diff --git a/.dockerignore b/.dockerignore
new file mode 100644
index 0000000..ee75bcb
--- /dev/null
+++ b/.dockerignore
@@ -0,0 +1,14 @@
+.git
+.github
+.gitignore
+*.md
+.env
+.env.*
+Dockerfile
+.dockerignore
+coverage.out
+docs/
+deploy/
+api/
+go.work
+go.work.sum
diff --git a/.env.example b/.env.example
new file mode 100644
index 0000000..c15da9a
--- /dev/null
+++ b/.env.example
@@ -0,0 +1,5 @@
+# sight environment variables — copy to .env and fill in
+# LLM API key for code review (consumers provide their own via the Provider interface)
+# sight itself doesn't read this — configure your LLM client before passing it to sight.NewReviewer()
+SIGHT_PROVIDER=
+SIGHT_API_KEY=
diff --git a/.github/workflows/docker.yml b/.github/workflows/docker.yml
new file mode 100644
index 0000000..7b40db5
--- /dev/null
+++ b/.github/workflows/docker.yml
@@ -0,0 +1,63 @@
+name: Docker
+
+on:
+  push:
+    branches: [main]
+    tags: ["v*"]
+  pull_request:
+    branches: [main]
+    paths:
+      - "Dockerfile"
+      - "**.go"
+      - "go.mod"
+      - "go.sum"
+
+permissions:
+  contents: read
+  packages: write
+
+env:
+  REGISTRY: ghcr.io
+  IMAGE_NAME: graycodeai/sight
+
+jobs:
+  build-and-push:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@v3
+
+      - name: Log in to GHCR
+        if: github.event_name != 'pull_request'
+        uses: docker/login-action@v3
+        with:
+          registry: ${{ env.REGISTRY }}
+          username: ${{ github.actor }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+
+      - name: Docker metadata
+        id: meta
+        uses: docker/metadata-action@v5
+        with:
+          images: ${{ env.REGISTRY }}/${{ env.IMAGE_NAME }}
+          tags: |
+            type=ref,event=branch
+            type=semver,pattern={{version}}
+            type=semver,pattern={{major}}.{{minor}}
+            type=sha,prefix=sha-
+
+      - name: Build and push
+        uses: docker/build-push-action@v6
+        with:
+          context: .
+          push: ${{ github.event_name != 'pull_request' }}
+          tags: ${{ steps.meta.outputs.tags }}
+          labels: ${{ steps.meta.outputs.labels }}
+          cache-from: type=gha
+          cache-to: type=gha,mode=max
+          build-args: |
+            VERSION=${{ github.ref_name }}
+            COMMIT=${{ github.sha }}
+            BUILD_DATE=${{ github.event.head_commit.timestamp }}
diff --git a/Dockerfile b/Dockerfile
new file mode 100644
index 0000000..8dd91f6
--- /dev/null
+++ b/Dockerfile
@@ -0,0 +1,29 @@
+FROM golang:1.26.3-alpine AS builder
+
+RUN apk add --no-cache git ca-certificates tzdata
+
+WORKDIR /build
+COPY go.mod go.sum ./
+RUN go mod download && go mod verify
+
+COPY . .
+ARG VERSION=dev
+ARG COMMIT=none
+ARG BUILD_DATE=unknown
+RUN CGO_ENABLED=0 GOOS=linux go build -trimpath \
+    -ldflags="-s -w \
+      -X main.Version=${VERSION} \
+      -X main.Commit=${COMMIT} \
+      -X main.BuildDate=${BUILD_DATE}" \
+    -o sight ./cmd/sight
+
+FROM alpine:3.21
+RUN apk add --no-cache ca-certificates tini && \
+    adduser -D -u 1000 sight
+
+COPY --from=builder /build/sight /usr/local/bin/sight
+COPY --from=builder /usr/share/zoneinfo /usr/share/zoneinfo
+
+USER sight
+ENTRYPOINT ["tini", "--", "sight"]
+CMD ["mcp"]
diff --git a/api/openapi.yaml b/api/openapi.yaml
new file mode 100644
index 0000000..8507b5c
--- /dev/null
+++ b/api/openapi.yaml
@@ -0,0 +1,95 @@
+openapi: "3.1.0"
+info:
+  title: sight — Code Review Tool Reference
+  description: |
+    sight is an AI-powered code review library for diffs.
+    It parses unified diffs, enriches with code context and git history,
+    and runs parallel multi-concern reviews through an LLM provider.
+
+    sight operates as a Go library and MCP server (stdio or HTTP).
+    This document describes the MCP tool surface as a machine-readable reference.
+  version: "0.1.0"
+  license:
+    name: MIT
+    url: https://github.com/GrayCodeAI/sight/blob/main/LICENSE
+  contact:
+    url: https://github.com/GrayCodeAI/sight
+
+# No HTTP server by default — MCP uses stdio transport.
+# HTTP transport is available via: sight mcp --transport http --addr 127.0.0.1:8080
+
+tags:
+  - name: review
+    description: Code review tools
+  - name: taint
+    description: Security taint analysis
+
+x-mcp-server:
+  transport: stdio
+  binary: sight
+  start_command: ["sight", "mcp"]
+  http_transport_command: ["sight", "mcp", "--transport", "http", "--addr", "127.0.0.1:8080"]
+
+x-mcp-tools:
+  sight_review:
+    description: AI code review on a unified diff — security, correctness, style, performance
+    inputSchema:
+      type: object
+      required: [diff]
+      properties:
+        diff:
+          type: string
+          description: Unified diff text (output of `git diff`)
+        preset:
+          type: string
+          enum: [Quick, Standard, Thorough, SecurityFocus, CI]
+          default: Standard
+          description: Review thoroughness preset
+        exclude_files:
+          type: array
+          items:
+            type: string
+          description: File patterns to exclude from review (glob)
+        max_tokens:
+          type: integer
+          description: Maximum tokens for the LLM review
+
+  sight_describe:
+    description: Generate a pull request description from a unified diff
+    inputSchema:
+      type: object
+      required: [diff]
+      properties:
+        diff:
+          type: string
+          description: Unified diff text
+
+  sight_improve:
+    description: Suggest specific code improvements for a diff
+    inputSchema:
+      type: object
+      required: [diff]
+      properties:
+        diff:
+          type: string
+        focus:
+          type: string
+          description: Focus area for improvements (e.g. "performance", "readability")
+
+  sight_taint:
+    description: SSA-based cross-function taint analysis for security vulnerabilities
+    inputSchema:
+      type: object
+      required: [path]
+      properties:
+        path:
+          type: string
+          description: Absolute path to the Go package or module root
+        patterns:
+          type: string
+          default: "./..."
+          description: Go package patterns to analyze
+        json:
+          type: boolean
+          default: false
+          description: Output results as JSON
diff --git a/deploy/docker/docker-compose.yml b/deploy/docker/docker-compose.yml
new file mode 100644
index 0000000..271dd99
--- /dev/null
+++ b/deploy/docker/docker-compose.yml
@@ -0,0 +1,12 @@
+name: sight
+
+services:
+  sight:
+    build:
+      context: ../../
+      dockerfile: Dockerfile
+    image: ghcr.io/graycodeai/sight:dev
+    env_file:
+      - path: ../../.env.example
+        required: false
+    command: ["mcp"]
diff --git a/docs/architecture.md b/docs/architecture.md
new file mode 100644
index 0000000..d6ca4d3
--- /dev/null
+++ b/docs/architecture.md
@@ -0,0 +1,124 @@
+<div align="center">
+
+# 👁️ sight Architecture
+
+**AI-Powered Code Review on Diffs**
+
+[![Go](https://img.shields.io/badge/Go-1.26+-00ADD8?logo=go)](https://go.dev/)
+[![Protocol](https://img.shields.io/badge/Protocol-MCP-purple)]()
+
+</div>
+
+---
+
+## 🎯 Overview
+
+sight is an AI-powered code review library for Go. It parses unified diffs, enriches with **code context** and **git history**, and runs **parallel multi-concern reviews** through an LLM provider.
+
+> 💡 No LLM client bundled — consumers inject their own via the `Provider` interface.
+
+---
+
+## 🧱 Components
+
+```
+sight/
+├── api/openapi.yaml          📜 MCP tool surface reference
+├── cmd/sight/main.go         🖥️ CLI entry (mcp, taint subcommands)
+├── sight.go                  📤 Public API: Review(), Finding, Result, Stats
+├── reviewer.go               🔄 Reviewer: parallel concern orchestration
+├── options.go                ⚙️ config, With* functions, presets
+├── provider.go               🔌 Provider interface (consumers implement)
+├── severity.go               📊 Re-exports from hawk/shared/types
+├── static_rules.go           🛡️ 30+ static analysis rules
+├── taint_analysis.go         🔗 SSA-based taint tracking
+├── sast_integration.go       🔒 SAST-LLM fusion
+├── autofix.go                🔧 Fix suggestion pipeline
+├── eval.go                   📊 Evaluation harness
+├── mcp/                      🔌 MCP server (stdio + HTTP)
+└── internal/
+    ├── diff/                 📄 Unified diff parser
+    ├── review/               🧠 Concerns, prompts, response parsing
+    ├── comment/              💬 Inline comment formatting
+    ├── context/              📖 Code context + git blame
+    └── output/               📊 SARIF and terminal formatters
+```
+
+---
+
+## 📤 Public API
+
+```go
+// 🚀 One-shot review
+result, err := sight.Review(ctx, diffText,
+    sight.WithProvider(myLLMProvider),
+    sight.Thorough,
+)
+
+// 🔄 Reusable reviewer
+reviewer := sight.NewReviewer(sight.WithProvider(myLLMProvider))
+result, err := reviewer.Review(ctx, diffText)
+
+// ❌ Check if any findings are above threshold
+if result.Failed() {
+    fmt.Printf("Review failed: %d findings\n", len(result.Findings))
+}
+```
+
+---
+
+## 🔌 Provider Interface
+
+```go
+type Provider interface {
+    Chat(ctx context.Context, messages []Message, opts ChatOpts) (*Response, error)
+}
+```
+
+> Consumers implement this with their LLM client (e.g., using eyrie). hawk wires eyrie as the provider via `internal/bridge/sight/bridge.go`.
+
+---
+
+## ⚡ Presets
+
+| Preset | Concerns | Speed |
+|--------|----------|:-----:|
+| 🏃 `Quick` | security, correctness | Fast |
+| 📊 `Standard` | security, correctness, style, docs | Medium |
+| 🔬 `Thorough` | all concerns | Slow |
+| 🔒 `SecurityFocus` | security, taint only | Fast |
+| 🤖 `CI` | Standard, fail on High+ | Medium |
+
+---
+
+## 🔌 MCP Server
+
+```bash
+sight mcp                                    # 📡 stdio transport
+sight mcp --transport http --addr :8080      # 🌐 HTTP transport
+```
+
+**Tools:** `sight_review` · `sight_describe` · `sight_improve` · `sight_taint`
+
+---
+
+## 🔎 Findings
+
+| Field | Description |
+|-------|-------------|
+| `Concern` | Review category (security, style, etc.) |
+| `Severity` | 🟢 Info · 🟡 Low · 🟠 Medium · 🔴 High · 🟥 Critical |
+| `File` / `Line` | Location in the diff |
+| `Message` | What was found and why |
+| `Fix` | Suggested fix |
+| `CWE` | CWE reference (security findings) |
+| `Confidence` | 0.0–1.0 score |
+| `InlineComment` | PR-ready inline comment |
+
+---
+
+## 🛡️ Static Rules + Taint Analysis
+
+**30+ built-in rules** run without LLM overhead — hardcoded secret patterns, SQL injection sinks, unsafe deserialization, etc. Fused with LLM results.
+
+**Taint analysis** (`sight taint --path .`) uses SSA-based cross-function tracking to detect source→sink data flows. Sources, sinks, and sanitizers are configurable.
diff --git a/examples/basic/main.go b/examples/basic/main.go
new file mode 100644
index 0000000..a2ea31c
--- /dev/null
+++ b/examples/basic/main.go
@@ -0,0 +1,47 @@
+package main
+
+import (
+	"context"
+	"fmt"
+	"log"
+
+	"github.com/GrayCodeAI/sight"
+)
+
+type mockProvider struct{}
+
+func (m *mockProvider) Complete(ctx context.Context, messages []sight.Message) (string, error) {
+	return "Code looks good. Consider adding error handling for edge cases.", nil
+}
+
+func main() {
+	diff := `diff --git a/main.go b/main.go
+--- a/main.go
++++ b/main.go
+@@ -10,6 +10,10 @@ func main() {
+ 	if err != nil {
+ 		log.Fatal(err)
+ 	}
++
++	// Process the data
++	processData(result)
++
+ 	fmt.Println("Done")
+ }
+`
+
+	reviewer := sight.NewReviewer(
+		sight.WithProvider(&mockProvider{}),
+		sight.Thorough,
+	)
+
+	result, err := reviewer.Review(context.Background(), diff)
+	if err != nil {
+		log.Fatal(err)
+	}
+
+	fmt.Printf("Found %d findings:\n", len(result.Findings))
+	for _, f := range result.Findings {
+		fmt.Printf("[%s] %s:%d - %s\n", f.Severity, f.File, f.Line, f.Message)
+	}
+}