Upgrade Zed to v1.25.0-pre and rotate leaked remote-agent MCP secrets
Zed v1.25.0-pre adds ACP context compaction and session notices for external agents, Gemini 4 Argon plus new Bedrock models, and stops remote-project agent/MCP secrets from appearing in the local process list. Upgrade, rotate those secrets, and point agents at gemini-4-argon or the new Bedrock models.Why this loop
This prerelease is worth adopting immediately if you run external agents or MCP servers over SSH remotes: environment variables, including secrets, were visible in the local process list (#65174). After you upgrade, rotate every key that was injected that way. The same build enables session notices and context compaction for ACP external agents (#65414), so long threads can stay open instead of being restarted to manage context. It also fixes prompts that were slow or never sent because of unneeded Git checkpoints (#65304) and first updates that were dropped in new threads (#65339). Update model routing now that the Google AI provider includes gemini-4-argon and Amazon Bedrock includes GPT-6 Astra, Claude Fable 5.1, and Claude Opus 5.5. In the editor, apply inlay hint edits with a double-click or editor: accept inlay hint, and rely on automatic unfolding when jumping to a location. Do not infer anything from the unpublished extra change.
Proposed actions
- Install Zed v1.25.0-pre from https://github.com/zed-industries/zed/releases/tag/v1.25.0-pre and fully restart so ACP session notices, context compaction, the remote MCP/agent env-var process-list fix (#65174), and faster external-agent prompts (#65304) are active.
- Rotate every API key and secret previously passed as an environment variable to external agents or MCP servers in SSH remote projects; those values were visible in the local process list before this build, then reconnect the remote project on v1.25.0-pre.
- In Zed Agent / Google AI settings, set the model id to gemini-4-argon. If you use Amazon Bedrock, add GPT-6 Astra, Claude Fable 5.1, and Claude Opus 5.5 to the Bedrock provider.
- Keep long-running external ACP agent threads open and use the new context compaction and session notices instead of starting a fresh thread; re-send a Gemini tool call that returns an image to confirm the post-image failure (#65576) is gone.
- From the command palette run editor: accept inlay hint with the cursor on a hint (or double-click the hint) to apply inlay hint text edits; use go to definition, outline, project symbols, multibuffer results, or navigation history and confirm folded regions auto-unfold at the jump target.
Agent prompt
Paste into your agent or query via MCP (get_agent_prompt) — free, no extra AI costAGENTS.md / CLAUDE.md / GEMINI.md rule update
DevAgentRadar → Repo Harness Rule Patch
Goal: update our repository's permanent harness instructions based on this release.
Update the file that tool actually reads. Do not dump everything into one file.
- Claude Code → CLAUDE.md plus
.claude/(skills, hooks, settings, agents, commands). It does not read AGENTS.md natively; start CLAUDE.md with@AGENTS.md. - Codex, Copilot, Cursor, Factory Droid, Grok Build, Roo Code, Goose, OpenCode, Amp, Zed, Aider → AGENTS.md.
- Gemini CLI / Antigravity → GEMINI.md. AGENTS.md only if
context.fileNameis set. - Cursor glob-scoped rules →
.cursor/rules/*.mdc(plain.mdis ignored), not a second constitution. - Codex MCP →
.codex/config.toml, not.mcp.json. - Copilot extra instructions →
.github/copilot-instructions.md; custom agents →.github/agents/. - Windsurf →
.windsurf/rules(do not assume AGENTS.md). - Cline →
.clinerules. - Procedures → a skill (
SKILL.md). Enforcement the model must not skip → a hook. Isolated roles → subagents. - Do not fork the same rule into three tool files.
Context
Assistant: Zed AI Proposal: Upgrade Zed to v1.25.0-pre and rotate leaked remote-agent MCP secrets Summary: Zed v1.25.0-pre adds ACP context compaction and session notices for external agents, Gemini 4 Argon plus new Bedrock models, and stops remote-project agent/MCP secrets from appearing in the local process list. Upgrade, rotate those secrets, and point agents at gemini-4-argon or the new Bedrock models. Primary source: https://github.com/zed-industries/zed/releases/tag/v1.25.0-pre
Why it matters
This prerelease is worth adopting immediately if you run external agents or MCP servers over SSH remotes: environment variables, including secrets, were visible in the local process list (#65174). After you upgrade, rotate every key that was injected that way. The same build enables session notices and context compaction for ACP external agents (#65414), so long threads can stay open instead of being restarted to manage context. It also fixes prompts that were slow or never sent because of unneeded Git checkpoints (#65304) and first updates that were dropped in new threads (#65339). Update model routing now that the Google AI provider includes gemini-4-argon and Amazon Bedrock includes GPT-6 Astra, Claude Fable 5.1, and Claude Opus 5.5. In the editor, apply inlay hint edits with a double-click or editor: accept inlay hint, and rely on automatic unfolding when jumping to a location. Do not infer anything from the unpublished extra change.
Suggested actions
-
Install Zed v1.25.0-pre from https://github.com/zed-industries/zed/releases/tag/v1.25.0-pre and fully restart so ACP session notices, context compaction, the remote MCP/agent env-var process-list fix (#65174), and faster external-agent prompts (#65304) are active.
-
Rotate every API key and secret previously passed as an environment variable to external agents or MCP servers in SSH remote projects; those values were visible in the local process list before this build, then reconnect the remote project on v1.25.0-pre.
-
In Zed Agent / Google AI settings, set the model id to gemini-4-argon. If you use Amazon Bedrock, add GPT-6 Astra, Claude Fable 5.1, and Claude Opus 5.5 to the Bedrock provider.
-
Keep long-running external ACP agent threads open and use the new context compaction and session notices instead of starting a fresh thread; re-send a Gemini tool call that returns an image to confirm the post-image failure (#65576) is gone.
-
From the command palette run editor: accept inlay hint with the cursor on a hint (or double-click the hint) to apply inlay hint text edits; use go to definition, outline, project symbols, multibuffer results, or navigation history and confirm folded regions auto-unfold at the jump target.
Config surfaces this release may change
- context and compaction settings — check your repo before applying
- MCP servers — check your repo before applying
- settings files — check your repo before applying
- background and headless runs — check your repo before applying
After you finish
Do not report this as applied to DevAgentRadar. You cannot write the visitor's loop.
Tell the human: open https://devagentradar.com/proposals/zed-ai-v1-25-0-pre-upgrade-zed-to-v1-25-0-pre-and-rotate-leaked-remote-a and mark Applied, Skipped, or Failed. Proposal id: 618c64aa-38a6-4ec6-8d2b-3bc4b62293f7
Instructions:
- Read the existing context files and settings for the tools in this repo.
- Draft an explicit Git diff. Update ONLY agent harness rules or tool configuration.
- Call out deprecated flags, obsolete habits, or changed permission boundaries.
- Do not touch application logic.
Start with the proposed diff and say which layer and which tool's file it belongs in.
Your loop
This browser · no sign-in · not shared as “you”After you run the prompt
Only you can mark this. Agents cannot write your loop.
Your decision stays on this device. A public tally appears after a few votes.