Agent Tools Today, 10 Nov 2026: Claude Fable 5.2, an Opus 5.5 Nerf Thread, Orca's Phone Agents Tab

Anthropic released Claude Fable 5.2 and Mythos 5.2 on 9 Nov 2026, describing the two as "the same underlying model" and pricing them at $8 per million input tokens and $40 per million output tokens, with outputs up to 30% faster than Claude Fable 5.1 and Claude Mythos 5.1. Anthropic said Fable "has strong cybersecurity and biology safeguards to guard against misuse, which allows us to make it available to all users," and the release drew 431 points and 732 comments on Hacker News by 10 Nov. Claude Code 2.1.321, published 9 Nov at 19:22 UTC, made Claude Fable 5.2 the default Fable model and fixed an MCP OAuth flaw that sent secrets such as refresh tokens to another host when a server redirected them, plus a case where a read-only Bash command ran outside the sandbox without a permission prompt.

Agent Tools Today, 10 Nov 2026: Claude Fable 5.2, an Opus 5.5 Nerf Thread, Orca's Phone Agents Tab The top launches, skills, multi agent moves, rising repos and talking points in AI agents from 9 to 10 Nov 2026, each with its source and date. Agent tools today, 10 Nov 2026, in one line: Anthropic released Claude Fable 5.2, “Is Opus 5.5 nerfed?” drew 778 Hacker News points, and Orca 1.4.255 adds an Agents tab on your phone. It covers 9 Nov 04:00 UTC to 10 Nov 04:00 UTC, checked 10 Nov. Previous edition: 9 Nov https://munderdiffl.in/blog/agent-tools-today-2026-11-09/ . Top launch top-launch 1. Claude Fable 5.2 and Mythos 5.2. Anthropic’s page, dated 9 Nov, says the two “are the same underlying model”, and that Fable “has strong cybersecurity and biology safeguards to guard against misuse, which allows us to make it available to all users.” They “generate outputs up to 30% faster than Claude Fable 5.1 and Claude Mythos 5.1”, the vendor’s own figures, at “$8 per million input tokens and $40 per million output tokens” as of 9 Nov. 431 points and 732 comments on Hacker News https://news.ycombinator.com/item?id=50429068 by 10 Nov. Source https://www.anthropic.com/claude-fable-and-mythos-5-2 Top 2 skills, MCP servers and plugins top-2-skills-mcp-servers-and-plugins 1. Orca’s plugin directory. On 10 Nov the directory shows a “Featured” row, “Most installed, one per category”, above a “Newly added” list that reads “128 added this week”, with entries “Added Nov 9, 2026”. The top featured plugin, Ayu for Orca, shows 2.7k installs. Source https://www.onorca.dev/plugins 2. ha-mcp. “Configure and automate Home Assistant with Claude, ChatGPT, and other AI assistants.” On GitHub trending on 10 Nov with 31 stars in a day. MIT, 6,150 stars. Source https://github.com/homeassistant-ai/ha-mcp Top 5 moves from multi agent tools top-5-moves-from-multi-agent-tools 1. Claude Code 2.1.321. Published 9 Nov at 19:22 UTC. It adds Claude Fable 5.2, “now the default Fable model”, and fixes “MCP OAuth requests sending secrets such as refresh tokens to another host when a server redirected them” and “a case where a read-only Bash command ran outside the sandbox without a permission prompt”. Version 2.1.322, at 02:29 UTC on 10 Nov, adds “fast mode on Vertex in the global, us and eu regions”. See Claude Code privacy and secrets https://munderdiffl.in/blog/claude-code-privacy-and-secrets/ . Source https://github.com/anthropics/claude-code/releases/tag/v2.1.321 2. Orca 1.4.255. Published 9 Nov at 12:24 UTC. “A new Agents tab lists agents across your connected desktops.” Also: “Switching Codex accounts moves your open Codex chats to the account you choose.” Git Graph, new a day earlier, now lets you “Check out, cherry-pick or revert commits”. Source https://github.com/stablyai/orca/releases/tag/v1.4.255 3. T3 Code v0.1.8. A stable release, published 9 Nov at 22:46 UTC. The notes list “agents ask for secrets through request secret instead of chat”, “render MCP Apps from Claude tool calls” and “list Claude models reported by Claude Code”. See T3 Code alternatives https://munderdiffl.in/blog/t3-code-alternatives/ . Source https://github.com/pingdotgg/t3code/releases/tag/v0.1.8 4. Pi v1.5.1. Published 9 Nov at 23:47 UTC. You can now “use pi auth login <provider and pi auth logout <provider without starting a session”, and “eligible caches now stay warm for up to 30 minutes between runs when a refresh is expected to pay for itself.” The anthropic provider’s default model “is now Claude Opus 5.5”. See Pi agent alternatives https://munderdiffl.in/blog/pi-agent-alternatives/ . Source https://github.com/earendil-works/pi/releases/tag/v1.5.1 5. Goose 1.58.0. Published 9 Nov at 19:07 UTC. New features include a “Context window report and viewer”, a “Lemonade provider” and “Collapse thinking by default with a setting to control it desktop ”. One fix: “Use recipe settings in scheduled runs”. Source https://github.com/aaif-goose/goose/releases/tag/v1.58.0 Top 2 rising repos top-2-rising-repos 1. Astryx. Meta’s “open source design system that’s fully customizable and agent ready”. The README says it “grew inside Meta over the last eight years” and is “Currently in Beta”. 3,278 stars in a day on GitHub trending on 10 Nov, up from 96 in a day on 8 Nov, 20,792 in total. MIT. Source https://github.com/meta/astryx 2. numbrone. “A chief of staff for your coding agents.” Each project’s manager runs “workers in parallel, each in its own git worktree”, all inside herdr, and the README says it “is not made by or affiliated with the herdr team.” MIT, created 8 Nov at 21:30 UTC and missed by the 9 Nov edition; 111 stars on 10 Nov. See herdr alternatives https://munderdiffl.in/blog/herdr-alternatives/ . Source https://github.com/stevenmichaelthomas/numbrone Stars can rise for reasons other than use: names to watch, not picks. Top 5 things people are talking about top-5-things-people-are-talking-about 1. Is Opus 5.5 nerfed? Some users say so. The Hacker News item “Is Opus 5.5 nerfed? Thinking patterns changed on November 3” had 778 points and 484 comments by 10 Nov, posted 9 Nov at 07:04 UTC. This rests on the title and the thread. Anthropic’s 9 Nov page still calls Opus 5.5 its “strongest daily driver”. Source https://news.ycombinator.com/item?id=50422069 2. Does ChatGPT Desktop forward your SSH agent? The poster says it does: “The agent running on a remote machine can authenticate using your local SSH credentials. That means it can push to your private repos on GitHub, for example.” The post points to openai/codex pull request 28131, titled “ codex Refresh SSH agent for app-server proxy”. 207 points by 10 Nov, posted 9 Nov at 08:17 UTC. Source https://news.ycombinator.com/item?id=50422628 3. Does Gemini send Drive files to human reviewers? A GitHub repo created 9 Nov claims so, under the title “Google Gemini routes personal Drive files and emails to human annotators”. It says the capture is “probabilistic”. These are the author’s claims. 523 points on Hacker News https://news.ycombinator.com/item?id=50429426 by 10 Nov. Source https://github.com/thisisme6598/gemini files 4. Can Fable 5.2 build a whole game? Lex Toumbourou says it “one-shotted my nephew’s dream game from a bunch of drawings”. His first version, shipped in June, was built “mostly with Claude Code, using Opus 4.8 and Fable 5”. 306 points and 220 comments on Hacker News https://news.ycombinator.com/item?id=50433659 by 10 Nov. Source https://notesbylex.com/fable-5-2-one-shotted-my-nephews-dream-game 5. Is Fable 5.2 the top model on Artificial Analysis? The Hacker News title says “highest so far”. The Artificial Analysis page says Fable 5.2 “scores 65 on the Artificial Analysis Intelligence Index” against a median of 40, and calls it “notably slow” at 46.4 minutes per task. Smart and slow: a familiar office type. 329 points and 239 comments on Hacker News https://news.ycombinator.com/item?id=50431144 by 10 Nov. Source https://artificialanalysis.ai/models/claude-fable-5-2 How this list is built how-this-list-is-built We read tracked tools’ GitHub releases, GitHub trending, the Hacker News top 30 and leads from Google and X, and opened every page linked above. A name only makes a list with an opened link and a signal: points, stars or a dated release. Codex shipped only alpha builds, and Gemini CLI, OpenCode, Crush, Paperclip, AionUi, OpenClaw and Cline shipped no stable release in the window. Copilot CLI 1.0.108 adds “/index commands”, and Qwen Code v0.27.1 can “enable LSP through a user-scoped setting and env pair”.