Snack Pack #1: Your agents got a babysitter
Week of Oct 9–15, 2026 · 7 snacks · ~7 min
Your agents got a babysitter.
Copilot CLI’s Autopilot no longer stalls on a hook’s “ask” when nobody is watching, and Cline changed its default models for unpinned setups on Cline, OpenRouter and about 20 other providers. Plus OpenCode stops auto-updating and Copilot learns to fix your security alerts.
Nutrition Facts
Serving: 1 Snack Pack · ~7 min
Counts from this issue · 15 cards · Hype: 0 g.
Catch up on the week’s agent changes that touch your setup. Snack Pack #1 lives right here on the web, and every bite links to its source.
No funding fluff. No “AI will change everything.” Just what shipped, what changed, and why it might matter to you. A fresh pack lands every Thursday.
Copilot CLI: hooks that don’t hang the babysitter
Copilot CLI Autopilot no longer stalls on a hook’s “ask” when nobody is watching.
- One more folder:
/add-dirgrants sandbox access for the session; interactive settings nudge you to mask likely secrets. - No more stalled Autopilot: agent-scoped declarative hooks; Autopilot now denies a preToolUse hook’s “ask” instead of stopping on a permission dialog nobody is watching.
- New model, easier answers: Gemini 4 Argon support;
/askexpands with Ctrl+E, copies with Ctrl+Y.
Who cares Terminal Copilot users running Autopilot / unattended loops.
Source: Copilot CLI v1.0.97 release notes · Copilot CLI v1.0.96 release notes · Copilot CLI v1.0.98 release notes
Cline swapped your default model
Check your model: unpinned Cline setups just got a new default. Cline + OpenRouter users now land on Claude Sonnet 5.5. About 20 other providers also shuffle defaults, so unpinned users will wake up on a different model. Pin the model you want to keep.
Who cares Cline users who never pinned a model. Check the cost dashboard before you blame the agent.
Source: Cline v4.1.24 release notes · Cline Desktop v0.0.46 release notes
OpenCode 1.19: the on-ramp to OpenCode 2 (and auto-update is off)
Action needed: OpenCode 2 is here, and updates are no longer automatic. opencode upgrade --major (or /update in the TUI) walks you there, so upgrade on purpose. It also allows OpenAI’s ultrafast service tier.
Who cares OpenCode folks. Deliberate upgrades > surprise majors.
Source: OpenCode v1.19.0 release notes
Copilot will now fix your security warnings for you
Copilot can now fix your security alerts and prove it. Assign a CodeQL alert and it patches the code, re-runs CodeQL to prove the alert is gone, then opens a draft PR. Up to 25 alerts per batch. Now GA.
Default-branch alerts only: an enterprise admin has to turn on the “Agentic autofix” policy, and each run spends Code Security AI Credits + Actions minutes. You can also start it via the REST API (copilot-swe-agent[bot]).
Who cares Security folks staring at a CodeQL mountain on GHEC / GHAS.
Source: GitHub Changelog
HydraFusion stops making you pick a model (and adds a “Max” mode)
HydraFusion gets a bigger model pool and a Max tier. Its model pool now includes Claude Opus 5.5, GPT-6 Astra, GPT-6.1 Sol, and GPT-6 Luna. Cascade gets a judging step; Critique can tool-snoop the workspace with parallel critics. New HydraFusion Max is the “go deep” button.
Lives in VS Code, the Copilot app, and Copilot CLI.
Who cares Copilot Pro, Pro+, Business and Enterprise (with preview features on).
Source: GitHub Changelog
Claude Code: idle chats now compact before the cache expires
Walk away from a long chat and Claude Code tidies it up. After ~54 idle minutes, a 200K+ token chat compacts before the prompt cache expires, so your next message doesn’t re-chew the whole transcript. Tune the threshold with CLAUDE_, or opt out with "idle.
Who cares Anyone with marathon Claude Code sessions — or a token bill that makes them wince.
Source: Claude Code v2.1.299 release notes · Claude Code v2.1.300 release notes
Cursor Origin opens the snack aisle
Cursor’s Origin now plugs into the apps you already use. That’s 20+ apps: Cloudflare, Vercel, Buildkite, CircleCI, CodeRabbit, Greptile, Semgrep, Socket, Sentry, Linear, and friends. Public submissions welcome. Apps auth with app JWT + install tokens, get webhooks, write check runs / PR comments / reviews.
Who cares Dev-tool makers hunting distribution, and teams wiring CI/review bots into Origin.
Source: Cursor changelog
What’s a hook?
Hooks are house rules your agent can’t skip. A hook is a small command that Copilot cloud agent and Copilot CLI run at set moments: when a session starts, when you enter a prompt, or right before it uses a tool.
The before-a-tool one (preToolUse) can approve or deny that step (GitHub Docs: About hooks).
Try this: peek in your repo for a . folder, then read our Copilot CLI hooks Daily to see why an “ask” no longer freezes Autopilot. New here? Start here.
Source: GitHub Docs: About hooks
Agent Skills: recipe cards for your AI
Tired of re-explaining the same steps to your AI every single chat? An Agent Skill is a recipe card it keeps in a folder (SKILL.md): only the name and blurb sit on the menu at startup (about 100 tokens), and the full recipe comes out when your task matches.
Skills teach your agent how to cook; new hands to grab your tools are MCP’s job.
Full Main Course: 6-min read
Sources
Show all 6 sourcesShow fewer
Pin your model before the defaults move
Try it now:
- Open your agent settings (Cline, Copilot, Claude Code, Cursor — wherever you live).
- Explicitly pick the model you want this week so a quiet release cannot swap the chef under you.
- Recheck cost dashboards after any "default model" note in release notes — surprises usually show up on the bill first.
Source: Cline v4.1.24 release notes
Claude Agent Skills Explained
Plays from YouTube when you press play.
Video Summary
- Skills are expertise Claude loads only when it needs it. Each one is a folder that kicks in automatically when a task matches.
- Only each skill’s name and description load at startup; the full SKILL.md loads on demand.
- Compares skills with CLAUDE.md (project facts), MCP (connects to data) and subagents (fixed roles).
On screen: Mostly a code editor, with skill .md files, CLAUDE.md, an mcp.json with Playwright, database-migrations.md (model: sonnet), and full-stack-features.md stacking skills.
After watching you can: Tell skills, CLAUDE.md, MCP servers and subagents apart, and know when to use each.
Short on time? Watch just:
MCP reference servers
The classics grew up. Filesystem, memory, everything and sequential-thinking reached their first stable release on npm; fetch, git and time got fresh PyPI builds.
Anthropic claude-api skill
Meet Claude’s cheat sheet for the Messages API. It covers dynamic workflows, four quickstarts, and Haiku 5.5 as the current Haiku.
GitHub MCP Server
Your agents can stop guessing what GitHub sends back. GitHub’s official server now ships structured outputs and output schemas for Code Mode agents (schemas need MCP 2026-07-28+ clients).
The agents got a babysitter
Auto-denied permission prompts, idle compaction, autofixes that recheck themselves — this week was less “smarter model,” more “you can actually walk away.”
- Update now: Gemini CLI now checks folder trust in headless runs too; update to 0.64.0 or later (Gemini CLI v0.64.0 release notes).
- Devin Desktop speaks a newer MCP revision and lets admins pin the extension marketplace.
- Codex CLI 0.163 lets you steer it mid-answer (our Daily).
- No Aider release in the window.
Source: Gemini CLI v0.64.0 release notes
That’s the pack. Pass a snack to a friend.