Big News: Socket raises $60M Series C at a $1B valuation to secure software supply chains for AI-driven development.Announcement
Sign In

frontagent

Package Overview
Dependencies
Maintainers
1
Versions
16
Alerts
File Explorer

Advanced tools

Socket logo

Install Socket

Detect and block malicious and high-risk dependencies

Install

frontagent - npm Package Compare versions

Comparing version
1.0.3
to
1.0.4
+14
-0
CHANGELOG.md

@@ -7,2 +7,16 @@ # Changelog

## [1.0.4] - 2026-05-19
### Added
- Added RAG query sub-stage timing in agent benchmark output and summaries.
- Added RAG query result cache hit reporting for quantitative cold/warm analysis.
### Changed
- Optimized warm RAG retrieval by reusing local knowledge-base indexes when `syncOnQuery` is disabled.
- Reused the runtime knowledge-base instance so in-process RAG query caching can take effect.
- Updated the agent-flow benchmark so `BENCH_CLEAR_CACHE=0` preserves `.frontagent` cache for warm-cache measurements.
### Fixed
- Avoided repeatedly treating warm RAG benchmark runs as cold starts by preserving the benchmark workspace cache.
## [1.0.1] - 2026-04-30

@@ -9,0 +23,0 @@

+17
-11
{
"name": "frontagent",
"version": "1.0.3",
"version": "1.0.4",
"description": "FrontAgent CLI and VS Code extension for frontend AI engineering with SDD constraints, MCP-controlled execution, and RAG planning",

@@ -20,2 +20,16 @@ "type": "module",

],
"scripts": {
"build": "turbo build && node build.mjs",
"bundle": "node build.mjs",
"dev": "turbo dev",
"lint": "turbo lint",
"typecheck": "turbo typecheck",
"test": "turbo test",
"clean": "turbo clean && rm -rf node_modules dist",
"prepublishOnly": "pnpm build",
"bench:flow": "node benchmarks/frontagent-flow-benchmark.mjs",
"bench:flow:smoke": "BENCH_MODE=smoke RUNS=3 node benchmarks/frontagent-flow-benchmark.mjs",
"bench:agent-flow": "node benchmarks/frontagent-agent-flow-benchmark.mjs",
"bench:agent-flow:smoke": "BENCH_MODE=smoke RUNS=1 node benchmarks/frontagent-agent-flow-benchmark.mjs"
},
"keywords": [

@@ -56,14 +70,6 @@ "ai-agent",

},
"packageManager": "pnpm@9.0.0",
"engines": {
"node": ">=20.0.0"
},
"scripts": {
"build": "turbo build && node build.mjs",
"bundle": "node build.mjs",
"dev": "turbo dev",
"lint": "turbo lint",
"typecheck": "turbo typecheck",
"test": "turbo test",
"clean": "turbo clean && rm -rf node_modules dist"
}
}
}

@@ -17,3 +17,3 @@ # FrontAgent

> **Distilled Planner Model**: FrontAgent's Planner stage has been distilled into a standalone small model [frontagent-planner-7B-lora](https://huggingface.co/ceilf6/frontagent-planner-7B-lora). Load the LoRA adapter on top of Qwen2.5-Coder-7B to generate frontend execution plans directly, without calling large LLM APIs.
> **Distilled Planner Model**: FrontAgent's Planner stage has been distilled into a standalone small model [frontagent-planner-7B-lora](https://huggingface.co/ceilf6/frontagent-planner-7B-lora). Load the LoRA adapter on top of Qwen2.5-Coder-7B 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](models/frontagent-planner).

@@ -36,2 +36,3 @@ - ✅ **Two-Stage Architecture** - Separate planning and execution to avoid JSON parsing errors and enable dynamic code generation

- ✅ **Planner Skills Layer** - Reusable planning skills for task decomposition and phase injection
- ✅ **Distilled Planner Assets** - Repository-native training, evaluation, and release assets for the Planner LoRA model
- ✅ **Skill Lab** - Benchmark, improve, and promote content skills with local eval suites

@@ -123,3 +124,3 @@ - ✅ **Repository Management Phase** - Auto git/gh workflow after acceptance (commit, push, PR)

Most MCP hosts use the same `mcpServers` shape. Put this JSON in the host's MCP configuration file or UI:
Most MCP hosts use the same `mcpServers` shape. Start with the simple config:

@@ -140,4 +141,16 @@ ```json

If the host cannot find `fa` on `PATH`, use the built CLI file directly:
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:
```bash
which node
which fa
```
```json

@@ -147,4 +160,21 @@ {

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

@@ -254,2 +284,3 @@ "mcp",

- 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)

@@ -288,2 +319,16 @@ CLI options:

# 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

@@ -359,2 +404,9 @@ fa run "How to build a custom selector" \

```bash
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"

@@ -382,2 +434,10 @@ export FRONTAGENT_RAG_BRANCH="main"

export FRONTAGENT_RAG_WEAVIATE_COLLECTION_PREFIX="FrontAgentRagChunk"
# 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"
```

@@ -406,2 +466,4 @@

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:

@@ -1164,3 +1226,3 @@

- [x] **Cross-session memory system** (NEW!) -- Four-phase durable memory with structured Markdown storage, runtime recall, and prompt zone separation
- [x] **Distilled Planner Model** -- SFT fine-tuned from FrontAgent Planner prompts, published as [frontagent-planner-7B-lora](https://huggingface.co/ceilf6/frontagent-planner-7B-lora) (Qwen2.5-Coder-7B + LoRA, 100% JSON validity, 100% complete plan rate)
- [x] **Distilled Planner Model** -- SFT fine-tuned from FrontAgent Planner prompts, published as [frontagent-planner-7B-lora](https://huggingface.co/ceilf6/frontagent-planner-7B-lora), with training and release assets in [models/frontagent-planner](models/frontagent-planner) (Qwen2.5-Coder-7B + LoRA, 100% JSON validity, 100% complete plan rate)

@@ -1167,0 +1229,0 @@ ### In Progress 🚧

Sorry, the diff of this file is too big to display

Sorry, the diff of this file is too big to display

Sorry, the diff of this file is too big to display

Sorry, the diff of this file is too big to display