FrontAgent

Enterprise-grade frontend AI coding agent and MCP-powered automation system - constrained by Specification Driven Development (SDD) for controlled planning, code generation, browser-aware execution, and repository workflows

中文文档 | Quick Start | Architecture | Design Doc

FrontAgent is an open-source AI coding agent for frontend engineering. It helps teams build, modify, validate, and ship web applications through an agentic CLI, VS Code extension, desktop app, local MCP server, RAG planning, browser-aware automation, and SDD guardrails.

Distilled Planner Models: FrontAgent's Planner stage has been distilled into Hugging Face planner assets collected under FrontAgent: Frontend Engineering Agent. Load the published adapters on their supported Qwen Coder base models to generate frontend execution plans directly, without calling large LLM APIs. The training workflow, prompts, evaluation scripts, and Hugging Face release metadata live in models/frontagent-planner.

Use FrontAgent when you need a frontend AI agent that can:

• Generate and refactor React, TypeScript, Vite, Tailwind CSS, and modern web UI code from structured execution plans.

• Run as an AI agent CLI, VS Code AI extension, desktop agent app, or stdio MCP server for Claude Desktop, Cursor, Codex, and other MCP hosts.

• Combine repository-aware RAG, Filesense navigation, facts memory, and module dependency tracking to reduce path hallucinations.

• Execute browser-aware validation, page inspection, shell commands, and git/gh repository workflows behind explicit safety controls.

• Enforce Specification Driven Development (SDD), minimal patches, self-healing error recovery, and quality gates for production frontend teams.

• Use distilled Qwen Coder planner models from Hugging Face for local or lower-cost frontend execution planning.

• ✅ Two-Stage Architecture - Separate planning and execution to avoid JSON parsing errors and enable dynamic code generation

• ✅ Phase-Based Execution - Steps grouped by phases with error recovery within each phase

• ✅ Self-Healing - Tool Error Feedback Loop automatically analyzes errors and generates fix steps

• ✅ Facts Memory - Structured facts-based context system for precise project state tracking

• ✅ Module Dependency Tracking - Automatic import/export parsing to detect path hallucinations

• ✅ Hallucination Prevention - Multi-layer hallucination detection and interception

• ✅ SDD Constraints - Specification Driven Development as hard constraints for agent behavior

• ✅ MCP Protocol - Controlled tool invocation via Model Context Protocol

• ✅ Minimal Changes - Patch-based code modifications with rollback support

• ✅ Web Awareness - Understand page structure through browser MCP

• ✅ Shell Integration - Terminal command execution (requires user approval)

• ✅ Pre-Planning Scan - Scan project structure before planning to generate accurate file paths

• ✅ Auto Port Detection - Automatically detect dev server ports from config files

• ✅ Remote Hybrid RAG - Full-repository indexing with submodule exclusion, combining BM25 keyword search and embedding-based semantic search

• ✅ Filesense Navigation - Budgeted current-repository navigation with generated schemas and notes

• ✅ LangGraph Engine (Optional) - Switchable graph-based execution engine with optional checkpoints

• ✅ Planner Skills Layer - Reusable planning skills for task decomposition and phase injection

• ✅ Distilled Planner Assets - Repository-native training, evaluation, and release assets for the Hugging Face Planner model collection

• ✅ Skill Lab - Benchmark, improve, and promote content skills with local eval suites

• ✅ VS Code Sidebar - Marketplace extension with task runs, SDD helpers, secure settings, and run logs

• ✅ OSS Harness - Local contract, quality, GitNexus, and workflow gates for maintainer-friendly changes

• ✅ Repository Management Phase - Auto git/gh workflow after acceptance (commit, push, PR)

• ✅ Cross-Session Memory - Four-phase memory system (preload, runtime recall, post-task persistence, structured storage) that persists project facts, error resolutions, and dependency state across runs

Current Release Snapshot

The repository is currently aligned on frontagent@2.1.1 for both the npm CLI package and the VS Code extension.

• Runtime requirements: Node.js >=20.0.0; VS Code extension engine ^1.120.0.
• Build output: pnpm build builds the monorepo, bundles the CLI, syncs the VS Code version, and packages apps/vscode/frontagent-2.1.1.vsix.
• Quality gates: pnpm quality:predev, pnpm quality:precommit, pnpm quality:ci, and pnpm quality:local combine contract checks, linting, typechecking, tests, workflow tests, and build verification.
• v2.1.1 focus: smaller agent/executor/context/Filesense/memory/runtime/webview modules, hardened VS Code webview nonce generation, restored GitNexus contract checks, and expanded focused tests.

Three Ways to Use FrontAgent

FrontAgent supports terminal, VS Code, and standalone desktop workflows:

• CLI: use fa init, fa run, RAG commands, Skill Lab, and automation-friendly workflows directly from your terminal.
• VS Code Extension: use the FrontAgent sidebar task console to run tasks, attach the current file or selection, provide a browser URL, review phase/step progress, approve sensitive actions, initialize/validate SDD, and open run logs from inside VS Code.
• Desktop App: a standalone Electron GUI (apps/desktop) that reuses the same Node runtime spine — a Task Console for launching tasks, watching live phase/step telemetry, and approving sensitive actions, plus a Settings panel for the LLM provider/model.

Install the VS Code extension from the Marketplace by searching for FrontAgent or the extension id ceilf6.frontagent.

Desktop App

The desktop client is a sandboxed Electron window wired to the real runtime (window.frontagent → IPC → runFrontAgentTask); it persists settings under your OS user-data directory and degrades gracefully if the runtime is unavailable.

• Download: grab the unsigned per-platform archive frontagent-desktop-${version}-${os}-${arch}.zip from the GitHub Releases page (published automatically for each v* tag), unzip, and launch the frontagent executable. The LLM API key is read from the environment (PROVIDER_API_KEY / API_KEY); set provider/model/base URL in the in-app Settings panel.
• Build locally: pnpm --filter @frontagent/desktop package produces an unpacked app under apps/desktop/release/, and pnpm --filter @frontagent/desktop dev runs it against the Vite dev server. The release script (pnpm --filter @frontagent/desktop run release) produces the distributable zip.

Desktop archives are currently unsigned; code signing, notarization, and native installers (dmg/nsis/AppImage) are planned follow-ups.

CLI Quick Start

# 1. Install globally via npm
npm install -g frontagent
# or using pnpm
pnpm add -g frontagent
# or using yarn
yarn global add frontagent

# 2. Configure LLM (supports OpenAI and Anthropic)
# OpenAI config
export PROVIDER="openai"
export BASE_URL="https://api.openai.com/v1"
export MODEL="gpt-4"
export API_KEY="sk-..."

# Or Anthropic config
export PROVIDER="anthropic"
export BASE_URL="https://api.anthropic.com"
export MODEL="claude-sonnet-4-20250514"
export API_KEY="sk-ant-..."

# 3. Navigate to your project directory and initialize SDD
cd your-project
fa init

# 4. Let AI help you complete tasks
fa run "Create a user login page"
fa run "Optimize homepage loading performance"
fa run "Add dark mode support"
# Use LangGraph engine + checkpoint (optional)
fa run "Add route guards and open a PR" --engine langgraph --langgraph-checkpoint

MCP Server

FrontAgent can run as a local stdio MCP Server for MCP hosts such as Claude Desktop, Cursor, Codex, and other clients that can launch a command-based MCP server.

MCP mode exposes FrontAgent's upper-level agent capabilities only. It does not expose raw internal tools such as read_file, apply_patch, run_command, browser tools, or rag_query directly to the external host.

Start the Server

# Use the installed CLI
fa mcp serve

# Or run from a source checkout after pnpm build
node /absolute/path/to/FrontAgent-app/apps/cli/dist/index.js \
  mcp serve

By default, FrontAgent resolves the project root from the MCP host's workspace roots when the host exposes exactly one file root. If host roots are unavailable, it falls back to the MCP server process current working directory.

Use --project-root only when you want to pin the server to a specific project, or when the host exposes multiple workspace roots and FrontAgent cannot choose safely:

fa mcp serve --project-root /absolute/path/to/your-project

One MCP server process is bound to one resolved project root.

Useful server options:

fa mcp serve \
  --engine native \
  --security-mode balanced \
  --rag-repo https://github.com/ceilf6/Lab.git \
  --rag-branch main \
  --filesense-enabled true \
  --log-file .frontagent/runs/mcp-server.log

Host Configuration

Most MCP hosts use the same mcpServers shape. Start with the simple config:

{
  "mcpServers": {
    "frontagent": {
      "command": "fa",
      "args": ["mcp", "serve"]
    }
  }
}

If your host UI has separate fields, use:

• Command: fa
• Args: mcp, serve

Do not put fa mcp serve into the command field as one string.

If the host reports command "fa" not found or env: node: No such file or directory, the GUI host probably does not inherit your terminal shell PATH. Then use absolute paths:

which node
which fa

{
  "mcpServers": {
    "frontagent": {
      "command": "/opt/homebrew/bin/node",
      "args": ["/opt/homebrew/bin/fa", "mcp", "serve"]
    }
  }
}

If you are using a source checkout instead of a globally linked package, point node at the built CLI file after pnpm build:

{
  "mcpServers": {
    "frontagent": {
      "command": "/opt/homebrew/bin/node",
      "args": [
        "/absolute/path/to/FrontAgent-app/apps/cli/dist/index.js",
        "mcp",
        "serve"
      ]
    }
  }
}

For direct LLM fallback, pass environment variables through the host config:

{
  "mcpServers": {
    "frontagent": {
      "command": "fa",
      "args": ["mcp", "serve"],
      "env": {
        "PROVIDER": "openai",
        "BASE_URL": "https://api.openai.com/v1",
        "MODEL": "gpt-4",
        "API_KEY": "sk-..."
      }
    }
  }
}

Examples of where to put the config:

• Claude Desktop: add the server under mcpServers in claude_desktop_config.json.
• Cursor: add the server under mcpServers in your Cursor MCP config, for example .cursor/mcp.json.
• Codex or other MCP hosts: use the same command, args, and env values in the host's MCP server configuration surface.

Exposed MCP Tools

FrontAgent exposes six MCP tools:

• frontagent_status: returns project root, SDD status, visible skills, LLM backend status, RAG status, and run-log directory.
• frontagent_run_task: runs a full FrontAgent task. Inputs include task, type, files, url, sddPath, and securityMode.
• frontagent_plan_task: generates a FrontAgent execution plan without executing tools or writing files.
• frontagent_validate_sdd: validates the project SDD file.
• frontagent_list_skills: lists visible content skills.
• frontagent_init_sdd: creates an SDD template. Existing files are not overwritten unless force=true.

frontagent_run_task returns structured JSON text with:

{
  "success": true,
  "taskId": "task_...",
  "output": "...",
  "error": null,
  "duration": 1234,
  "runLogPath": "/absolute/path/.frontagent/runs/...",
  "executedStepsSummary": [],
  "securityDecisions": []
}

LLM Backend Behavior

MCP mode uses auto LLM backend selection:

1. If the host supports MCP Sampling, FrontAgent asks the host model through sampling/createMessage.
2. If Sampling is unsupported or unavailable, FrontAgent falls back to direct LLM configuration.

Direct fallback uses the same environment variables and flags as fa run:

export PROVIDER="openai"
export BASE_URL="https://api.openai.com/v1"
export MODEL="gpt-4"
export API_KEY="sk-..."

Read-only tools such as frontagent_status, frontagent_list_skills, frontagent_validate_sdd, and frontagent_init_sdd do not require LLM configuration. frontagent_run_task and frontagent_plan_task require either host Sampling support or a valid direct LLM fallback.

Security Model

MCP mode keeps FrontAgent's internal safety boundary:

• External MCP hosts cannot directly call internal file, shell, browser, or RAG tools.
• Internal file writes, shell commands, browser actions, and other side effects still go through SecurityManager.
• The default security mode is balanced.
• Because stdio MCP does not provide FrontAgent's interactive approval UI, any action that requires an ask decision fails closed.
• frontagent_init_sdd only writes SDD files inside the configured project root.

Remote RAG

FrontAgent now supports a full remote repository knowledge base flow for planning and code generation:

• It syncs the remote repository into .frontagent/rag-cache/repo
• It indexes the full repository by chunk, and automatically excludes Git submodule paths
• It runs BM25 keyword retrieval and embedding-based semantic retrieval in parallel
• It applies metadata filters to each candidate list, then fuses the ranked results
• Built indexes and embedding vectors are cached under .frontagent/rag-cache

Default knowledge source:

• Repository: https://github.com/ceilf6/Lab.git
• Source mode: git by default; when FRONTAGENT_OPENVIKING_ENDPOINT is configured FrontAgent defaults to composite (OpenViking first, Git RAG fallback)

CLI options:

fa run "Explain React setState behavior" \
  --provider openai \
  --base-url https://yunwu.ai/v1 \
  --api-key YOUR_TOKEN \
  --rag-repo https://github.com/ceilf6/Lab.git \
  --rag-branch main \
  --rag-keyword-candidates 40 \
  --rag-semantic-candidates 40 \
  --rag-keyword-weight 0.45 \
  --rag-semantic-weight 0.55

# When provider=openai, RAG embeddings inherit the same base-url/api-key by default.
# Override them only if your embedding endpoint is different.
fa run "Explain React setState behavior" \
  --provider openai \
  --base-url https://yunwu.ai/v1 \
  --api-key YOUR_TOKEN \
  --rag-embedding-model text-embedding-3-small \
  --rag-embedding-batch-size 32 \
  --rag-embedding-timeout-ms 30000

# Use Weaviate as the semantic vector store (BM25 stays local)
fa run "Explain React setState behavior" \
  --provider openai \
  --base-url https://yunwu.ai/v1 \
  --api-key YOUR_TOKEN \
  --rag-embedding-model text-embedding-3-small \
  --rag-vector-store-provider weaviate \
  --rag-weaviate-url http://127.0.0.1:8080 \
  --rag-weaviate-collection-prefix FrontAgentRagChunk \
  --rag-weaviate-batch-size 64 \
  --rag-weaviate-timeout-ms 30000

# Use OpenViking Wiki as the primary knowledge provider, with Git RAG fallback
fa run "Where is FrontAgent RAG implemented?" \
  --rag-source composite \
  --open-viking-endpoint https://openviking.example.com/query \
  --open-viking-corpus wiki \
  --open-viking-namespace docs/openviking \
  --open-viking-l1-entry docs/openviking/frontagent-l1.md

# Require OpenViking only and disable Git fallback
fa run "Where is FrontAgent RAG implemented?" \
  --rag-source openviking \
  --open-viking-endpoint https://openviking.example.com/query \
  --disable-open-viking-fallback

# Disable LLM query rewrite before retrieval
fa run "How to build a custom selector" \
  --disable-rag-query-rewrite

# Cross-encoder reranking is enabled by default after BM25 + embedding candidate retrieval
fa run "Explain React setState behavior" \
  --provider openai \
  --base-url https://yunwu.ai/v1 \
  --api-key YOUR_TOKEN \
  --rag-embedding-model text-embedding-3-small \
  --rag-reranker-model jina-reranker-v2-base-multilingual \
  --rag-reranker-base-url https://your-reranker-endpoint/v1 \
  --rag-reranker-timeout-ms 30000

# Disable reranking for a run
fa run "Explain React setState behavior" \
  --disable-rag-reranker

# Disable semantic retrieval and use BM25 only
fa run "Explain React setState behavior" \
  --disable-rag-semantic

# Disable remote RAG for a run
fa run "Create a page" --disable-rag

# Force a remote git sync before this query; by default FrontAgent reuses the local cache
fa run "Explain React setState behavior" --rag-sync-on-query

Skill Lab

FrontAgent now includes a local Skill Lab workflow for iterating on content skills under skills/.

# List visible content skills
fa skill list

# Scaffold a new content skill
fa skill scaffold pricing-audit

# Generate starter trigger evals for a skill
fa skill init-evals frontend-design

# Generate starter behavior evals (binary checks for output quality)
fa skill init-behavior-evals frontend-design

# Benchmark current trigger behavior
fa skill benchmark frontend-design

# Benchmark trigger + behavior together
fa skill benchmark frontend-design --behavior

# Generate a candidate revision and compare it against baseline
fa skill improve frontend-design

# Improve with both trigger and behavior eval suites
fa skill improve frontend-design --behavior

# Promote a candidate after review
fa skill promote frontend-design 20260331T120000Z

The current Skill Lab flow supports two eval tracks for content skills:

• Trigger evals: whether the skill activates correctly.
• Behavior evals: whether the final output quality passes binary checks.

You can run trigger-only (default) or trigger + behavior (--behavior) in benchmark/improve.

Environment variables:

export FRONTAGENT_RAG_SOURCE="composite" # git | openviking | composite
export FRONTAGENT_OPENVIKING_ENDPOINT="https://openviking.example.com/query"
export FRONTAGENT_OPENVIKING_API_KEY=""
export FRONTAGENT_OPENVIKING_CORPUS="wiki"
export FRONTAGENT_OPENVIKING_NAMESPACE="docs/openviking"
export FRONTAGENT_OPENVIKING_L1_ENTRY="docs/openviking/frontagent-l1.md"
export FRONTAGENT_OPENVIKING_TIMEOUT_MS="30000"
export FRONTAGENT_RAG_REPO="https://github.com/ceilf6/Lab.git"
export FRONTAGENT_RAG_BRANCH="main"
export FRONTAGENT_RAG_SYNC_ON_QUERY="false"
export FRONTAGENT_RAG_MAX_RESULTS="5"
export FRONTAGENT_RAG_KEYWORD_CANDIDATES="40"
export FRONTAGENT_RAG_SEMANTIC_CANDIDATES="40"
export FRONTAGENT_RAG_KEYWORD_WEIGHT="0.45"
export FRONTAGENT_RAG_SEMANTIC_WEIGHT="0.55"
export FRONTAGENT_RAG_QUERY_REWRITE_MAX_TOKENS="160"
export FRONTAGENT_RAG_QUERY_REWRITE_TEMPERATURE="0.1"
export FRONTAGENT_RAG_RERANKER_MODEL="jina-reranker-v2-base-multilingual"
export FRONTAGENT_RAG_RERANKER_BASE_URL="https://your-reranker-endpoint/v1"
export FRONTAGENT_RAG_RERANKER_API_KEY="sk-..."
export FRONTAGENT_RAG_RERANKER_CANDIDATE_COUNT="20"
export FRONTAGENT_RAG_RERANKER_MAX_DOCUMENT_CHARS="1800"
export FRONTAGENT_RAG_RERANKER_TIMEOUT_MS="30000"
export FRONTAGENT_RAG_EMBEDDING_MODEL="text-embedding-3-small"
export FRONTAGENT_RAG_EMBEDDING_BASE_URL="https://api.openai.com/v1"
export FRONTAGENT_RAG_EMBEDDING_API_KEY="sk-..."
export FRONTAGENT_RAG_EMBEDDING_DIMENSIONS=""
export FRONTAGENT_RAG_EMBEDDING_BATCH_SIZE="32"
export FRONTAGENT_RAG_EMBEDDING_TIMEOUT_MS="30000"
export FRONTAGENT_RAG_VECTOR_STORE_PROVIDER="weaviate"
export FRONTAGENT_RAG_WEAVIATE_URL="http://127.0.0.1:8080"
export FRONTAGENT_RAG_WEAVIATE_API_KEY=""
export FRONTAGENT_RAG_WEAVIATE_COLLECTION_PREFIX="FrontAgentRagChunk"
export FRONTAGENT_RAG_WEAVIATE_BATCH_SIZE="64"
export FRONTAGENT_RAG_WEAVIATE_TIMEOUT_MS="30000"

# Filesense lightweight repository navigation
export FRONTAGENT_FILESENSE_ENABLED="true"
export FRONTAGENT_FILESENSE_OUTPUT="summary"       # summary | candidates | verbose
export FRONTAGENT_FILESENSE_WRITE_MODE="cache"     # cache | workspace | none
export FRONTAGENT_FILESENSE_MAX_ENTRIES="300"
export FRONTAGENT_FILESENSE_MAX_BYTES="131072"
export FRONTAGENT_FILESENSE_TIMEOUT_MS="3000"

If provider=openai, and FRONTAGENT_RAG_EMBEDDING_BASE_URL / FRONTAGENT_RAG_EMBEDDING_API_KEY are not set, FrontAgent will reuse the LLM base-url and api-key automatically.

Main LLM sampling controls:

fa run "Explain React createElement" \
  --temperature 0.2 \
  --top-p 0.9

• --temperature is supported.
• --top-p is supported through the AI SDK call settings.
• --top-k is exposed, but only some providers/models support it. For example, Anthropic models can use it, while OpenAI-compatible chat models may ignore it as unsupported.
• repetition_penalty is not exposed yet in FrontAgent because the current AI SDK/provider stack does not provide a stable cross-provider path for it.

Before retrieval, FrontAgent now sends the user's original request through a separate LLM rewrite step to generate a more retrieval-friendly frontend search query. This rewrite uses the same provider/base-url/model/api-key as the main agent, but the rewritten query is only used for RAG and does not replace the user's original task.

After BM25 + embedding recall, FrontAgent will by default send the top candidate chunks to a reranker endpoint (/rerank, Jina/Cohere-compatible) for cross-encoder-style final ordering when reranker model/base-url/api-key are available. Use --disable-rag-reranker to turn it off for a run.

When FRONTAGENT_RAG_VECTOR_STORE_PROVIDER=weaviate, FrontAgent keeps BM25 in the local index.json, but semantic vectors are written to and queried from Weaviate instead of embeddings.json.

Filesense is used as a current-repository navigation provider. FrontAgent prefers filesense_navigate for structure/location/create/refactor preparation because it is budgeted and avoids full-repo persistent sync. filesense_sync_and_summarize remains available for explicit index maintenance, but is not the normal planning path.

Prebuilt cache bundle workflow:

• Do not commit .frontagent/rag-cache into Git
• Export a prebuilt cache bundle and upload it to GitHub Releases or object storage
• Other users can import the bundle locally before their first query

# Export the current cache directory as a distributable tar.gz bundle
fa rag export

# Export to a custom path
fa rag export --output ./artifacts/frontagent-rag-cache.tar.gz

# Import from a local file
fa rag import ./artifacts/frontagent-rag-cache.tar.gz --force

# Import from a remote URL
fa rag import https://example.com/frontagent-rag-cache.tar.gz --force

Architecture Overview

System Architecture

┌─────────────────────────────────────────────────────────────────────┐
│                         FrontAgent System                            │
├─────────────────────────────────────────────────────────────────────┤
│                                                                      │
│  ┌─────────────┐     ┌─────────────┐     ┌─────────────┐           │
│  │ User Input  │────▶│ Agent Core  │────▶│   Output    │           │
│  └─────────────┘     └──────┬──────┘     └─────────────┘           │
│                             │                                        │
│           ┌─────────────────┼─────────────────┐                     │
│           ▼                 ▼                 ▼                     │
│  ┌────────────────┐ ┌────────────┐ ┌────────────────┐              │
│  │  SDD Layer     │ │  Planner   │ │   Executor     │              │
│  │ (Constraints)  │ │ (Stage 1)  │ │   (Stage 2)    │              │
│  └───────┬────────┘ └─────┬──────┘ └───────┬────────┘              │
│          │                │                 │                       │
│          ▼                ▼                 ▼                       │
│  ┌──────────────────────────────────────────────────────┐          │
│  │           MCP Layer (Trusted Interface)               │          │
│  ├──────────────┬───────────────┬──────────────────────┤          │
│  │  MCP File    │   MCP Web     │     MCP Shell        │          │
│  └──────┬───────┴───────┬───────┴──────────┬───────────┘          │
└─────────┼───────────────┼──────────────────┼────────────────────────┘
          ▼               ▼                  ▼
   ┌──────────────┐ ┌──────────┐     ┌──────────┐
   │  File System │ │ Browser  │     │  Shell   │
   │  (Project)   │ │(Playwright)│    │Commands  │
   └──────────────┘ └──────────┘     └──────────┘

Execution Flow

User Task
   │
   ▼
┌──────────────────┐
│  Pre-Planning    │ ← Scan project structure (NEW!)
│  File Scan       │   Detect dev server port (NEW!)
└────────┬─────────┘
         │
         ▼
┌──────────────────┐
│  Planner (Stage1)│ ← SDD Constraints
└────────┬─────────┘   Project file list
         │             Dev server port
         │ Generates execution plan (with phase field)
         ▼
┌──────────────────────────────────────────┐
│           Executor (Stage 2)              │
│  ┌────────────────────────────────┐      │
│  │  Phase 1: Analysis              │      │
│  │  ├─ Step 1 ✓                  │      │
│  │  ├─ Step 2 ✗ (error)          │      │
│  │  └─ Error Recovery             │      │
│  │     ├─ Analyze error           │      │
│  │     ├─ Generate fix steps     │      │
│  │     └─ Execute fix ✓          │      │
│  └────────────────────────────────┘      │
│  ┌────────────────────────────────┐      │
│  │  Phase 2: Creation              │      │
│  │  ├─ Step 3 ✓                  │      │
│  │  └─ Step 4 ✓                  │      │
│  └────────────────────────────────┘      │
│                                           │
│  After each step:                         │
│  └─ Update Facts                          │
│     ├─ File system state                  │
│     ├─ Dependency state                   │
│     ├─ Module dependency graph            │
│     └─ Project state                      │
│                                           │
│  Phase completion validation:             │
│  └─ Check missing module references       │
│     └─ Auto-generate fix steps ✓          │
└───────────────────────────────────────────┘
         │
         ▼
    Task Complete ✓

Default phase taxonomy: Phase 1 Analysis -> Phase 2 Creation -> Phase 3 Installation -> Phase 4 Validation/Acceptance -> Phase 5 Startup -> Phase 6 Browser Validation -> Phase 7 Repository Management (git/gh)

Key Features

1. Pre-Planning Project Scan (NEW!)

Before generating the execution plan, FrontAgent now automatically scans the project structure to provide accurate file context to the LLM:

// Automatically executed before planning
const projectStructure = await scanProjectFiles();
// Returns: "Project files (245 files): src/App.tsx, src/components/Button.tsx, ..."

// LLM receives this context and generates more accurate file paths

Benefits:

• ✅ Accurate File Paths - LLM knows existing files and generates correct relative paths
• ✅ Reduced Hallucination - Fewer "file not found" errors
• ✅ Better Context - Planner understands the project structure before planning

Implementation: packages/core/src/agent/project-prescan-preparation.ts

2. Automatic Dev Server Port Detection (NEW!)

FrontAgent now automatically detects the development server port from your project configuration:

// Detection sources (in order):
// 1. vite.config.ts/js: server.port field
// 2. package.json scripts: --port or -p flags
// 3. Framework defaults: Vite (5173), Next.js (3000), CRA (3000), Angular (4200)
// 4. Fallback: 5173

const devServerPort = await detectDevServerPort();
// Used in browser navigation tasks

Benefits:

• ✅ Automatic Port Discovery - No manual port configuration needed
• ✅ Framework Awareness - Recognizes different framework defaults
• ✅ Browser Testing - Correct port used for browser validation tasks

Implementation: packages/core/src/agent/dev-server-detection.ts

3. Two-Stage Architecture

FrontAgent uses an innovative two-stage architecture that completely solves JSON parsing errors when generating large amounts of code:

Stage 1: Planner

• Input: User task + SDD constraints + project context + project file list (NEW!)
• Output: Structured execution plan (descriptions only, no code)
• Tech: Uses generateObject to produce Zod Schema-compliant JSON
• Key: No code in JSON, avoiding escape and parsing issues

{
  "summary": "Create login page",
  "steps": [
    {
      "description": "Create Login.tsx component file",
      "action": "create_file",
      "params": {
        "path": "src/pages/Login.tsx",
        "codeDescription": "Create a React component with username, password inputs and login button"
      },
      "needsCodeGeneration": true
    }
  ]
}

Stage 2: Executor

• Input: Structured execution plan
• Process: Execute each step in the plan sequentially
• Code Generation: When encountering needsCodeGeneration: true, dynamically generate code using generateText
• Tech: Use MCP tools for file operations, command execution, etc.

Advantages:

1. ✅ Completely avoid JSON parsing errors (code not in JSON)
2. ✅ Better controllability (each step validated individually)
3. ✅ Support large projects (no JSON size limit)
4. ✅ More precise code generation (based on real-time context)

4. Phase-Based Execution & Self-Healing

FrontAgent implements advanced phase-based execution and automatic error recovery:

Phase-Based Execution

The execution plan is automatically divided into multiple phases, each focused on a specific goal:

{
  "steps": [
    {
      "stepId": "step-1",
      "phase": "Analysis Phase",
      "description": "Read existing files, analyze project structure",
      "action": "read_file"
    },
    {
      "stepId": "step-2",
      "phase": "Creation Phase",
      "description": "Create new component files",
      "action": "create_file"
    },
    {
      "stepId": "step-3",
      "phase": "Installation Phase",
      "description": "Install necessary dependencies",
      "action": "run_command"
    },
    {
      "stepId": "step-4",
      "phase": "Validation Phase",
      "description": "Run tests to verify functionality",
      "action": "run_command"
    },
    {
      "stepId": "step-5",
      "phase": "Repository Management Phase",
      "description": "Commit changes, push branch, and create/update PR with gh",
      "action": "run_command"
    }
  ]
}

Advantages:

• 🎯 Clear Execution Flow - Each phase has a clear objective
• 🔄 Intra-Phase Error Recovery - Errors automatically fixed within phases
• 📊 Better Progress Tracking - Users see which phase is currently executing
• 🔀 Dependency-Aware Phase Ordering - Phase DAG scheduling reduces out-of-order skips
• 🚀 Post-Acceptance Automation - Optional repository management phase can handle git/gh flow

Tool Error Feedback Loop

When tool execution fails, the system automatically analyzes errors and generates fix steps:

// 1. Error detected
Error: Cannot apply patch: file not found in context: src/App.tsx

// 2. LLM analyzes error
{
  "canRecover": true,
  "analysis": "File src/App.tsx not read into context, need to read it first",
  "recoverySteps": [
    {
      "description": "Read src/App.tsx into context",
      "action": "read_file",
      "tool": "filesystem",
      "params": { "path": "src/App.tsx" }
    },
    {
      "description": "Reapply patch to src/App.tsx",
      "action": "apply_patch",
      "tool": "filesystem",
      "params": { /* original params */ }
    }
  ]
}

// 3. Auto-execute fix steps
// 4. Continue original flow

Features:

• 🔍 Smart Error Analysis - LLM understands error causes and finds root issues
• 🛠️ Auto-Generate Fixes - No manual intervention needed
• 📝 Common Error Patterns - Built-in handling for common errors
• ♻️ Phase-Level Recovery - Errors fixed within phases without blocking overall flow

5. LangGraph Execution Engine (NEW!)

FrontAgent now supports a switchable execution engine:

• native (default): Existing executor flow with phase DAG scheduling
• langgraph: Runs phase flow through StateGraph with optional MemorySaver checkpoint

CLI options:

# Use native engine (default)
fa run "Add login page" --engine native

# Use LangGraph engine
fa run "Add login page" --engine langgraph

# LangGraph + checkpoint + custom recovery attempts
fa run "Add login page" --engine langgraph --langgraph-checkpoint --max-recovery-attempts 5

6. Planner Skills Layer (NEW!)

FrontAgent adds a dedicated skills layer in Planner to encapsulate reusable planning logic.

• Built-in task skills: task.create, task.modify, task.query, task.debug, task.refactor, task.test
• Built-in phase skill: phase.repository-management (injects git/gh workflow after acceptance)
• Custom task skills with the same match condition override built-ins (latest registered wins)
• Executor also supports action-level skills (for params/codegen/error policy)
• Supports runtime extension via API:

import { createAgent, type TaskPlanningSkill } from "@frontagent/core";

const agent = createAgent(config);

const customSkill: TaskPlanningSkill = {
  name: "task.security-audit",
  supports: (task) =>
    task.type === "debug" && task.description.includes("security"),
  plan: ({ stepFactory }) => [
    stepFactory.createStep({
      description: "Scan for security-sensitive patterns",
      action: "search_code",
      tool: "search_code",
      params: {
        pattern: "eval|innerHTML|dangerouslySetInnerHTML",
        filePattern: "src/**/*.{ts,tsx,js,jsx}",
      },
    }),
  ],
};

agent.registerTaskSkill(customSkill);
console.log(agent.getPlannerSkillSnapshot());

agent.registerExecutorActionSkill({
  name: "action.run-command.noncritical-policy",
  action: "run_command",
  shouldSkipToolError: ({ errorMsg, params }) => {
    if (typeof params.command === "string" && params.command.includes("echo")) {
      return true;
    }
    return errorMsg.includes("already exists");
  },
});
console.log(agent.getExecutorSkillSnapshot());

7. Facts-Based Context System

Traditional agents use logs as context, leading to information redundancy and inaccuracy. FrontAgent uses a structured "facts" system:

Traditional approach (log-based):

Executed operation log:
1. Attempted to read src/App.tsx - failed
2. Attempted to create src/components/Button.tsx - success
3. Attempted to read src/App.tsx - success
4. Installed react-router-dom - success
...(lots of redundant info)

FrontAgent approach (facts-based):

## File System State

### Confirmed Existing Files:
- src/App.tsx
- src/components/Button.tsx
- package.json

### Confirmed Non-Existent Paths:
- src/pages/Login.tsx

## Dependency State

### Installed Packages:
react-router-dom, axios

### Missing Packages:
@types/node

## Created Modules

### component (3 modules):
- src/components/ui/Button.tsx (default export: Button)
- src/components/ui/Card.tsx (default export: Card)
- src/components/layout/Header.tsx (exports: Header, Navigation)

### page (2 modules):
- src/pages/HomePage.tsx (default export: HomePage)
- src/pages/LoginPage.tsx (default export: LoginPage)

### ⚠️ Missing Module References:
- src/pages/HomePage.tsx references non-existent module: ../components/ui/Spinner

## Project State
- Dev server: Running (port: 5173) ← Auto-detected!
- Build status: Success

## Recent Errors
- [apply_patch] Cannot apply patch: file not found in context

Advantages:

• 📊 Structured Information - Clear state categories (filesystem, dependencies, module graph, project state)
• 🎯 Deduplication - Automatic deduplication using Set/Map
• 💡 Context Awareness - LLM knows which files exist/don't exist
• 🔄 Real-time Updates - Auto-update facts after each tool execution
• 📉 Reduced Token Usage - Concise information reduces LLM input length
• 🔗 Module Tracking - Auto-parse import/export relationships for each created file

8. Cross-Session Memory System (NEW!)

FrontAgent now implements a four-phase memory architecture that persists knowledge across task runs, so the agent no longer starts from scratch every time.

Architecture

The memory system treats context as a time-phased pipeline:

1. Phase 1 -- Startup Preload: Load durable memories and seed ProjectFacts from the last snapshot before planning begins
2. Phase 2 -- Runtime Recall: Dynamically recall relevant memories during code generation based on file path, tags, and keywords
3. Phase 3 -- Post-Task Persistence: Extract and persist durable learnings (created files, error resolutions, dependency changes) after each task completes
4. Phase 4 -- Compaction: Deferred until multi-turn interactive mode is added

Storage Layout

<projectRoot>/.frontagent/memory/
  MEMORY.md              # Index entrypoint (concise topic list)
  topics/
    project-structure.md  # Filesystem layout, key modules
    dependencies.md       # Packages, versions, known issues
    errors.md             # Past error resolutions
  snapshots/
    facts-latest.json     # Last ProjectFacts snapshot

All memory files are human-readable Markdown (inspectable, editable). Facts snapshots use JSON for efficiency.

Prompt Separation

The system prompt is now structured into three independent zones, each with its own token budget:

1. Rules zone -- SDD constraints, behavioral instructions (immutable per task)
2. Memory zone -- Durable project knowledge loaded from .frontagent/memory/
3. Context zone -- Dynamic per-task data (files, RAG results, skills, facts)

Key Design Decisions

• Single-writer pattern: All writes go through MemoryStore to prevent conflicts
• Dedup tracking: Per-session injectedKeys set prevents re-injecting the same memory
• Budget enforcement: Configurable character limits for preload (default 8000) and per-step recall (default 2000)
• Non-blocking persistence: Memory writes run off the critical path with full error swallowing
• Backward compatible: Fresh projects with no prior memory run identically to previous behavior

Configuration

const agent = createAgent({
  // ...other config
  memory: {
    enabled: true, // default: true
    preloadBudgetChars: 8000, // max chars injected at startup
    recallBudgetChars: 2000, // max chars per code-gen recall
    maxTopicFiles: 10, // max topic files loaded at startup
  },
});

Implementation: packages/core/src/memory/

Core Modules

@frontagent/sdd - SDD Control Layer

Specification Driven Development (SDD) as hard constraints for agent behavior:

# sdd.yaml
version: "1.0"

project:
  name: "my-project"
  type: "react-spa"

tech_stack:
  framework: "react"
  version: "^18.0.0"
  language: "typescript"
  forbidden_packages:
    - "jquery"
    - "lodash"

code_quality:
  max_function_lines: 50
  max_file_lines: 300
  forbidden_patterns:
    - "any"
    - "// @ts-ignore"

modification_rules:
  protected_files:
    - "package.json"
  require_approval:
    - pattern: "src/api/*"
      reason: "API layer changes require approval"

@frontagent/mcp-file - File Operations MCP

Provides file operation MCP tools:

• read_file - Read file content
• list_directory - List directory content (supports recursion)
• create_file - Create new file (two-stage: generate code from description)
• apply_patch - Apply code patches (two-stage: generate changes from description)
• search_code - Search code
• get_ast - Get AST analysis
• rollback - Rollback changes

@frontagent/mcp-shell - Shell Command MCP

Provides terminal command execution (requires user approval):

• run_command - Execute shell commands
  • Custom working directory support
  • Timeout settings
  • User approval required before execution
  • Auto-distinguish warnings from errors
  • Use cases: npm install, git init, pnpm build, etc.

@frontagent/mcp-web - Web Awareness MCP

Provides browser interaction MCP tools:

• browser_navigate - Navigate to URL
• get_page_structure - Get page DOM structure
• get_accessibility_tree - Get accessibility tree
• get_interactive_elements - Get interactive elements
• browser_click / browser_type / browser_scroll - Page interactions
• browser_screenshot - Page screenshot
• browser_wait_for_selector - Wait for element availability

@frontagent/hallucination-guard - Hallucination Prevention

Multi-layer hallucination detection:

1. File Existence Check - Verify referenced files exist
2. Import Validity Check - Verify imports are resolvable
3. Syntax Validity Check - Verify code syntax is correct
4. SDD Compliance Check - Verify compliance with SDD constraints

Tech Stack

• Language: TypeScript
• Runtime: Node.js 20+
• Package Manager: pnpm
• MCP SDK: @modelcontextprotocol/sdk
• Browser Automation: Playwright
• AST Analysis: ts-morph
• LLM Integration: Vercel AI SDK

Directory Structure

frontagent/
├── packages/
│   ├── shared/              # Shared types and utilities
│   ├── sdd/                 # SDD control layer
│   ├── mcp-file/            # File operations MCP client
│   ├── mcp-web/             # Web awareness MCP client
│   ├── mcp-shell/           # Shell commands MCP client
│   ├── hallucination-guard/ # Hallucination prevention
│   └── core/                # Agent core (two-stage architecture)
│       └── memory/          # Cross-session memory system
├── apps/
│   └── cli/                 # CLI tool
├── examples/
│   ├── sdd-example.yaml     # SDD config example
│   └── e-commerce-frontend/ # E-commerce frontend example
└── docs/
    ├── architecture.md      # Architecture design
    └── design.md            # Original requirements

Usage Examples

Example 1: Create New Project

cd examples
fa run "Create an e-commerce frontend project using React + TypeScript + Vite + Tailwind CSS"

Agent will automatically:

1. Analyze project requirements
2. Generate execution plan
3. Create package.json and config files
4. Request to execute npm install (requires user approval)
5. Generate page components and style files

Example 2: Modify Existing Files

fa run "Modify vite.config.ts to add path alias configuration"

Agent will:

1. Read existing vite.config.ts
2. Understand current configuration
3. Generate new config code
4. Apply minimal patches

Example 3: Add New Features

fa run "Add user authentication feature, including login, registration, and token management"

Agent will:

1. Analyze existing project structure
2. Plan files to create
3. Generate auth-related components
4. Create API integration code
5. Update route configuration

Example 4: Performance Optimization

fa run "Analyze and optimize homepage loading performance"

Agent will:

1. Read relevant component code
2. Analyze performance issues
3. Propose optimization solutions
4. Implement code-level optimizations (lazy loading, code splitting, etc.)

Example 5: Auto Error Recovery

fa run "Add route configuration in App.tsx"

Execution process shows self-healing:

Phase 1: Analysis Phase
  ✓ Step 1: Read package.json

Phase 2: Creation Phase
  ✗ Step 2: Modify App.tsx
     Error: Cannot apply patch: file not found in context

  🔄 Error recovery in progress...
     Analysis: App.tsx not read into context

  ✓ Recovery Step 1: Read src/App.tsx into context
  ✓ Recovery Step 2: Reapply patch to App.tsx

Phase 3: Validation Phase
  ✓ Step 3: Run type check

✅ Task complete! Auto-fixed 1 error

Key Features:

• 🎯 Phase-Based Execution - Clear execution phases (Analysis, Creation, Validation)
• 🔄 Auto-Fix - Detected file not read, auto-insert read step
• 📊 Facts Tracking - System knows which files are read/unread
• ⚡ No Retry Needed - One-shot completion, no manual re-runs needed

Example 6: Enable LangGraph Engine

fa run "Implement user profile page and open PR" \
  --type create \
  --engine langgraph \
  --langgraph-checkpoint \
  --max-recovery-attempts 5

Notes:

• --engine langgraph enables graph-based phase orchestration
• --langgraph-checkpoint enables in-memory checkpointing for the run
• After acceptance passes, repository management phase can run git/gh actions

Environment Variables

Required Configuration

OpenAI Configuration Example

export PROVIDER="openai"
export BASE_URL="https://api.openai.com/v1"
export MODEL="gpt-4"
export API_KEY="sk-..."

Anthropic Configuration Example

export PROVIDER="anthropic"
export BASE_URL="https://api.anthropic.com"
export MODEL="claude-sonnet-4-20250514"
export API_KEY="sk-ant-..."

Development

# Development mode
pnpm dev

# Type check
pnpm typecheck

# Build
pnpm build

# Clean
pnpm clean

Roadmap

Completed ✅

• Two-stage agent architecture (Planner + Executor)
• Phase-based execution
• Tool Error Feedback Loop (self-healing)
• Facts-based context system
• Module dependency graph
• Post-generation validation
• Path hallucination detection
• Multi-LLM provider support (OpenAI, Anthropic)
• Shell command execution (with user approval)
• Dynamic code generation (avoid JSON parsing errors)
• MCP tool integration (File, Web, Shell)
• Type auto-normalization (handle LLM output uncertainty)
• Unlimited steps (support complex tasks with many steps)
• LLM schema constraint optimization (multi-strategy auto-fix, smart retry)
• Pre-planning file scan (NEW!)
• Auto dev server port detection (NEW!)
• Dependency-aware phase DAG scheduling (NEW!)
• LangGraph execution engine (optional) (NEW!)
• Repository management phase (git/gh automation) (NEW!)
• Cross-session memory system (NEW!) -- Four-phase durable memory with structured Markdown storage, runtime recall, and prompt zone separation
• Distilled Planner Models -- SFT fine-tuned from FrontAgent Planner prompts, published in the FrontAgent: Frontend Engineering Agent Hugging Face collection, with training and release assets in models/frontagent-planner
• VS Code extension -- Sidebar task console, current file/selection context, SDD commands, secure configuration, run logs, and packaged Marketplace artifact
• Local stdio MCP server -- Host-facing FrontAgent task, planning, status, skill, and SDD tools with fail-closed internal execution security
• Filesense repository navigation -- Budgeted current-repository structure lookup with generated JSON schemas and explicit cache/workspace/none write modes
• OSS Harness quality gates -- Bootstrap, local contract, precommit, CI, GitNexus, and workflow-rule checks

In Progress 🚧

• Enhanced SDD constraints (finer-grained rule control)

Planned 📋

• Memory-driven pattern learning (auto-extract coding conventions from past tasks)
• GUI agent auto-testing (Playwright-based)
• Multi-agent collaboration (decompose large tasks)
• Custom MCP server support (user-defined tools)
• Code review mode (auto-check code quality)
• Incremental update mode (only modify necessary parts)

Friendly Links

• Linux.do - Chinese AI learning and developer community.
• Aionui - Mobile remote-control UI for letting AI agents operate tasks from a phone.
• OfficeCLI - Office suite designed for AI agents.
• deepseek-pp - Browser extension for DeepSeek web conversations.
• MuseAI - Local AI companion, text adventure, and interactive fiction app.
• RedBox - Local AI creation workspace for Xiaohongshu creators.
• 1flowbase - Virtual model gateway for publishing multi-model workflows as OpenAI/Claude-compatible endpoints, with trace, token, latency, and cost visibility.

Contributing

Welcome to contribute! Submit issues, bugs, or suggestions:

1. Fork the repository
2. Create a new branch git checkout -b feature/amazing-feature
3. Commit your changes git commit -m 'Add amazing feature'
4. Push to the branch git push origin feature/amazing-feature
5. Open a Pull Request

License

MIT

Contributions are welcome! Please feel free to submit issues, bug reports, or suggestions.

1. Fork this repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

License

MIT