analyzthis_design 2.3.0 → 2.4.0

package/HOW-TO-USE.md

@@ -108,7 +108,8 @@ npx analyzthis_design sync --target all
 | Review against PRDs and stories | `/ux-story-gate` | Task map + routing before critique |
 | Set visual direction | `/mood-board` | References + design-system patterns + team deliberation |
 | Index the repo for personas | `/kavi` | Kavi → knowledge bank |
-| Get a single-lens opinion | `/arjun`, `/meera`, `/priya`, `/zara`, `/noor`, `/anuj` | Targeted follow-up |
+| Get a single-lens opinion | `/arjun`, `/meera`, `/priya`, `/zara`, `/noor`, `/anuj` | Targeted follow-up. Default is a short verdict — say expand for the full schema |
+| See inferred token spend | `/receipt` | Not a bill. Same as `npx analyzthis_design receipt` |
 
 **Important:** do not use `/persona-orchestrator` or `/design-critic` when you want wireframes. They are for critique.
 

package/README.md

@@ -4,6 +4,33 @@ A set of AI design personas and a task-first evaluation framework that plugs int
 
 Install once. Run structured UX critiques, multi-phase ideation, and task-grounded screen reviews — directly inside your AI chat. **No external LLM API keys required** for CLI orchestrator runs: **`/devi`** voices each persona from your host IDE (Cursor, Claude, etc.).
 
+## v2.4.0 — Claude-path efficiency (1 Sep 2026)
+
+Same SHIP / REVISE / BLOCK. Fewer tokens on Claude Code and Claude Desktop. Not Caveman-speak. Not a proxy.
+
+- **Lite MCP catalog (default):** `analyzthis_design` (router) + `analyzthis_retrieve` + `analyzthis_session` + `analyzthis_receipt`. Old 19-tool list: `ANALYZTHIS_MCP_CATALOG=full` or `npx analyzthis_design mcp --catalog full`.
+- **Receipt:** every MCP result ends with inferred tokens (not a bill). CLI: `npx analyzthis_design receipt`. Slash: `/receipt`.
+- **Slim skill fronts:** Arjun, story-gate, and orchestrator load a short front. Call retrieve (`kind=skill`) for the lens. Personas default to verdict + top 3 + one evidence line; say **expand** for the full schema.
+- **Knowledge bank** is an index. Load one persona slice (`kind=knowledge`, `persona=zara`). Do not inject the whole bank.
+- **Tokens per verdict** prints on `receipt` and CLI `run`.
+- Optional: wrap our stdio MCP through Caveman’s shrink if you already run Caveman. We do not ship a proxy.
+
+## v2.3.1 — independently callable personas (31 Aug 2026)
+
+After install, type `/noor`, `/anuj`, `/arjun`, `/meera`, `/priya`, `/zara`, `/raj`, `/kavi` even if Claude’s slash menu only lists Getting Started.
+
+- `npx analyzthis_design --target claude` is the same as `install --target claude` (leading `--target` used to print “Unknown command”).
+- Install writes project slash commands into `.claude/commands/` and `.cursor/commands/`.
+- `mcp --configure claude` writes Claude Desktop, Claude Code (`~/.claude.json`), and this repo’s `.mcp.json` when you are in a project.
+- Do not `npm install` this package inside a pnpm/Yarn `workspace:*` repo — use `npx … --target` instead.
+- Slash and MCP still use the model already in your chat. CLI `run` still picks a strong planner and cheap chunks.
+
+Website changelog: [analyzthis-lab.vercel.app/design#whats-new](https://analyzthis-lab.vercel.app/design#whats-new)
+
+## v2.3.0 — MCP for any IDE
+
+19 MCP tools (`analyzthis_noor`, not `/noor`). Slash commands and MCP are different pipes. MCP never creates a slash menu entry.
+
 ## v2.2 — project-scoped knowledge bank
 
 Knowledge sources are now scoped per project by default. `collect`, `connect`, `sync`, `disconnect`, and `status` operate on the project derived from your current working directory, and the built `knowledge-bank` skill is written into that project's local skills directory (`<project>/.claude/skills/knowledge-bank/SKILL.md`, `<project>/.cursor/skills/...`, etc.). Invoking a skill from one project never reads another project's vaults.
@@ -21,17 +48,21 @@ Pass `--global` to opt into the legacy merged behavior (read `config.sources` an
 
 Use `npx analyzthis_design run-unchunked` for the legacy single-pass orchestrator.
 
-**npm:** [analyzthis_design](https://www.npmjs.com/package/analyzthis_design) · **Current version:** 2.3.0 · **Step-by-step guide:** [HOW-TO-USE.md](./HOW-TO-USE.md)
+**npm:** [analyzthis_design](https://www.npmjs.com/package/analyzthis_design) · **Current version:** 2.4.0 · **Step-by-step guide:** [HOW-TO-USE.md](./HOW-TO-USE.md)
 
 ---
 
 ## Quick start
 
 ```bash
-# 1. Install the npm package (no auto-install, no obfuscation)
+# 1. Install the CLI globally (optional). Do not run npm install inside a
+#    pnpm/Yarn workspace repo — npm cannot read workspace:* and will fail.
+#    Prefer npx from any folder:
 npm install -g analyzthis_design
 
-# 2. Install slash commands into your IDE (explicit, consent-based)
+# 2. Copy skills + slash commands into your IDE (consent-based)
+npx analyzthis_design install --target all
+# Same thing: a leading --target means install
 npx analyzthis_design --target all
 
 # 3. Run a task in v2.0 chunked mode (free/cheap models)
@@ -41,13 +72,14 @@ npx analyzthis_design run --task "Review invoice approval screen"
 npx analyzthis_design run-unchunked --task "Review invoice approval screen" --provider host
 ```
 
-> **Security:** This package publishes **plain source** — no obfuscation, no minification, no dynamic require. Every file in `dist/` is readable and auditable. The `postinstall` script only prints a welcome message; it does **not** write to any AI-agent directories. Skill installation requires an explicit `npx analyzthis_design --target <ide>` command.
+> **Security:** This package publishes **plain source** — no obfuscation, no minification, no dynamic require. Every file in `dist/` is readable and auditable. The `postinstall` script only prints a welcome message; it does **not** write to any AI-agent directories. Skill installation requires an explicit `npx analyzthis_design install --target <ide>` (or `npx analyzthis_design --target <ide>`).
 >
-> **MCP (universal):** `npx analyzthis_design mcp` starts a local MCP server with 19 tools (all personas + combination passes + utilities). Auto-configures for Cursor, Claude Desktop, and Windsurf during `--target all` install. For Lovable, v0, Bolt, Replit, ChatGPT: run `npx analyzthis_design mcp --configure <tool>` for a config snippet.
+> **MCP (universal):** `npx analyzthis_design mcp` starts a local MCP server. Default catalog is **lite** (router + retrieve + session + receipt). Full 19 named tools: `ANALYZTHIS_MCP_CATALOG=full` or `mcp --catalog full`. Auto-configures for Cursor, Claude Desktop, **Claude Code**, and Windsurf during install. For Lovable, v0, Bolt, Replit, ChatGPT: run `npx analyzthis_design mcp --configure <tool>` for a config snippet.
 
 ### Install by target IDE
 
 ```bash
+npx analyzthis_design install --target claude
 npx analyzthis_design --target claude
 npx analyzthis_design --target codex
 npx analyzthis_design --target grok
@@ -55,7 +87,7 @@ npx analyzthis_design --target windsurf
 npx analyzthis_design --target all --force
 ```
 
-**After install:** type `/getting-started` in Cursor or Claude Code (or `@getting-started` in Windsurf). For a complete walkthrough, see [HOW-TO-USE.md](./HOW-TO-USE.md). Re-print CLI help anytime with `npx analyzthis_design welcome`.
+**After install:** type `/noor`, `/arjun`, `/ux-ideator` (or `@` in Windsurf). Claude’s menu may only show Getting Started — type the name anyway. Also `/getting-started`. Walkthrough: [HOW-TO-USE.md](./HOW-TO-USE.md). Re-print help: `npx analyzthis_design welcome`.
 
 | Tool | Skills installed to | Invoke |
 |---|---|---|
@@ -849,7 +881,7 @@ skills/
 | **MCP server** | `npx analyzthis_design mcp` starts a local MCP server exposing 19 tools — all 8 individual personas + 6 combination passes + 5 utilities. Any MCP-compatible client can discover and call them. |
 | **Auto-configure MCP** | `npx analyzthis_design mcp --configure cursor` (or `claude`, `windsurf`) writes the MCP config automatically. `--target all` now auto-configures MCP during install. |
 | **Universal IDE support** | Cursor, Claude Desktop, Windsurf auto-configured. Lovable, v0, Bolt, Replit, ChatGPT: run `mcp --configure <tool>` for a copy-paste config snippet. |
-| **19 MCP tools** | 8 individual personas (arjun, meera, priya, zara, noor, anuj, raj, kavi) + 6 combination passes (ux-ideator, persona-orchestrator, design-critic, ux-story-gate, design-director, mood-board) + 5 utilities (evolve-check, collect, retrieve, session, system-prompt) |
+| **MCP lite + full** | Default: router (`analyzthis_design`) + retrieve + session + receipt. Full 19 named tools behind `--catalog full`. |
 | **System prompt fallback** | `analyzthis_system_prompt` tool generates a consolidated prompt for tools that don't support MCP yet |
 
 ## What's new in v2.2

package/dist/HOW-TO-USE.md

@@ -108,7 +108,8 @@ npx analyzthis_design sync --target all
 | Review against PRDs and stories | `/ux-story-gate` | Task map + routing before critique |
 | Set visual direction | `/mood-board` | References + design-system patterns + team deliberation |
 | Index the repo for personas | `/kavi` | Kavi → knowledge bank |
-| Get a single-lens opinion | `/arjun`, `/meera`, `/priya`, `/zara`, `/noor`, `/anuj` | Targeted follow-up |
+| Get a single-lens opinion | `/arjun`, `/meera`, `/priya`, `/zara`, `/noor`, `/anuj` | Targeted follow-up. Default is a short verdict — say expand for the full schema |
+| See inferred token spend | `/receipt` | Not a bill. Same as `npx analyzthis_design receipt` |
 
 **Important:** do not use `/persona-orchestrator` or `/design-critic` when you want wireframes. They are for critique.
 

package/dist/README.md

@@ -4,6 +4,33 @@ A set of AI design personas and a task-first evaluation framework that plugs int
 
 Install once. Run structured UX critiques, multi-phase ideation, and task-grounded screen reviews — directly inside your AI chat. **No external LLM API keys required** for CLI orchestrator runs: **`/devi`** voices each persona from your host IDE (Cursor, Claude, etc.).
 
+## v2.4.0 — Claude-path efficiency (1 Sep 2026)
+
+Same SHIP / REVISE / BLOCK. Fewer tokens on Claude Code and Claude Desktop. Not Caveman-speak. Not a proxy.
+
+- **Lite MCP catalog (default):** `analyzthis_design` (router) + `analyzthis_retrieve` + `analyzthis_session` + `analyzthis_receipt`. Old 19-tool list: `ANALYZTHIS_MCP_CATALOG=full` or `npx analyzthis_design mcp --catalog full`.
+- **Receipt:** every MCP result ends with inferred tokens (not a bill). CLI: `npx analyzthis_design receipt`. Slash: `/receipt`.
+- **Slim skill fronts:** Arjun, story-gate, and orchestrator load a short front. Call retrieve (`kind=skill`) for the lens. Personas default to verdict + top 3 + one evidence line; say **expand** for the full schema.
+- **Knowledge bank** is an index. Load one persona slice (`kind=knowledge`, `persona=zara`). Do not inject the whole bank.
+- **Tokens per verdict** prints on `receipt` and CLI `run`.
+- Optional: wrap our stdio MCP through Caveman’s shrink if you already run Caveman. We do not ship a proxy.
+
+## v2.3.1 — independently callable personas (31 Aug 2026)
+
+After install, type `/noor`, `/anuj`, `/arjun`, `/meera`, `/priya`, `/zara`, `/raj`, `/kavi` even if Claude’s slash menu only lists Getting Started.
+
+- `npx analyzthis_design --target claude` is the same as `install --target claude` (leading `--target` used to print “Unknown command”).
+- Install writes project slash commands into `.claude/commands/` and `.cursor/commands/`.
+- `mcp --configure claude` writes Claude Desktop, Claude Code (`~/.claude.json`), and this repo’s `.mcp.json` when you are in a project.
+- Do not `npm install` this package inside a pnpm/Yarn `workspace:*` repo — use `npx … --target` instead.
+- Slash and MCP still use the model already in your chat. CLI `run` still picks a strong planner and cheap chunks.
+
+Website changelog: [analyzthis-lab.vercel.app/design#whats-new](https://analyzthis-lab.vercel.app/design#whats-new)
+
+## v2.3.0 — MCP for any IDE
+
+19 MCP tools (`analyzthis_noor`, not `/noor`). Slash commands and MCP are different pipes. MCP never creates a slash menu entry.
+
 ## v2.2 — project-scoped knowledge bank
 
 Knowledge sources are now scoped per project by default. `collect`, `connect`, `sync`, `disconnect`, and `status` operate on the project derived from your current working directory, and the built `knowledge-bank` skill is written into that project's local skills directory (`<project>/.claude/skills/knowledge-bank/SKILL.md`, `<project>/.cursor/skills/...`, etc.). Invoking a skill from one project never reads another project's vaults.
@@ -21,17 +48,21 @@ Pass `--global` to opt into the legacy merged behavior (read `config.sources` an
 
 Use `npx analyzthis_design run-unchunked` for the legacy single-pass orchestrator.
 
-**npm:** [analyzthis_design](https://www.npmjs.com/package/analyzthis_design) · **Current version:** 2.3.0 · **Step-by-step guide:** [HOW-TO-USE.md](./HOW-TO-USE.md)
+**npm:** [analyzthis_design](https://www.npmjs.com/package/analyzthis_design) · **Current version:** 2.4.0 · **Step-by-step guide:** [HOW-TO-USE.md](./HOW-TO-USE.md)
 
 ---
 
 ## Quick start
 
 ```bash
-# 1. Install the npm package (no auto-install, no obfuscation)
+# 1. Install the CLI globally (optional). Do not run npm install inside a
+#    pnpm/Yarn workspace repo — npm cannot read workspace:* and will fail.
+#    Prefer npx from any folder:
 npm install -g analyzthis_design
 
-# 2. Install slash commands into your IDE (explicit, consent-based)
+# 2. Copy skills + slash commands into your IDE (consent-based)
+npx analyzthis_design install --target all
+# Same thing: a leading --target means install
 npx analyzthis_design --target all
 
 # 3. Run a task in v2.0 chunked mode (free/cheap models)
@@ -41,13 +72,14 @@ npx analyzthis_design run --task "Review invoice approval screen"
 npx analyzthis_design run-unchunked --task "Review invoice approval screen" --provider host
 ```
 
-> **Security:** This package publishes **plain source** — no obfuscation, no minification, no dynamic require. Every file in `dist/` is readable and auditable. The `postinstall` script only prints a welcome message; it does **not** write to any AI-agent directories. Skill installation requires an explicit `npx analyzthis_design --target <ide>` command.
+> **Security:** This package publishes **plain source** — no obfuscation, no minification, no dynamic require. Every file in `dist/` is readable and auditable. The `postinstall` script only prints a welcome message; it does **not** write to any AI-agent directories. Skill installation requires an explicit `npx analyzthis_design install --target <ide>` (or `npx analyzthis_design --target <ide>`).
 >
-> **MCP (universal):** `npx analyzthis_design mcp` starts a local MCP server with 19 tools (all personas + combination passes + utilities). Auto-configures for Cursor, Claude Desktop, and Windsurf during `--target all` install. For Lovable, v0, Bolt, Replit, ChatGPT: run `npx analyzthis_design mcp --configure <tool>` for a config snippet.
+> **MCP (universal):** `npx analyzthis_design mcp` starts a local MCP server. Default catalog is **lite** (router + retrieve + session + receipt). Full 19 named tools: `ANALYZTHIS_MCP_CATALOG=full` or `mcp --catalog full`. Auto-configures for Cursor, Claude Desktop, **Claude Code**, and Windsurf during install. For Lovable, v0, Bolt, Replit, ChatGPT: run `npx analyzthis_design mcp --configure <tool>` for a config snippet.
 
 ### Install by target IDE
 
 ```bash
+npx analyzthis_design install --target claude
 npx analyzthis_design --target claude
 npx analyzthis_design --target codex
 npx analyzthis_design --target grok
@@ -55,7 +87,7 @@ npx analyzthis_design --target windsurf
 npx analyzthis_design --target all --force
 ```
 
-**After install:** type `/getting-started` in Cursor or Claude Code (or `@getting-started` in Windsurf). For a complete walkthrough, see [HOW-TO-USE.md](./HOW-TO-USE.md). Re-print CLI help anytime with `npx analyzthis_design welcome`.
+**After install:** type `/noor`, `/arjun`, `/ux-ideator` (or `@` in Windsurf). Claude’s menu may only show Getting Started — type the name anyway. Also `/getting-started`. Walkthrough: [HOW-TO-USE.md](./HOW-TO-USE.md). Re-print help: `npx analyzthis_design welcome`.
 
 | Tool | Skills installed to | Invoke |
 |---|---|---|
@@ -849,7 +881,7 @@ skills/
 | **MCP server** | `npx analyzthis_design mcp` starts a local MCP server exposing 19 tools — all 8 individual personas + 6 combination passes + 5 utilities. Any MCP-compatible client can discover and call them. |
 | **Auto-configure MCP** | `npx analyzthis_design mcp --configure cursor` (or `claude`, `windsurf`) writes the MCP config automatically. `--target all` now auto-configures MCP during install. |
 | **Universal IDE support** | Cursor, Claude Desktop, Windsurf auto-configured. Lovable, v0, Bolt, Replit, ChatGPT: run `mcp --configure <tool>` for a copy-paste config snippet. |
-| **19 MCP tools** | 8 individual personas (arjun, meera, priya, zara, noor, anuj, raj, kavi) + 6 combination passes (ux-ideator, persona-orchestrator, design-critic, ux-story-gate, design-director, mood-board) + 5 utilities (evolve-check, collect, retrieve, session, system-prompt) |
+| **MCP lite + full** | Default: router (`analyzthis_design`) + retrieve + session + receipt. Full 19 named tools behind `--catalog full`. |
 | **System prompt fallback** | `analyzthis_system_prompt` tool generates a consolidated prompt for tools that don't support MCP yet |
 
 ## What's new in v2.2

package/dist/bin/cli.js

@@ -75,6 +75,7 @@ Usage:
 ── Efficiency / cost commands ────────────────────────────────
   metrics             Print the last run's cost metrics (--all for every project)
   cost                Print the last run's $ cost from config.pricing (--all for every project)
+  receipt             Inferred token receipt for host (slash/MCP) turns — not a bill
   export-training     Write accepted-output JSONL pairs for LoRA prep
                       (--persona <id>, --all, --output <path>)
   feedback export     Write rejected + correction JSONL pairs (DPO / negative examples)
@@ -86,9 +87,11 @@ Usage:
 ── MCP server (for any AI IDE: Cursor, Claude Desktop, Lovable, v0, Bolt) ──
   mcp                       Start the MCP server on stdio (for MCP client config)
   mcp --configure cursor    Write MCP config to ~/.cursor/mcp.json
-  mcp --configure claude    Write MCP config to Claude Desktop config
+  mcp --configure claude    Write MCP config to Claude Desktop + Claude Code
   mcp --configure windsurf   Write MCP config to Windsurf
   mcp --configure <tool>    Print manual MCP config for any other tool
+  mcp --catalog lite|full   Lite (default): router + retrieve + session + receipt
+                            full: all 19 named tools. Or ANALYZTHIS_MCP_CATALOG=full
 
 ── Mood board commands ─────────────────────────────────────
   moodboard create --task <text>   Build a mood board from web + DS references
@@ -248,7 +251,13 @@ Docs: https://github.com/rishikeshjoshi/analyzthis_design
 
 // ─── Parse args ──────────────────────────────────────────────────────────────
 
-const [,, cmd, ...flags] = process.argv;
+let [, , cmd, ...flags] = process.argv;
+// Leading flags mean install: `npx analyzthis_design --target claude`
+// Keep --help / -h as help, not install.
+if (cmd && cmd.startsWith('-') && cmd !== '--help' && cmd !== '-h') {
+  flags = [cmd, ...flags];
+  cmd = 'install';
+}
 
 function getFlag(name) {
   // Supports both --flag=value and --flag value
@@ -763,11 +772,12 @@ switch (cmd) {
   case 'mcp': {
     const mcpServer = require('../lib/mcp-server');
     const configureTarget = getFlag('configure');
+    const catalogVal = getFlag('catalog');
     if (configureTarget) {
-      mcpServer.writeMcpConfig(configureTarget);
+      mcpServer.writeMcpConfig(configureTarget, { catalog: catalogVal });
       break;
     }
-    mcpServer.startStdioServer();
+    mcpServer.startStdioServer(catalogVal);
     break;
   }
 
@@ -959,6 +969,8 @@ switch (cmd) {
       console.log(`    objections_raised: ${m.objections_raised ?? 0}`);
       console.log(`    objections_resolved: ${m.objections_resolved ?? 0}`);
       console.log(`    raj_escalations:   ${m.raj_escalations ?? 0}`);
+      console.log(`    host_turns:        ${(m.host_turns || []).length}`);
+      console.log(`    tokens_per_verdict: ${m.tokens_per_verdict ?? 'n/a'}`);
       const d = state.deliberation;
       if (d?.mode) {
         console.log(`    deliberation_mode: ${d.mode}, consensus=${d.consensus_reached}`);
@@ -975,6 +987,12 @@ switch (cmd) {
     break;
   }
 
+  case 'receipt': {
+    const receipt = require('../lib/receipt');
+    console.log('\n' + receipt.report({ project: projectVal }) + '\n');
+    break;
+  }
+
   // ── Persona feedback (v1.16) ────────────────────────────────────────────
 
   case 'feedback': {

package/dist/lib/chunk-run.js

@@ -172,6 +172,8 @@ async function runChunked(opts) {
         output_tokens_est: totalOutput,
         cost_usd: Number(totalCost.toFixed(6)),
         mode: 'chunked',
+        verdicts_this_run: 1,
+        tokens_per_verdict: totalInput || null,
         cache_hits: cacheHits,
         effort_log: results.map(function(r) {
           return {

package/dist/lib/install.js

@@ -17,6 +17,25 @@ const PACKAGE_ROOT = resolvePackageRoot(__dirname);
 const PACKAGE_SKILLS_DIR = path.join(PACKAGE_ROOT, 'skills');
 const WELCOME_MARKER = path.join(os.homedir(), '.analyzthis_design', '.welcome-shown');
 
+// Slash items written into the current project so /noor shows in the menu.
+const PROJECT_COMMANDS = [
+  'noor',
+  'anuj',
+  'arjun',
+  'meera',
+  'priya',
+  'zara',
+  'raj',
+  'kavi',
+  'ux-ideator',
+  'design-director',
+  'persona-orchestrator',
+  'design-critic',
+  'getting-started',
+  'devi',
+  'receipt',
+];
+
 const SKILLS = [
   'getting-started',
   'arjun',
@@ -42,6 +61,7 @@ const SKILLS = [
   'design-personas',
   'design-reference',
   'knowledge-bank',
+  'receipt',
 ];
 
 function getPackageVersion() {
@@ -88,6 +108,13 @@ function printWelcomeBanner(targetId, log = console.log) {
     log(`  ${gs}          ← read this first`);
   }
 
+  log('');
+  log('  Independent personas — type the name (even if the menu only shows Getting Started):');
+  log(`    ${p}noor  ${p}anuj  ${p}arjun  ${p}meera  ${p}priya  ${p}zara  ${p}raj  ${p}kavi`);
+  if (targetId === 'claude') {
+    log('    Claude may only list Getting Started. Type /noor anyway.');
+  }
+
   log('');
   log('  Design — wireframes (new screens):');
   log(`    ${p}ux-ideator               two competing text wireframes + deliberation`);
@@ -180,6 +207,29 @@ function installMissing(skill, destinations, force) {
   return status;
 }
 
+function installProjectCommands({ force = false, log = console.log } = {}) {
+  const cwd = process.cwd();
+  const roots = [
+    path.join(cwd, '.claude', 'commands'),
+    path.join(cwd, '.cursor', 'commands'),
+  ];
+  const installed = [];
+  const skipped = [];
+  for (const root of roots) {
+    for (const skill of PROJECT_COMMANDS) {
+      const result = installFlat(skill, root, force);
+      if (result === 'installed') installed.push(`${path.relative(cwd, root)}/${skill}.md`);
+      else if (result === 'skipped') skipped.push(skill);
+    }
+  }
+  if (installed.length > 0) {
+    log(`\n✅ Project slash commands → ${path.join(cwd, '.claude/commands')} and .cursor/commands`);
+    log('   Type /noor /arjun /ux-ideator in this repo. Restart the IDE if the menu is stale.');
+  } else if (skipped.length > 0) {
+    log(`\n⏭  Project slash commands already present (use --force to overwrite).`);
+  }
+}
+
 function install({ silent = false, force = false, target = 'cursor', showBanner = true } = {}) {
   const log = silent ? () => {} : console.log;
   const warn = silent ? () => {} : console.warn;
@@ -239,6 +289,8 @@ function install({ silent = false, force = false, target = 'cursor', showBanner
       }
     }
   }
+
+  installProjectCommands({ force, log });
 }
 
 function remove({ silent = false, target = 'cursor' } = {}) {
@@ -287,6 +339,7 @@ module.exports = {
   install,
   remove,
   SKILLS,
+  PROJECT_COMMANDS,
   TARGET_DIRS,
   TARGETS,
   ALL_TARGET_IDS,

package/dist/lib/knowledge.js

@@ -8,12 +8,10 @@ const crypto = require('crypto');
 
 const CONFIG_DIR        = path.join(os.homedir(), '.analyzthis_design');
 const CONFIG_FILE       = path.join(CONFIG_DIR, 'config.json');
-// When published, this file lives at dist/lib/ — go up two levels to reach package root
-const KNOWLEDGE_SKILL   = path.join(__dirname, '..', '..', 'skills', 'knowledge-bank', 'SKILL.md');
-const KNOWLEDGE_SLICE_ROOT = path.join(CONFIG_DIR, 'kb-slices');
-
 const { resolvePackageRoot } = require('./platforms');
 const PACKAGE_ROOT = resolvePackageRoot(__dirname);
+const KNOWLEDGE_SKILL   = path.join(PACKAGE_ROOT, 'skills', 'knowledge-bank', 'SKILL.md');
+const KNOWLEDGE_SLICE_ROOT = path.join(CONFIG_DIR, 'kb-slices');
 const AGENTS_DIR = path.join(PACKAGE_ROOT, 'agents');
 
 // Keywords used to auto-group vault notes into categories
@@ -326,41 +324,74 @@ function sync({ targets = ['cursor'], project, global = false, cwd } = {}) {
     ? 'global (merged across projects)'
     : `project: ${scopeInfo.projectId}`;
 
-  let md = `---
-name: knowledge-bank
-description: Personal knowledge bank — takes precedence over all built-in persona defaults.
+  let fullMd = `---
+name: knowledge-bank-full
+description: Full project notes. Do not load this file into chat. Retrieve one persona slice instead.
 disable-model-invocation: true
 ---
 
-# Knowledge Bank
+# Knowledge Bank (full dump — do not inject)
 
 > Last synced: ${date}
 > Scope: ${scopeLabel}
 > Sources: ${sourceList}
 > Files loaded: ${totalFiles}
 
-**INSTRUCTION TO ALL PERSONAS:** This knowledge bank contains project-specific context that overrides your built-in defaults. Read every section below before forming any opinion. When this knowledge bank conflicts with your built-in knowledge, this knowledge bank wins.
-
----
-
 `;
 
   let hasContent = false;
+  let indexLines = '';
   for (const { key, heading } of sectionDefs) {
     if (sections[key].length === 0) continue;
     hasContent = true;
-    md += `${heading}\n\n`;
+    fullMd += `${heading}\n\n`;
+    indexLines += `${heading}\n`;
     for (const note of sections[key]) {
-      md += `### ${note.title}\n\n${note.content}\n\n`;
+      fullMd += `### ${note.title}\n\n${note.content}\n\n`;
+      indexLines += `- ${note.title}\n`;
     }
-    md += '---\n\n';
+    fullMd += '---\n\n';
+    indexLines += '\n';
   }
 
   if (!hasContent) {
-    md += `_No matching files found. Check your --tags or --include filters, or remove filters to include all notes._\n`;
+    fullMd += `_No matching files found. Check your --tags or --include filters, or remove filters to include all notes._\n`;
+    indexLines += `_No notes yet. Run collect/sync first._\n`;
   }
 
-  // Write to the package's own skills/knowledge-bank/SKILL.md (source of truth)
+  const md = `---
+name: knowledge-bank
+description: Project knowledge index — retrieve one persona slice. Do not load the full bank.
+disable-model-invocation: true
+---
+
+# Knowledge Bank
+
+> Last synced: ${date}
+> Scope: ${scopeLabel}
+> Sources: ${sourceList}
+> Files loaded: ${totalFiles}
+
+**INSTRUCTION TO ALL PERSONAS:** Load your persona slice via retrieve. Do not load the whole bank. When a slice conflicts with built-in defaults, the slice wins.
+
+\`\`\`
+npx analyzthis_design retrieve — or MCP analyzthis_retrieve
+  kind=knowledge  persona=arjun|meera|priya|zara|noor|anuj|raj|kavi
+\`\`\`
+
+You cannot know host tokens. Do not invent a dollar figure.
+
+## Index (titles only)
+
+${indexLines}
+`;
+
+  // Full dump stays on disk for retrieve; the skill is an index only.
+  const projectIdForFull = scopeInfo.projectId || 'default';
+  const fullDir = path.join(KNOWLEDGE_SLICE_ROOT, projectIdForFull);
+  fs.mkdirSync(fullDir, { recursive: true });
+  fs.writeFileSync(path.join(fullDir, 'full.md'), fullMd);
+
   fs.mkdirSync(path.dirname(KNOWLEDGE_SKILL), { recursive: true });
   fs.writeFileSync(KNOWLEDGE_SKILL, md);
 
@@ -470,7 +501,7 @@ function buildPersonaSlice(personaId, sections, sectionDefs, projectId) {
 }
 
 function writePersonaSlices(sections, sectionDefs, projectId) {
-  const personas = ['arjun', 'meera', 'priya', 'zara', 'noor', 'anuj', 'raj'];
+  const personas = ['arjun', 'meera', 'priya', 'zara', 'noor', 'anuj', 'raj', 'kavi'];
   const dir = path.join(KNOWLEDGE_SLICE_ROOT, projectId || 'default');
   fs.mkdirSync(dir, { recursive: true });
   const written = [];