@triedotdev/mcp 1.0.18 → 1.0.20

package/README.md

@@ -8,11 +8,12 @@
 
 - **20 Built-in Agents** - Security, Privacy, SOC 2, Legal, Architecture, Performance, E2E, Visual QA, Data Flow, Agent Smith, and more
 - **Super Reviewer** - Interactive PR reviews: walks through changes file-by-file with AI guidance
-- **Agent Smith v2.0** - Ultimate vibe code enforcer: 38 hunters targeting AI-generated anti-patterns, file-level analysis, cross-file detection, persistent memory
+- **Agent Smith** - Ultimate AI code enforcer: 38 hunters targeting AI-generated anti-patterns, file-level analysis, cross-file detection, persistent memory
 - **Parallel Execution** - All agents run simultaneously for fast scans
 - **YOLO Mode** - Autonomous auto-fixing as you code
 - **Custom Agents** - Create agents from PDFs, docs, or style guides
-- **No API Key Required** - Works with any MCP-compatible AI tool (Cursor, Claude Code, VS Code)
+- **Works Everywhere** - Works with any MCP-compatible AI tool (Cursor, Claude Code, VS Code)
+- **AI-Enhanced Mode** - Optional: Set `ANTHROPIC_API_KEY` for deeper AI analysis
 - **Smart Triaging** - Only activates relevant agents based on code context
 - **Docker Support** - Run in containers for CI/CD or isolated environments
 
@@ -47,6 +48,54 @@ claude mcp add Trie --scope user -- npx @triedotdev/mcp
 
 **Restart Claude Code after adding the MCP server** for changes to take effect.
 
+## AI-Enhanced Mode (Optional)
+
+Trie agents work in two modes:
+
+### Pattern-Only Mode (Default)
+- Fast pattern detection using regex and static analysis
+- No API key required
+- Still finds many issues
+
+### AI-Enhanced Mode
+- Pattern detection + AI validation and expansion
+- Validates findings (reduces false positives)
+- Finds deeper issues (logic bugs, race conditions)
+- Provides intelligent, contextual fixes
+- Requires `ANTHROPIC_API_KEY`
+
+### Setup API Key
+
+**Option 1: Environment Variable**
+```bash
+export ANTHROPIC_API_KEY=sk-ant-...
+```
+
+**Option 2: MCP Server Config (Cursor)**
+```json
+{
+  "mcpServers": {
+    "Trie": {
+      "command": "npx",
+      "args": ["@triedotdev/mcp"],
+      "env": {
+        "ANTHROPIC_API_KEY": "sk-ant-..."
+      }
+    }
+  }
+}
+```
+
+**Option 3: MCP Server Config (Claude Code)**
+```bash
+claude mcp add Trie --scope user -e ANTHROPIC_API_KEY=sk-ant-... -- npx @triedotdev/mcp
+```
+
+When AI is enabled, you'll see:
+- `🤖 AI-powered analysis enabled` in the output
+- `[AI VALIDATED]` and `[AI FOUND]` tags on issues
+- Richer fix recommendations
+
 ## Usage
 
 Once configured, just ask your AI assistant:
@@ -90,9 +139,9 @@ Use trie_pr_review with pr:"12345"
 
 **Manually invoked** - Use `trie_agent_smith` directly. Not included in `trie_scan`.
 
-*"I'm going to be honest with you... I hate this vibe code."*
+*"I'm going to be honest with you... I hate this AI code."*
 
-Agent Smith v2.0 is the **ultimate vibe code enforcer** — specifically designed to hunt down AI-generated code anti-patterns from Cursor, v0, Lovable, Bolt, and other AI tools.
+Agent Smith v2.0 is the **ultimate AI code enforcer** — specifically designed to hunt down AI-generated code anti-patterns from Cursor, v0, Lovable, Bolt, and other AI tools.
 
 ```
 Use trie_agent_smith
@@ -184,23 +233,28 @@ Use trie_smith
 | `emoji-overflow-hunter` | Any emoji usage | Use proper icons (Lucide, Heroicons) |
 | `inter-font-hunter` | Inter/system-ui font | Try Space Grotesk, DM Sans, Outfit |
 
-### 3-Phase Analysis
+### How Agent Smith Works
+
+Agent Smith uses a **hybrid pattern + AI architecture**:
+
+**Phase 1: Pattern Detection (Fast, ~0.1s)**
+- 38 specialized regex hunters scan files in parallel
+- File-level metrics (giant files, hook counts, import chaos)
+- Cross-file pattern detection (issues appearing in 5+ files)
 
-1. **Pattern Hunting** - 38 sub-agents scan every file in parallel
-2. **File-Level Analysis** - Detects structural issues:
-   - Giant files (500+ lines)
-   - State explosion (10+ useState)
-   - Effect hell (5+ useEffect)
-   - Any explosion (5+ `any` types)
-   - Console flood (10+ console.log)
-   - Import chaos (30+ imports)
-3. **Cross-File Detection** - Patterns appearing in 5+ files flagged as codebase-wide issues
+**Phase 2: AI Enhancement (If API key is set, ~5-10s)**
+- Validates pattern findings (TRUE_POSITIVE vs FALSE_POSITIVE)
+- Finds deeper issues that patterns miss (logic bugs, race conditions)
+- Provides "inevitability scores" (0-100) for prioritization
+- Generates specific, copy-paste-ready fixes
+- Adds Agent Smith philosophical commentary
 
 ### What Makes Agent Smith Different
 
 | Feature | Description |
 |---------|-------------|
-| **Vibe Code Focus** | Specifically targets patterns AI tools commonly get wrong |
+| **Hybrid AI** | Fast regex detection + deep AI reasoning on findings |
+| **AI Code Focus** | Specifically targets patterns AI tools commonly get wrong |
 | **Multiplier Effect** | Finds one issue → searches for EVERY similar instance |
 | **Persistent Memory** | Remembers dismissed issues, brings them back if they multiply |
 | **Inevitability Score** | 0-100 rating of how likely to cause production problems |
@@ -393,7 +447,7 @@ docker-compose up
 | Agent | Command | What It Does |
 |-------|---------|--------------|
 | **Super Reviewer** | `trie_pr_review` | Interactive PR review: walks through changes, explains code, finds issues |
-| **Agent Smith** | `trie_agent_smith` | Ultimate vibe code enforcer: 38 hunters, file analysis, cross-file detection |
+| **Agent Smith** | `trie_agent_smith` | Ultimate AI code enforcer: 38 hunters, file analysis, cross-file detection |
 | **Comprehension** | `trie_explain` | Plain-language explanations of code, issues, or risks |
 
 **Note:** Super Reviewer and Agent Smith are manually invoked agents. They don't run during `trie_scan`; use their dedicated commands directly.
@@ -414,7 +468,7 @@ Custom agents automatically activate during scans based on their rules.
 |------|-------------|
 | `trie_scan` | Intelligent scan with automatic agent selection |
 | `trie_pr_review` | Interactive PR review - walks through changes file-by-file |
-| `trie_agent_smith` | Ultimate vibe code enforcer - 38 hunters, file analysis, cross-file detection |
+| `trie_agent_smith` | Ultimate AI code enforcer - 38 hunters, file analysis, cross-file detection |
 | `trie_fix` | Apply high-confidence fixes |
 | `trie_explain` | Plain-language explanations |
 | `trie_watch` | Continuous scanning mode |

package/dist/{agent-smith-PRK7TYEI.js → agent-smith-BKHHC4G6.js}

@@ -1,10 +1,10 @@
 import {
   AgentSmithAgent,
   SUB_AGENT_PATTERNS
-} from "./chunk-WSBTQJMH.js";
+} from "./chunk-EEVALBHC.js";
 import "./chunk-DGUM43GV.js";
 export {
   AgentSmithAgent,
   SUB_AGENT_PATTERNS
 };
-//# sourceMappingURL=agent-smith-PRK7TYEI.js.map
+//# sourceMappingURL=agent-smith-BKHHC4G6.js.map