@polderlabs/bizar 6.2.5 → 7.0.0

package/config/AGENTS.md

@@ -54,7 +54,7 @@ which skills 2>/dev/null
 skills list --json
 
 # View installed skill files
-ls ~/.cline/skills/<skill-name>/SKILL.md
+ls ~/.claude/skills/<skill-name>/SKILL.md
 ```
 
 ### How to Install
@@ -72,7 +72,7 @@ skills add <owner/repo> -s "<skill-name>" -y
 1. **Assess**: When given a task, consider whether a skill might exist for it
 2. **Check installed**: Run `skills list --json` to see what's already available
 3. **Try known repos**: Based on the task domain, attempt installation from known skill repos (e.g., `skills add supabase/agent-skills --all -y` for database work, `skills add vercel-labs/agent-skills --all -y` for frontend work)
-4. **Use**: After installing, the skill instructions are at `~/.cline/skills/<skill-name>/SKILL.md` — load them with the `skill` tool
+4. **Use**: After installing, the skill instructions are at `~/.claude/skills/<skill-name>/SKILL.md` — load them with the `skill` tool
 5. **Skip**: If no skill is found after trying likely repos, proceed without
 
 ### Known Skill Repositories by Domain
@@ -146,26 +146,26 @@ Odin (`@odin`) is the All-Father and primary/default agent. He analyzes each req
 
 ### Frigg
 
-- **Model**: `cline/deepseek-v4-flash-free` (via Cline Zen — free tier)
+- **Model**: `claude-3-5-haiku` (via Claude Code — Anthropic API, free tier)
 - **Use for**: Read-only codebase Q&A. Use `@frigg` to ask questions about the project and get answers with file references — never modifies anything.
 - **Mode**: Primary (directly selectable by the user via `@frigg`)
 - **Cost**: Free
 
 ### Vör
 
-- **Model**: `cline/deepseek-v4-flash-free` (via Cline Zen — free tier)
+- **Model**: `claude-3-5-haiku` (via Claude Code — Anthropic API, free tier)
 - **Use for**: Clarifying ambiguous or incomplete requests. First reads project context (`.bizar/PROJECT.md`, obsidian, project files), then only asks targeted, project-specific questions if still unclear. Never asks generic questions.
 - **Cost**: Free
 
 ### Heimdall
 
-- **Model**: `cline/deepseek-v4-flash-free` (via Cline Zen — free tier)
+- **Model**: `claude-3-5-haiku` (via Claude Code — Anthropic API, free tier)
 - **Use for**: Simple tasks, mechanical work, quick edits, file operations. The ever-watchful guardian.
 - **Cost**: Free
 
 ### Mimir
 
-- **Model**: `cline/deepseek-v4-flash-free` (via Cline Zen — free tier)
+- **Model**: `claude-3-5-haiku` (via Claude Code — Anthropic API, free tier)
 - **Use for**: Deep codebase research, exploration, documentation analysis. Semble-first search approach.
 - **Cost**: Free
 
@@ -207,7 +207,7 @@ Odin (`@odin`) is the All-Father and primary/default agent. He analyzes each req
 
 ### Routing Heuristic
 
-Odin dispatches all tasks to subagents via the `task` tool. When work items are **independent**, he launches them as **parallel `task` calls in a single message**.
+Odin dispatches all tasks to subagents via the Claude Code `Agent` tool. When work items are **independent**, he launches them as **parallel `Agent` calls in a single message** with `subagent_type` set to the target agent's name.
 
 **Before dispatching any task, Odin runs `bizar memory status` to confirm the vault is reachable.** No `bank_id` is needed — `bizar memory status` resolves everything from the project root.
 
@@ -285,25 +285,25 @@ From the project root, the user (or heimdall via `/init` or any other agent prom
 
 ## General Agent Baseline — Always-On Behavior
 
-This section is the single source of truth for every Bizar agent's behavior. It is **adapted from the upstream system prompt and translated to Bizar**. Every Claude-specific reference has been mapped to the Bizar equivalent (BizarHarness, cline, obsidian, Semble, Skills CLI, agent-browser, the cline tool set). All agents **MUST** follow these rules at all times.
+This section is the single source of truth for every Bizar agent's behavior. It is **adapted from the upstream system prompt and translated to Bizar**. Every Claude-specific reference has been mapped to the Bizar equivalent (BizarHarness, Claude Code, obsidian, Semble, Skills CLI, agent-browser, Claude Code's native tool set). All agents **MUST** follow these rules at all times.
 
-> **Tool name translation table** (used throughout this baseline):
+> **Tool name translation table** (used throughout this baseline): this section maps Claude.ai tool names (e.g. `view`) to Claude Code tool names (e.g. `Read`). Starting with v6.3.0 (DEC-011), Claude Code's native tools are the source of truth — the harness no longer maintains its own tool set. The Bizar column below is the Claude Code column when they match.
 >
-> | Upstream (Claude.ai) | Bizar equivalent |
+> | Upstream (Claude.ai) | Claude Code |
 > |---|---|
-> | `view` | `read` |
-> | `str_replace` | `edit` |
-> | `create_file` | `write` |
-> | `bash_tool` | `bash` |
-> | `web_search` | `websearch` (cline built-in) |
-> | `web_fetch` | `webfetch` (cline built-in) |
+> | `view` | `Read` |
+> | `str_replace` | `Edit` |
+> | `create_file` | `Write` |
+> | `bash_tool` | `Bash` |
+> | `web_search` | `WebSearch` |
+> | `web_fetch` | `WebFetch` |
 > | `present_files` | not applicable — Bizar delivers files via the dashboard (`@polderlabs/bizar-dash/src/server/routes/artifacts.mjs`) or by writing to the workspace |
-> | `image_search` / `places_*` / `weather_fetch` / `recipe_display_v0` / `fetch_sports_data` / `message_compose_v1` / `recommend_claude_apps` | not available in Bizar — do not assume these exist |
+> | `image_search` / `places_*` / `weather_fetch` / `recipe_display_v0` / `fetch_sports_data` / `message_compose_v1` / `recommend_claude_apps` | not available in Claude Code — do not assume these exist |
 > | `search_mcp_registry` / `suggest_connectors` | use the `skills` CLI (`skills add <owner/repo> -s <name>`) to discover and install skills instead |
-> | `ask_user_input_v0` | Bizar has a `question` tool — same shape, single high-value question |
-> | `skill` | `skill` — load a SKILL.md from `~/.cline/skills/<name>/` or installed equivalent |
-> | `task` (subagent dispatch) | `task` — same — used by Odin to dispatch subagents |
-> | MCP servers | `semble` (codebase search) and any user-added servers in `config/cline.json` |
+> | `ask_user_input_v0` | `AskUserQuestion` — single high-value question |
+> | `skill` | `Skill` — load a SKILL.md from `~/.claude/skills/<name>/` or `.claude/skills/<name>/` |
+> | `task` (subagent dispatch) | `Agent` (with `subagent_type`, `run_in_background`, `isolation: "worktree"`) — used by Odin to dispatch subagents |
+> | MCP servers | `semble` (codebase search) and any user-added servers in `.claude/mcp.json` |
 
 ### Simplicity Rule — do not overcomplicate
 
@@ -321,7 +321,7 @@ This section is the single source of truth for every Bizar agent's behavior. It
 
 ### Identity preamble
 
-- Bizar is a Norse-pantheon multi-agent system for cline. Odin is the default primary agent; Frigg, Vör, Mimir, Heimdall, Hermod, Thor, Baldr, Tyr, Vidarr, and Forseti are the subagents.
+- Bizar is a Norse-pantheon multi-agent system for Claude Code. Odin is the default primary agent; Frigg, Vör, Mimir, Heimdall, Hermod, Thor, Baldr, Tyr, Vidarr, and Forseti are the subagents.
 - The agent does not have a fixed identity outside its role definition. Do not claim to be Claude, Anthropic, or any other AI.
 - Treat the user as a capable adult working on engineering work unless the context clearly indicates otherwise.
 
@@ -388,7 +388,7 @@ This section is the single source of truth for every Bizar agent's behavior. It
 - For facts that change quickly (current positions, prices, breaking news) or anything that could have changed recently, **search before answering**: use `websearch` and `webfetch` or delegate to `@mimir` for deep research.
 - For stable technical knowledge (language semantics, well-established APIs, mathematical truths), answer directly without search.
 - Default to running `bizar memory search "<topic>"` at session start to retrieve prior project context before answering anything project-specific.
-- When formulating date-sensitive queries, use the actual current date (Bizar's cline environment provides this). Do not hardcode years.
+- When formulating date-sensitive queries, use the actual current date (Bizar's Claude Code environment provides this). Do not hardcode years.
 - Do not over-rely on memory; if uncertain, search. Confabulating costs the user more than searching.
 
 ### mcp_servers_and_skills
@@ -407,7 +407,7 @@ The `skills` CLI (`npm install -g skills`) installs skill packs from skills.sh.
 ```bash
 which skills 2>/dev/null
 skills list --json
-ls ~/.cline/skills/<skill-name>/SKILL.md
+ls ~/.claude/skills/<skill-name>/SKILL.md
 ```
 
 If not installed but relevant, install it from a known repo by domain:
@@ -424,7 +424,7 @@ Load the SKILL.md via the `skill` tool before writing code or making changes cov
 
 #### Browser interaction
 
-For browser-driven E2E validation, use **agent-browser** (the Python tool from https://github.com/browser-use/agent-browser). Invoke via `bash` heredoc:
+For browser-driven E2E validation, use **agent-browser** (the Rust-native tool from https://github.com/vercel-labs/agent-browser; the earlier Python build at https://github.com/browser-use/agent-browser is archived as of the v6.0.0 migration). Invoke via `bash` heredoc:
 
 ```bash
 agent-browser <<'PY'
@@ -434,7 +434,7 @@ print(page_info())
 PY
 ```
 
-Common operations: `new_tab`, `goto_url`, `wait_for_load`, `page_info`, `click_at_xy`, `fill_input`, `press_key`, `scroll`, `capture_screenshot`, `js`, `cdp("Domain.method", ...)`. Read the SKILL.md at `~/.cline/skills/agent-browser/SKILL.md` on first use. Do **not** install headless Chrome via raw shell commands when agent-browser is available.
+Common operations: `new_tab`, `goto_url`, `wait_for_load`, `page_info`, `click_at_xy`, `fill_input`, `press_key`, `scroll`, `capture_screenshot`, `js`, `cdp("Domain.method", ...)`. Read the SKILL.md at `~/.claude/skills/agent-browser/SKILL.md` on first use. Do **not** install headless Chrome via raw shell commands when agent-browser is available.
 
 ### skills_mandatory_read
 
@@ -445,8 +445,8 @@ Concrete triggers:
 - Backend/API work → framework-specific skill
 - Browser E2E → `agent-browser` SKILL.md
 - Skill creation → `skill-creator` SKILL.md
-- BizarHarness-specific work → `~/.cline/skills/bizar/` SKILL.md (always)
-- Self-improvement logging → `~/.cline/skills/self-improvement/` SKILL.md (always)
+- BizarHarness-specific work → `~/.claude/skills/bizar/` SKILL.md (always)
+- Self-improvement logging → `~/.claude/skills/self-improvement/` SKILL.md (always)
 
 ### file_creation_advice
 

package/config/trigger-patterns.json

@@ -0,0 +1,186 @@
+{
+  "$schema": "./trigger-patterns.v1.schema.json",
+  "version": 1,
+  "description": "Trigger-pattern table for the Bizar Background Workers dispatcher. Each worker maps zero-or-more regex patterns to a Bizar skill and/or agent. The dispatcher in cli/worker-dispatcher.mjs reads this file, matches user prompts against the regex arrays (case-insensitive), and emits 0..N suggestions ranked by weight.",
+  "workers": [
+    {
+      "id": "testgaps",
+      "regex": [
+        "missing tests",
+        "low coverage",
+        "no tests for",
+        "tests? for .+ that",
+        "test gaps?",
+        "test coverage",
+        "coverage report",
+        "add tests"
+      ],
+      "weight": 0.8,
+      "skill": "harness-engineering",
+      "agent": "forseti",
+      "description": "Identify test coverage gaps and untested code paths."
+    },
+    {
+      "id": "audit",
+      "regex": [
+        "security scan",
+        "security audit",
+        "vulnerab",
+        "cve",
+        "owasp",
+        "pentest"
+      ],
+      "weight": 0.9,
+      "skill": "harness-engineering",
+      "agent": "forseti",
+      "description": "Security analysis and vulnerability scanning."
+    },
+    {
+      "id": "deepdive",
+      "regex": [
+        "deep dive into",
+        "deep dive",
+        "explain .{20,}",
+        "how does .+ work",
+        "walk me through",
+        "in-depth analysis",
+        "comprehensive review"
+      ],
+      "weight": 0.7,
+      "skill": null,
+      "agent": "mimir",
+      "description": "Deep code analysis and thorough explanation."
+    },
+    {
+      "id": "refactor",
+      "regex": [
+        "refactor",
+        "clean up .{1,30}",
+        "simplify",
+        "restructure",
+        "improve code quality",
+        "make .+ more readable"
+      ],
+      "weight": 0.7,
+      "skill": null,
+      "agent": "thor",
+      "description": "Code refactoring suggestions and structural cleanup."
+    },
+    {
+      "id": "document",
+      "regex": [
+        "write docs",
+        "write readme",
+        "add jsdoc",
+        "document .{1,40}",
+        "api docs",
+        "generate docs",
+        "add documentation"
+      ],
+      "weight": 0.6,
+      "skill": null,
+      "agent": "baldr",
+      "description": "Auto-documentation generation (README, JSDoc, API docs)."
+    },
+    {
+      "id": "optimize",
+      "regex": [
+        "optimize .{1,40} for",
+        "performance issue",
+        "make .+ faster",
+        "speed up",
+        "reduce .{1,20}(memory|time)",
+        "improve performance"
+      ],
+      "weight": 0.7,
+      "skill": null,
+      "agent": "thor",
+      "description": "Performance optimization and tuning."
+    },
+    {
+      "id": "ultralearn",
+      "regex": [
+        "learn about",
+        "understand how",
+        "understand what",
+        "understand why",
+        "comprehensive guide",
+        "master .+"
+      ],
+      "weight": 0.65,
+      "skill": "harness-engineering",
+      "agent": "mimir",
+      "description": "Deep knowledge acquisition and learning."
+    },
+    {
+      "id": "consolidate",
+      "regex": [
+        "consolidate",
+        "merge memories",
+        "clean up memory",
+        "deduplicate",
+        "memory maintenance"
+      ],
+      "weight": 0.5,
+      "skill": "harness-engineering",
+      "agent": null,
+      "description": "Memory consolidation and cleanup."
+    },
+    {
+      "id": "predict",
+      "regex": [
+        "what will happen",
+        "predict",
+        "forecast",
+        "anticipate",
+        "prepare for"
+      ],
+      "weight": 0.55,
+      "skill": null,
+      "agent": "mimir",
+      "description": "Predictive preloading and anticipation."
+    },
+    {
+      "id": "map",
+      "regex": [
+        "map .{0,5}codebase",
+        "architecture overview",
+        "project structure",
+        "dependency graph",
+        "code map",
+        "explore codebase"
+      ],
+      "weight": 0.6,
+      "skill": "harness-engineering",
+      "agent": "mimir",
+      "description": "Codebase mapping and architecture analysis."
+    },
+    {
+      "id": "preload",
+      "regex": [
+        "preload",
+        "cache ahead",
+        "prefetch",
+        "warm .+ cache"
+      ],
+      "weight": 0.4,
+      "skill": null,
+      "agent": null,
+      "description": "Resource preloading and cache warming."
+    },
+    {
+      "id": "benchmark",
+      "regex": [
+        "benchmark",
+        "performance test",
+        "measure speed",
+        "stress test",
+        "load test"
+      ],
+      "weight": 0.65,
+      "skill": null,
+      "agent": "thor",
+      "description": "Performance benchmarking."
+    }
+  ]
+}

package/install.sh

@@ -1,26 +1,23 @@
 #!/usr/bin/env bash
 #
-# install.sh — Cross-platform BizarHarness installer (v5.x — thin shell wrapper).
+# install.sh — Cross-platform BizarHarness installer (Claude Code).
 #
-# v5.x — Updated for issue #7. The install/update flow now:
-#   1. On first install: registers the system service that auto-starts
-#      the dashboard (`bizar service install`).
-#   2. On update: stops the running service before files are replaced
-#      (`bizar service stop` → npm upgrade → `bizar service restart`).
+# v6.3.0 — Migrated from Cline to Claude Code.
 #
-# Almost all logic still lives in `cli/provision.mjs:runProvision`, which
-# is shared between `bizar install` and `bizar update`. This bash script
-# exists only for the platform-specific steps that need sudo + a system
-# package manager:
+#   1. Detects platform (Linux/macOS/Windows)
+#   2. Installs Node.js 18+, jq, gh, uv, python3 (cross-platform)
+#   3. Installs Claude Code via npm:
+#        npm install -g @anthropic-ai/claude-agent-sdk
+#        npm install -g @anthropic-ai/claude-code
+#   4. Optionally installs agent-browser (Rust native CLI from vercel-labs)
+#   5. Shells out to `node cli/provision-claude.mjs`, which writes the
+#      `~/.claude/` config tree, registers the Bizar MCP server, and
+#      copies agent files / skills / commands / hooks / rules.
 #
-#   - Linux:  ensure node is installed (apt/dnf/pacman/zypper), install uv,
-#             python3.12, jq, gh, agent-browser (npm), Chrome for Testing.
-#   - macOS:  ensure homebrew is installed; everything else is via brew/npm.
-#   - Windows: a stub that prints "use install.ps1".
-#
-# Agent files / plugin copy / cline.json patching / service registration /
-# skills install / doctor check are ALL handled by the unified provisioner
-# after this script returns.
+# Almost all logic still lives in `cli/provision-claude.mjs:runProvision`,
+# which is shared between `bizar install` and `bizar update`. This bash
+# script exists only for the platform-specific steps that need sudo +
+# a system package manager.
 #
 # Flags:
 #   --non-interactive   skip prompts (CI safe)
@@ -72,7 +69,7 @@ MODE="install"
 
 show_usage() {
   cat <<'EOF'
-install.sh — BizarHarness cross-platform installer
+install.sh — BizarHarness cross-platform installer (Claude Code)
 
 Usage:
   ./install.sh                     Interactive install (Linux/macOS)
@@ -84,9 +81,10 @@ Usage:
   ./install.sh --mode=<mode>       install | update | install-only-system
 
 This script handles platform-specific system dependencies (apt/dnf/pacman/
-zypper on Linux, brew on macOS) and then shells to `node cli/provision.mjs`,
-which performs agent-file sync, plugin copy, cline.json patching,
-service registration, skills install, and the post-install doctor check.
+zypper on Linux, brew on macOS) and then shells to
+`node cli/provision-claude.mjs`, which writes ~/.claude/, registers the
+Bizar MCP server, copies agents / skills / commands / hooks / rules,
+and runs the post-install doctor check.
 EOF
 }
 
@@ -159,7 +157,7 @@ ensure_node() {
       action "Detected NixOS — Node.js must be installed via nix-shell"
       warn "Run the following, then re-run this installer:"
       warn "  nix-shell -p nodejs python3 jq gh git"
-      warn "  node cli/provision.mjs"
+      warn "  node cli/provision-claude.mjs"
       exit 0
       ;;
     void)
@@ -175,6 +173,29 @@ ensure_node() {
   note "Node.js installed: $(node --version)"
 }
 
+# ── Claude Code install ────────────────────────────────────────────────────
+install_claude_code() {
+  if have_cmd claude; then
+    note "claude $(claude --version 2>/dev/null | head -1) present"
+  else
+    section "Installing Claude Code"
+    if ! have_cmd npm; then
+      err "npm not available — install Node.js 18+ first: https://nodejs.org/"
+      exit 1
+    fi
+    action "Installing @anthropic-ai/claude-agent-sdk..."
+    dry npm install -g @anthropic-ai/claude-agent-sdk 2>&1 || \
+      warn "claude-agent-sdk install failed — retry later: npm install -g @anthropic-ai/claude-agent-sdk"
+    action "Installing @anthropic-ai/claude-code..."
+    dry npm install -g @anthropic-ai/claude-code 2>&1
+    if have_cmd claude; then
+      note "claude $(claude --version 2>/dev/null | head -1) installed"
+    else
+      warn "claude CLI not found on PATH after install — check npm global bin dir"
+    fi
+  fi
+}
+
 # ── Dependency detection + install (cross-platform) ────────────────────────
 check_deps() {
   section "Checking dependencies"
@@ -194,7 +215,7 @@ install_agent_browser() {
   # the v5.x browser-harness (Python CDP wrapper). Installs via npm.
   #
   # If npm is not available, this is a soft-fail: agent-browser is
-  # optional, and `cli/provision.mjs:ensureAgentBrowser` will retry it
+  # optional, and `cli/provision-claude.mjs:ensureAgentBrowser` will retry it
   # once npm is available.
   if have_cmd agent-browser; then
     local ab_ver
@@ -305,7 +326,7 @@ install_missing_deps_linux() {
       warn "NixOS detected — LightRAG and other uv tools require manual setup"
       warn "  After nix-shell: nix-shell -p nodejs python3 jq gh git"
       warn "  Then run: uv tool install \"lightrag-hku[api]\""
-      warn "  And re-run this installer: node cli/provision.mjs"
+      warn "  And re-run this installer: node cli/provision-claude.mjs"
       return 0
       ;;
   esac
@@ -337,15 +358,13 @@ install_service() {
   # We still call it from here because the bash script may run standalone
   # (e.g. for first-boot provisioning before the npm package is set up).
   #
-  # v5.x — issue #7: On first install, register the service so the
-  # dashboard auto-starts at login. On update, the running service is
-  # stopped by the provisioner before files are replaced, then restarted
-  # by the provisioner after — see cli/provision.mjs. This function
-  # covers the FIRST install path only.
+  # v6.3.0 — Claude Code does NOT have an out-of-process "server" the way
+  # Cline did, so we skip `bizar service install` here. The dashboard
+  # (bizar-dash) is launched via `bizar dash start --bg` instead.
   local bin="$REPO_DIR/cli/bin.mjs"
   if [ ! -f "$bin" ]; then
     warn "cli/bin.mjs not found — service registration deferred"
-    warn "    (will complete when `bizar service install` is run later)"
+    warn "    (will complete when \`bizar service install\` is run later)"
     return 0
   fi
   if [ "$DRY_RUN" -eq 1 ]; then
@@ -369,15 +388,15 @@ print_banner() {
   section "Install complete"
   echo ""
   echo -e "${BOLD}${CYAN}┌────────────────────────────────────────────────────────────┐${NC}"
-  echo -e "${BOLD}${CYAN}│${NC}  ${BOLD}BizarHarness ready.${NC}                                             │"
+  echo -e "${BOLD}${CYAN}│${NC}  ${BOLD}BizarHarness ready (Claude Code).${NC}                          │"
   echo -e "${BOLD}${CYAN}│${NC}                                                          │"
-  echo -e "${BOLD}${CYAN}│${NC}  ${GREEN}✓${NC} Cross-platform installer v4.4.7                             │"
+  echo -e "${BOLD}${CYAN}│${NC}  ${GREEN}✓${NC} Cross-platform installer v6.3.0                             │"
   echo -e "${BOLD}${CYAN}│${NC}                                                          │"
   echo -e "${BOLD}${CYAN}│${NC}  Dashboard: ${CYAN}$dashboard_url${NC}                                    │"
   echo -e "${BOLD}${CYAN}│${NC}                                                          │"
   echo -e "${BOLD}${CYAN}│${NC}  ${DIM}Next:${NC}                                                            │"
-  echo -e "${BOLD}${CYAN}│${NC}  ${DIM}  1. Restart cline to pick up new config${NC}                     │"
-  echo -e "${BOLD}${CYAN}│${NC}  ${DIM}  2. Run /connect in cline to add API keys${NC}                   │"
+  echo -e "${BOLD}${CYAN}│${NC}  ${DIM}  1. Restart claude to pick up new config${NC}                   │"
+  echo -e "${BOLD}${CYAN}│${NC}  ${DIM}  2. Run \`claude auth login\` to add API keys${NC}                 │"
   echo -e "${BOLD}${CYAN}│${NC}  ${DIM}  3. Run 'bizar dash start' to launch the dashboard${NC}             │"
   echo -e "${BOLD}${CYAN}│${NC}  ${DIM}  4. Visit $dashboard_url in your browser${NC}                │"
   echo -e "${BOLD}${CYAN}└────────────────────────────────────────────────────────────┘${NC}"
@@ -393,6 +412,7 @@ install_linux() {
   note "Detected Linux ($(uname -m))"
   ensure_node
   check_deps
+  install_claude_code
   install_missing_deps_linux
   install_agent_browser
   install_service
@@ -402,6 +422,7 @@ install_macos() {
   note "Detected macOS ($(uname -m))"
   check_deps
   install_missing_deps_macos
+  install_claude_code
   install_agent_browser
   install_service
 }
@@ -417,7 +438,7 @@ main() {
   fi
 
   echo ""
-  echo -e "${BOLD}${CYAN}  ⚡ BizarHarness Installer v6.0.0${NC}"
+  echo -e "${BOLD}${CYAN}  ⚡ BizarHarness Installer v6.3.0 (Claude Code)${NC}"
   if [ "$UPDATE_MODE" -eq 1 ]; then
     echo -e "  ${DIM}Update mode${NC}"
   fi
@@ -441,16 +462,17 @@ main() {
   esac
 
   # ── Hand off to the unified provisioner (Node) ─────────────────────────
-  # All cross-OS work (agent files, plugin copy, cline.json patching,
-  # skills install, doctor check) lives in `cli/provision.mjs:runProvision`.
-  # It's idempotent — skipping it (e.g. via --install-only-system) is fine
-  # for first-boot scenarios where the npm package isn't yet set up.
+  # All cross-OS work (agent files, command copy, ~/.claude/ settings,
+  # MCP server registration, skills/rules/hooks mirror) lives in
+  # `cli/provision-claude.mjs:runProvision`. It's idempotent — skipping
+  # it (e.g. via --install-only-system) is fine for first-boot scenarios
+  # where the npm package isn't yet set up.
   if [ "$MODE" = "install-only-system" ]; then
     note "skipped node provisioner (--mode=install-only-system)"
     exit 0
   fi
 
-  if [ -f "$REPO_DIR/cli/provision.mjs" ]; then
+  if [ -f "$REPO_DIR/cli/provision-claude.mjs" ]; then
     echo ""
     section "Running unified provisioner"
     local args=(--mode "$MODE")
@@ -459,7 +481,7 @@ main() {
     [ "$NON_INTERACTIVE" -eq 1 ] && args+=(--yes)
     # `set +e` so we capture the provisioner's exit code and surface it.
     set +e
-    node "$REPO_DIR/cli/provision.mjs" "${args[@]}"
+    node "$REPO_DIR/cli/provision-claude.mjs" "${args[@]}"
     local rc=$?
     set -e
     if [ "$rc" -ne 0 ]; then
@@ -468,9 +490,9 @@ main() {
     fi
     print_banner
   else
-    warn "cli/provision.mjs not found — agent files / plugin / cline.json"
+    warn "cli/provision-claude.mjs not found — agent files / MCP / settings.json"
     warn "    were NOT synced. Run \`bizar install\` from a checkout to fix."
   fi
 }
 
-main "$@"
+main "$@"

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@polderlabs/bizar",
-  "version": "6.2.5",
-  "description": "Norse-pantheon multi-agent system for cline — 13 agents across 4 cost tiers with cost-aware routing, plans, and a configurable agent harness. v4 ships as a single npm package bundling the dashboard server, cline plugin, and typed SDK.",
+  "version": "7.0.0",
+  "description": "Norse-pantheon multi-agent system for Claude Code — 14 agents across 4 cost tiers with cost-aware routing, plans, and a configurable agent harness. Ships as a single npm package with the dashboard server, Claude Code MCP server, and typed SDK.",
   "type": "module",
   "bin": {
     "bizar": "cli/bin.mjs"
@@ -9,13 +9,12 @@
   "files": [
     "cli/",
     "bizar-dash/",
-    "plugins/",
     "packages/sdk/",
+    ".claude/",
     "config/",
     "templates/",
     "scripts/",
-    "install.sh",
-    "cli/browser-harness-up.sh"
+    "install.sh"
   ],
   "scripts": {
     "typecheck": "tsc --noEmit",
@@ -24,12 +23,13 @@
     "test:sdk": "node_modules/.bin/vitest run --root packages/sdk",
     "test:web": "cd bizar-dash && npx vitest run",
     "test:sdk:watch": "node_modules/.bin/vitest --root packages/sdk",
-    "test": "npm run typecheck && npm run test:sdk && node --test cli/install.test.mjs cli/provision.test.mjs cli/commands/validate.test.mjs cli/commands/setup-provider.test.mjs cli/commands/rca.test.mjs && bun test plugins/bizar/tests/loop.test.ts plugins/bizar/tests/block.test.ts plugins/bizar/tests/stall-think.test.ts plugins/bizar/tests/tools/bg-get-comments.test.ts plugins/bizar/tests/tools/bg-spawn-delegation.test.ts plugins/bizar/tests/tools/cline-runner.test.ts plugins/bizar/tests/settings.test.ts plugins/bizar/tests/commands.test.ts plugins/bizar/tests/commands-impl.test.ts plugins/bizar/tests/tools/plan-action.test.ts plugins/bizar/tests/tools/wait-for-feedback.test.ts plugins/bizar/tests/tools/read-glyph-feedback.test.ts plugins/bizar/tests/reasoning-clean.test.ts plugins/bizar/tests/key-rotation.test.ts && node bizar-dash/tests/smoke-v2.mjs && node --test bizar-dash/tests/path-safe.test.mjs bizar-dash/tests/tmux-wrap.test.mjs bizar-dash/tests/cline-sessions-detail.test.mjs bizar-dash/tests/cline-runner.test.mjs bizar-dash/tests/mod-instructions.node.test.mjs bizar-dash/tests/mod-upgrade.node.test.mjs bizar-dash/tests/graphify-mod-spawn.node.test.mjs bizar-dash/tests/no-agent-browser.node.test.mjs bizar-dash/tests/providers-store-backup-keys.node.test.mjs bizar-dash/tests/dashboard-ports.test.mjs bizar-dash/tests/submit-feedback.test.mjs bizar-dash/tests/yaml.test.mjs bizar-dash/tests/memory-store.test.mjs bizar-dash/tests/memory-schema.test.mjs bizar-dash/tests/memory-secrets.test.mjs bizar-dash/tests/memory-git.test.mjs bizar-dash/tests/memory-sync.test.mjs bizar-dash/tests/obsidian-back-compat.test.mjs bizar-dash/tests/memory-lightrag.test.mjs bizar-dash/tests/memory-config.test.mjs bizar-dash/tests/memory-cli.test.mjs bizar-dash/tests/memory-cli-readlistdelete.test.mjs bizar-dash/tests/memory-cli-setup.test.mjs bizar-dash/tests/memory-conflicts.test.mjs bizar-dash/tests/memory-namespace.test.mjs bizar-dash/tests/memory-path-safety.test.mjs bizar-dash/tests/memory-protocol-drift.test.mjs bizar-dash/tests/memory-roundtrip.test.mjs bizar-dash/tests/cli-bugfixes.test.mjs bizar-dash/tests/server-bugfixes.test.mjs bizar-dash/tests/frontend-bugfixes.test.mjs",
+    "test": "npm run typecheck && npm run test:sdk && node --test cli/install.test.mjs cli/provision.test.mjs cli/worker-dispatcher.test.mjs cli/commands/setup-provider.test.mjs cli/commands/rca.test.mjs cli/__tests__/cost-gate.test.mjs cli/__tests__/feature-list-bridge.test.mjs && node --test bizar-dash/tests/path-safe.test.mjs",
     "build": "npm run build:sdk && npm run build:dash",
     "prepublishOnly": "npm run build"
   },
   "keywords": [
-    "cline",
+    "claude-code",
+    "claude-agent-sdk",
     "ai-agent",
     "norse",
     "multi-agent",
@@ -37,7 +37,8 @@
     "cli",
     "platform",
     "mods",
-    "schedules"
+    "schedules",
+    "mcp"
   ],
   "license": "MIT",
   "engines": {
@@ -45,6 +46,7 @@
     "bun": ">=1.1.0"
   },
   "dependencies": {
+    "@polderlabs/bizar-sdk": "^0.4.0",
     "@mdx-js/mdx": "^3.1.1",
     "@mdx-js/react": "^3.1.1",
     "@opentelemetry/api": "^1.9.1",
@@ -77,8 +79,14 @@
     "ws": "^8.18.0",
     "zod": "4.1.8"
   },
-  "peerDependencies": {},
-  "peerDependenciesMeta": {},
+  "peerDependencies": {
+    "@anthropic-ai/claude-agent-sdk": "*"
+  },
+  "peerDependenciesMeta": {
+    "@anthropic-ai/claude-agent-sdk": {
+      "optional": true
+    }
+  },
   "devDependencies": {
     "@testing-library/jest-dom": "^6.9.1",
     "@testing-library/react": "^16.3.2",
@@ -108,6 +116,6 @@
   },
   "homepage": "https://github.com/DrB0rk/BizarHarness#readme",
   "optionalDependencies": {
-    "@cline/sdk": "^0.0.58"
+    "@anthropic-ai/claude-agent-sdk": "^0.1.0"
   }
 }