-launches-salesforce-plugin-with-37-pre-built-sales-skills) extends that reach into CRM data.

### Claude Code hardens its session machinery

Claude Code shipped a dense run of releases, mostly about making long sessions survive contact with reality. The most consequential is [v2.1.275](/release/rel_1SjjEYFg34YFzPatqXk3o-claude-code-v2-1-275-fixes-prompt-cache-misses-and-hardens-plugin-credential), which fixes a restored memory file's age note drifting between requests after compaction or resume — a subtle bug that silently caused prompt cache misses on every turn — and stops plugin and marketplace output from leaking passwords or tokens embedded in git, ssh, or marketplace URLs. Alongside it, [2.1.274](/release/rel_uBw6beGnIDuiTP1Kk_lDV-claude-code-2-1-274-self-heals-corrupted-transcripts-and-bounds-mcp-startup) makes sessions that were stuck retrying "unexpected tool_use_id" errors self-heal corrupted transcripts or exit cleanly with a `/rewind` hint, and adds `CLAUDE_CODE_MCP_STARTUP_WAIT_MS` to bound how long the first non-interactive turn waits on MCP servers. [v2.1.277](/release/rel_cbueYirXLzGc0E7uNzZJ_-claude-code-v2-1-277-adds-agents-md-support-and-fixes-p-hang) brings AGENTS.md fallback for projects without a CLAUDE.md and fixes `claude -p` hanging after internal errors — which now exit with code 1 — while [2.1.271](/release/rel_FZcAuO_f0_zkSbkd7zvaS-claude-code-2-1-271-adds-fast-mode-to-remote-sessions-and-per-command-allowed) adds per-command `allowed_domains` to sandboxed auto mode so each Bash or PowerShell command reaches only the hosts it needs, plus fast mode in Remote sessions. Smaller follow-ups cleaned up after the rush: [v2.1.276](/release/rel_jNbTm_lkX5SXHDGV-dn_0-claude-code-v2-1-276-fixes-400-errors-behind-proxies-and-gateways) patched a 400-error regression for proxy and gateway users, and [v2.1.278](/release/rel_WabQz8DpxX8c3guIq3YEd-claude-code-v2-1-278-defaults-auto-mode-to-server-side-classifier-for-api) defaults auto mode to the server-side classifier for API, Bedrock, Vertex, and Foundry, avoiding classifier overhead charges.

### Codex goes multimodal, OpenAI's SDKs worldwide

OpenAI's [Codex CLI 0.155.0](/release/rel_JotzQfuFf_u8NV4btlouH-codex-cli-0-155-0-adds-experimental-voice-conversations-and-touch-id-for-mcp) added experimental `/voice` conversations with live transcripts and microphone controls, Touch ID verification for MCP requests in local TUI sessions on supported Macs, and a fix for a Windows-process escape from restricted WSL sandboxes — security work that pairs with its hardening of brokered shell snapshots. A quick [0.155.1](/release/rel_ouHUMdTjnl8p6YcHlBizT-codex-cli-0-155-1-disables-reasoning-summaries-by-default-in-new-tui-sessions) follow-up disabled reasoning summaries by default where providers reject them, and [Codex for iOS](/release/rel_yHCdCk5m7n-MZ5kfJ2Cz2-codex-for-ios-adds-folders-in-file-picker-and-writing-blocks) gained folders in the file picker and writing blocks.

The SDKs coordinated on a shared API surface: [Python 3.15.0](/release/rel_x03MYYurZiKM-sTEI-8Zv-openai-python-sdk-v3-15-0-adds-managed-responses-websocket-sessions) led with managed Responses WebSocket sessions, prompt-cache prewarming, and compaction progress events, with [Node v7.18.0](/release/rel_7WcQA0em--7h0-BKB6EiW-openai-node-sdk-v7-18-0-adds-managed-responses-websocket-sessions) and [Go v3.62.0](/release/rel_pEEe-voonq7GzpsN-xEw9-openai-go-sdk-v3-62-0-adds-managed-responses-websocket-connections) following with the same managed WebSocket sessions. All three then added [webhook endpoint management](/release/rel_IU1ZpWPnTj8v-tIysyvPC-openai-python-sdk-v3-16-0-adds-webhook-endpoint-management) and deprecated MCP `connector_id` across [Node](/release/rel_gxNBPlh3kUTXzxuINqZTC-openai-node-sdk-v7-19-0-adds-webhook-endpoint-management-deprecates-mcp) and [Go](/release/rel_3P7Avhd_iUxIub2ljw50b-openai-go-sdk-v3-63-0-adds-webhook-endpoint-management). Deeper in the week, Go and Node added [vault credentials and safety webhooks](/release/rel_yiwN_G9pUediUTUZVPnjt-openai-go-sdk-v3-64-0-adds-vault-credentials-storage-config-and-safety-webhooks), and a [memory leak in Python's `parse_response`](/release/rel_P8WaoBQzQavxv_WlJ-GYX-openai-python-sdk-v3-16-2-fixes-memory-leak-in-parse-response) got fixed. If you maintain an agent against this API, the connector_id deprecation is the migration to plan for.

## Releases covered

### Anthropic

- [Claude redesigns Projects as a single multi-threaded conversation](https://releases.sh/release/rel_PCagz-yiYqhvuxtsoCQts-projects-become-one-conversation-with-threads)
- [Anthropic merges Claude Cowork and chat into one unified Claude](https://releases.sh/release/rel_lu24lHJ2YPLlDCTvhxAJY-claude-cowork-and-chat-merge-into-one-claude)
- [Anthropic launches Claude Slides, Claude Design and Claude Docs in beta](https://releases.sh/release/rel_FpkOB2NeJrh0gN4cdMUtb-claude-slides-design-and-docs-enter-beta)
- [Claude launches Salesforce plugin with 37 pre-built sales skills](https://releases.sh/release/rel_BULe5EjELbqY3icbx4dUR-salesforce-plugin-brings-accounts-and-pipeline-into-claude)
- [Claude Code v2.1.275 fixes prompt cache misses and hardens plugin credential handling](https://releases.sh/release/rel_1SjjEYFg34YFzPatqXk3o-memory-file-age-no-longer-breaks-prompt-cache-plugin-urls-stop-leaking-tokens)
- [Claude Code 2.1.274 self-heals corrupted transcripts and bounds MCP startup waits](https://releases.sh/release/rel_uBw6beGnIDuiTP1Kk_lDV-corrupted-transcripts-self-heal-mcp-startup-wait-bounded)
- [Claude Code v2.1.277 adds AGENTS.md support and fixes -p hang](https://releases.sh/release/rel_cbueYirXLzGc0E7uNzZJ_-agents-md-supported-claude-p-no-longer-hangs-silently)
- [Claude Code 2.1.271 adds fast mode to Remote sessions and per-command allowed domains](https://releases.sh/release/rel_FZcAuO_f0_zkSbkd7zvaS-fast-mode-in-remote-sessions-per-command-allowed-domains)
- [Claude Code v2.1.276 fixes 400 errors behind proxies and gateways](https://releases.sh/release/rel_jNbTm_lkX5SXHDGV-dn_0-gateway-and-proxy-400-errors-fixed)
- [Claude Code v2.1.278 defaults auto mode to server-side classifier for API, Bedrock, Vertex, and Foundry](https://releases.sh/release/rel_WabQz8DpxX8c3guIq3YEd-auto-mode-defaults-to-server-side-classifier-on-bedrock-vertex-foundry)

### OpenAI

- [Codex CLI 0.155.0 adds experimental voice conversations and Touch ID for MCP](https://releases.sh/release/rel_JotzQfuFf_u8NV4btlouH-voice-conversations-and-touch-id-mcp-verification-arrive)
- [Codex CLI 0.155.1 disables reasoning summaries by default in new TUI sessions](https://releases.sh/release/rel_ouHUMdTjnl8p6YcHlBizT-new-tui-sessions-default-to-no-reasoning-summaries)
- [Codex for iOS adds folders in file picker and writing blocks](https://releases.sh/release/rel_yHCdCk5m7n-MZ5kfJ2Cz2-folder-creation-in-file-picker-queued-prompt-fixes)
- [OpenAI Python SDK v3.15.0 adds managed Responses WebSocket sessions](https://releases.sh/release/rel_x03MYYurZiKM-sTEI-8Zv-managed-responses-websocket-sessions-prompt-cache-prewarming)
- [OpenAI Node SDK v7.18.0 adds managed Responses WebSocket sessions](https://releases.sh/release/rel_7WcQA0em--7h0-BKB6EiW-responses-websocket-sessions-prompt-cache-prewarming)
- [OpenAI Go SDK v3.62.0 adds managed Responses WebSocket connections](https://releases.sh/release/rel_pEEe-voonq7GzpsN-xEw9-managed-responses-websockets-prompt-cache-prewarming)
- [OpenAI Python SDK v3.16.0 adds webhook endpoint management](https://releases.sh/release/rel_IU1ZpWPnTj8v-tIysyvPC-webhook-endpoint-management-added-mcp-connector-id-deprecated)
- [OpenAI Node SDK v7.19.0 adds webhook endpoint management, deprecates MCP connector_id](https://releases.sh/release/rel_gxNBPlh3kUTXzxuINqZTC-webhook-endpoint-management-added-mcp-connector-id-deprecated)
- [OpenAI Go SDK v3.63.0 adds webhook endpoint management](https://releases.sh/release/rel_3P7Avhd_iUxIub2ljw50b-webhook-endpoint-management-added-mcp-connector-id-deprecated)
- [OpenAI Go SDK v3.64.0 adds vault credentials, storage config, and safety webhooks](https://releases.sh/release/rel_yiwN_G9pUediUTUZVPnjt-vault-credentials-external-storage-config-safety-webhook-events-added)
- [OpenAI Python SDK v3.16.2 fixes memory leak in parse_response](https://releases.sh/release/rel_P8WaoBQzQavxv_WlJ-GYX-parse-response-memory-leak-fixed)
