chore(claude): replace agent-framework with hivemind plugin config

.claude/hooks/caveman-ultra-subagent.sh

@@ -0,0 +1,10 @@
+#!/usr/bin/env bash
+
+cat <<'EOF'
+{
+  "hookSpecificOutput": {
+    "hookEventName": "SubagentStart",
+    "additionalContext": "Caveman mode requirement for this project: operate in caveman ultra mode for this entire subagent conversation. Do not silently fall back to full, lite, or normal verbosity unless the user explicitly requests it."
+  }
+}
+EOF

.claude/settings.json

@@ -1,7 +1,57 @@
 {
   "enabledPlugins": {
-    "agent-framework@brenpike": true,
-    "claude-mem@thedotmack": true
+    "claude-mem@thedotmack": true,
+    "hivemind@brenpike": true,
+    "caveman@caveman": true,
+    "codex@openai-codex": true
   },
-  "agent": "agent-framework:orchestrator"
+  "agent": "hivemind:overlord",
+  "pluginConfigs": {
+    "caveman@caveman": {
+      "options": {
+        "defaultLevel": "ultra"
+      }
+    }
+  },
+  "hooks": {
+    "SubagentStart": [
+      {
+        "hooks": [
+          {
+            "type": "command",
+            "command": "\"$CLAUDE_PROJECT_DIR\"/.claude/hooks/caveman-ultra-subagent.sh"
+          }
+        ]
+      }
+    ]
+  },
+  "permissions": {
+    "allow": [
+      "Bash(echo *)",
+      "Bash(printf *)",
+      "Bash(cat *)",
+      "Bash(grep *)",
+      "Bash(jq *)",
+      "Bash(head *)",
+      "Bash(tail *)",
+      "Bash(ls *)",
+      "Bash(wc *)",
+      "Bash(sort *)",
+      "Bash(uniq *)",
+      "Bash(git ls-files *)",
+      "Bash(git ls-tree *)",
+      "Bash(git grep *)",
+      "Bash(git tag)",
+      "Bash(git tag -l*)",
+      "Bash(git tag --list*)",
+      "Bash(git stash list)",
+      "Bash(git stash show *)",
+      "Edit(.hivemind/review-loop/*)",
+      "Edit(.hivemind/runs/.init-inputs-*.json)",
+      "Edit(.hivemind/runs/.record-inputs-*.json)",
+      "Edit(.hivemind/runs/.markfb-inputs-*.json)",
+      "Edit(.hivemind/spawn-inputs.*.json)",
+      "Edit(.hivemind/seed-inputs-*.json)"
+    ]
+  }
 }

.gitattributes

@@ -0,0 +1,3 @@
+# Shell scripts must keep LF endings so their shebang works after checkout
+# on machines with core.autocrlf enabled.
+*.sh text eol=lf

.gitignore

@@ -465,4 +465,6 @@ nupkgs/
 .claude/settings.local.json
 
 .claude/agent-memory/*
-.claude/agent-memory/
+.claude/agent-memory/
+.hivemind/
+.claude/worktrees/

CLAUDE.md

@@ -36,20 +36,21 @@ Chatter.Rest.Hal is a .NET/C# implementation of the HAL (Hypertext Application L
 
 ## Multi-Agent Governance
 
-This repository uses the [agent-framework](https://github.com/brenpike/agent-framework) Claude Code plugin.
+This repository uses the [hivemind](https://github.com/brenpike/hivemind) Claude Code plugin.
 
 Enable the plugin in `.claude/settings.json`:
 ```json
 {
   "enabledPlugins": {
-    "agent-framework@brenpike": true
-  }
+    "hivemind@brenpike": true
+  },
+  "agent": "hivemind:overlord"
 }
 ```
 
 The plugin provides:
-- Agents: `agent-framework:orchestrator` (default), `agent-framework:planner`, `agent-framework:coder`, `agent-framework:designer`
-- Skills: `agent-framework:create-working-branch`, `agent-framework:checkpoint-commit`, `agent-framework:open-plan-pr`, and others
+- Agents: `hivemind:overlord` (default), `hivemind:cerebrate`, `hivemind:drone`, `hivemind:changeling`, `hivemind:local-reviewer`, `hivemind:github-reviewer`
+- Skills: `hivemind:create-working-branch`, `hivemind:molt`, `hivemind:open-plan-pr`, `hivemind:route-workflow`, `hivemind:seed-hive`, and others
 - Governance reference docs in the plugin's `governance/` directory
 
 Project-specific adapter details live in `CLAUDE.md` (this file). External reviewer guidance lives in `AGENTS.md`.

docs/performance-todo.md

@@ -1,6 +1,6 @@
 # Performance Improvement TODO
 
-Static-analysis-identified performance improvements for the Chatter.Rest.Hal packages. Each item is self-contained: a planner or coder agent can implement it without re-reading the original analysis.
+Static-analysis-identified performance improvements for the Chatter.Rest.Hal packages. Each item is self-contained: an implementing agent can implement it without re-reading the original analysis.
 
 ## Priority Summary