Skip to content
Proposals/Add pre_agent hooks to block, rewrite, or inject
proposalteamP5Worth a lookMVMistral Vibe

Add pre_agent hooks to block, rewrite, or inject context before the model reads a message

Vibe 3.1.1 adds pre_agent, notification, session_start, and session_end hooks, /vibe:import, agentWorktrees with enter_worktree, and vim_keybindings. Plugin skills now use the plugin name (my-plugin:review), not the tool namespace. Wire hooks and rename skill selectors.

Why this loop

v3.1.1 adds four hook points around the agent loop. pre_agent runs before the model reads a user message and can block it, rewrite its text, or add context. session_start runs when a session starts, resumes, is cleared, or forked and can add context for the next message. notification runs in the background when Vibe needs approval or an answer, or when a turn completes or fails. session_end runs in the background when a session ends, with reason exit, clear, or other. Quitting waits up to 5 seconds for still-running notification and session_end hooks. Use these instead of prompt prefixes for policy and context. The same release adds /vibe:import, which finds setup from other coding agents and imports instructions, skills, subagents, commands, hooks, MCP servers, and permissions. App servers can advertise agentWorktrees: sessions get the enter_worktree tool, the agent asks whether to work in a new git worktree or in place before editing the checkout, and each worktree is listed in artifacts/list and cleaned up when the session is archived. Plugin skills, knowledge, and agents are now named after the plugin as written (my-plugin:review) instead of its tool namespace (my_plugin:review). Custom agents whose enabled_skills or disabled_skills still use my_plugin:* will not match until they switch to my-plugin:*. vibe-acp notifies the client when an administrator changes the organization's Vibe Code configuration.

Proposed actions

  1. Add a pre_agent hook that runs before the model reads a user message and can block it, rewrite its text, or add context for the model.
  2. Add a session_start hook that runs when a session starts, resumes, is cleared, or forked and injects context into the next message, plus a background session_end hook that receives the reason (exit, clear, or other).
  3. Add a background notification hook that fires when Vibe needs your approval or an answer, or when a turn completes or fails; quitting waits up to 5 seconds for it.
  4. In the Vibe prompt, run /vibe:import to find your setup from other coding agents and import instructions, skills, subagents, commands, hooks, MCP servers, and permissions.
  5. In every custom agent, change enabled_skills and disabled_skills from the tool namespace (my_plugin:*) to the plugin name as written (my-plugin:*); skills are now my-plugin:review, not my_plugin:review.

Agent prompt

Paste into your agent or query via MCP (get_agent_prompt) — free, no extra AI cost

AGENTS.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.fileName is set.
  • Cursor glob-scoped rules → .cursor/rules/*.mdc (plain .md is 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: Mistral Vibe Proposal: Add pre_agent hooks to block, rewrite, or inject context before the model reads a message Summary: Vibe 3.1.1 adds pre_agent, notification, session_start, and session_end hooks, /vibe:import, agentWorktrees with enter_worktree, and vim_keybindings. Plugin skills now use the plugin name (my-plugin:review), not the tool namespace. Wire hooks and rename skill selectors. Primary source: https://github.com/mistralai/mistral-vibe/releases/tag/v3.1.1

Why it matters

v3.1.1 adds four hook points around the agent loop. pre_agent runs before the model reads a user message and can block it, rewrite its text, or add context. session_start runs when a session starts, resumes, is cleared, or forked and can add context for the next message. notification runs in the background when Vibe needs approval or an answer, or when a turn completes or fails. session_end runs in the background when a session ends, with reason exit, clear, or other. Quitting waits up to 5 seconds for still-running notification and session_end hooks. Use these instead of prompt prefixes for policy and context. The same release adds /vibe:import, which finds setup from other coding agents and imports instructions, skills, subagents, commands, hooks, MCP servers, and permissions. App servers can advertise agentWorktrees: sessions get the enter_worktree tool, the agent asks whether to work in a new git worktree or in place before editing the checkout, and each worktree is listed in artifacts/list and cleaned up when the session is archived. Plugin skills, knowledge, and agents are now named after the plugin as written (my-plugin:review) instead of its tool namespace (my_plugin:review). Custom agents whose enabled_skills or disabled_skills still use my_plugin:* will not match until they switch to my-plugin:*. vibe-acp notifies the client when an administrator changes the organization's Vibe Code configuration.

Suggested actions

  1. Add a pre_agent hook that runs before the model reads a user message and can block it, rewrite its text, or add context for the model.
  2. Add a session_start hook that runs when a session starts, resumes, is cleared, or forked and injects context into the next message, plus a background session_end hook that receives the reason (exit, clear, or other).
  3. Add a background notification hook that fires when Vibe needs your approval or an answer, or when a turn completes or fails; quitting waits up to 5 seconds for it.
  4. In the Vibe prompt, run /vibe:import to find your setup from other coding agents and import instructions, skills, subagents, commands, hooks, MCP servers, and permissions.
  5. In every custom agent, change enabled_skills and disabled_skills from the tool namespace (my_plugin:*) to the plugin name as written (my-plugin:*); skills are now my-plugin:review, not my_plugin:review.

Config surfaces this release may change

  • skills (high confidence) — check your repo before applying
  • context and compaction settings (high confidence) — check your repo before applying
  • hooks — check your repo before applying
  • permission rules — check your repo before applying
  • MCP servers — check your repo before applying
  • subagent definitions — 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/mistral-vibe-v3-1-1-add-pre-agent-hooks-to-block-rewrite-or-inject-conte and mark Applied, Skipped, or Failed. Proposal id: 8876757d-9406-4d2a-8505-de75489de554

Instructions:

  1. Read the existing context files and settings for the tools in this repo.
  2. Draft an explicit Git diff. Update ONLY agent harness rules or tool configuration.
  3. Call out deprecated flags, obsolete habits, or changed permission boundaries.
  4. Do not touch application logic.

Start with the proposed diff and say which layer and which tool's file it belongs in.

agentmcpmodelsecurityideRelease source ↗

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.

Originating release signal

MVMistral Vibev3.1.1Nov 9, 2026

v3.1.1

vibe-acp notifies the client when the administrator changes the organization's Vibe Code configuration. · Vim Normal and Insert modes in the prompt input, enabled with the vim_keybindings config option · The prompt of the conversation section at the top of the screen stays pinned while you scroll, and a click jumps back to it · enable_sticky_prompt = false turns the pinned prompt off · pre_agent hooks run before the model reads a user message: they can block it, rewrite its text, or add context for the model. · +88 more changes
Verified excerpt — the source's own words

Added

  • vibe-acp notifies the client when the administrator changes the organization's Vibe Code configuration.
  • Vim Normal and Insert modes in the prompt input, enabled with the vim_keybindings config option
  • The prompt of the conversation section at the top of the screen stays pinned while you scroll, and a click jumps back to it
  • enable_sticky_prompt = false turns the pinned prompt off
  • pre_agent hooks run before the model reads a user message: they can block it, rewrite its text, or add context for the model.
  • notification hooks run in the background when Vibe needs your approval or an answer, or when a turn completes or fails.
  • session_start hooks run when a session starts, resumes, is cleared or forked, and can add context for the model to the next message.
  • session_end hooks run in the background when a session ends, with the reason (exit, clear or other).
  • Built-in /vibe:import skill that finds your setup from other coding agents and imports instructions, skills, subagents, commands, hooks, MCP servers and permissions into Vibe
  • vibe -p reports each denied tool call on stderr and in the export's warnings

Excerpt ends here — this release continues at the source ↗.

Primary source ↗