Add modular Langfuse-backed AI observability

.env.example

@@ -29,4 +29,10 @@
 # Search Provider Configuration
 
 # SEARCH_PROVIDER=perplexity
-# PERPLEXITY_API_KEY=pplx-0000000000000000
+# PERPLEXITY_API_KEY=pplx-0000000000000000
+
+# Observability Configuration (optional)
+# OBSERVABILITY_BACKEND=langfuse
+# LANGFUSE_PUBLIC_KEY=pk-lf-0000000000000000
+# LANGFUSE_SECRET_KEY=sk-lf-0000000000000000
+# LANGFUSE_BASE_URL=https://cloud.langfuse.com

CONFIGURATION.md

@@ -99,6 +99,24 @@ SEARCH_PROVIDER=perplexity
 PERPLEXITY_API_KEY=pplx-...
 ```
 
+### Observability
+
+VectorLint can optionally emit AI execution telemetry to Langfuse. The first implementation is scoped to the Vercel AI SDK calls made through `VercelAIProvider`.
+
+**Example configuration for Langfuse:**
+
+```bash
+OBSERVABILITY_BACKEND=langfuse
+LANGFUSE_PUBLIC_KEY=pk-lf-...
+LANGFUSE_SECRET_KEY=sk-lf-...
+# Optional for self-hosted Langfuse. Defaults to cloud.langfuse.com.
+LANGFUSE_BASE_URL=https://cloud.langfuse.com
+```
+
+- Observability is best-effort and non-blocking.
+- If initialization or shutdown fails, VectorLint logs a warning and continues.
+- Prompts and outputs are recorded when Langfuse observability is enabled.
+
 ### False-Positive Filtering (PAT)
 
 VectorLint uses PAT (Pay A Tax) style gate checks to reduce false positives. The model may return many raw candidates, but only candidates that pass deterministic gate checks are surfaced in CLI output.

README.md

@@ -14,6 +14,8 @@ Install globally from npm:
 npm install -g vectorlint
 ```
 
+VectorLint currently requires Node.js `20.6+`.
+
 Verify installation:
 
 ```bash
@@ -123,6 +125,26 @@ VectorLint is bundled with a `VectorLint` preset containing rules for AI pattern
 
 👉 **[Learn how to create custom rules →](./CREATING_RULES.md)**
 
+### 4. Optional: Enable Langfuse observability
+
+VectorLint can emit AI execution telemetry through Langfuse without hardcoding Langfuse into the provider layer. This is best-effort instrumentation for the Vercel AI SDK calls used by `VercelAIProvider`.
+
+Add these environment variables to your global config or local `.env` file:
+
+```toml
+[env]
+OBSERVABILITY_BACKEND = "langfuse"
+LANGFUSE_PUBLIC_KEY = "pk-lf-..."
+LANGFUSE_SECRET_KEY = "sk-lf-..."
+# Optional for self-hosted Langfuse. Defaults to cloud.langfuse.com.
+LANGFUSE_BASE_URL = "https://cloud.langfuse.com"
+```
+
+Notes:
+- Observability is non-blocking. If Langfuse setup fails, VectorLint continues without telemetry.
+- Prompts and outputs are recorded when Langfuse observability is enabled.
+- Do not send secrets, credentials, or PII unless your policy explicitly allows observability tooling to access that data.
+
 ## Agent Mode
 
 Agent mode is for reviews that need context from several files such as