SLM Web Agent
Autonomous browser navigation agent powered by Playwright and local small language models for headless form filling and data extraction.
01 / Overview & Capabilities
Autonomous browser navigation agent powered by Playwright and local small language models for headless form filling and data extraction.
Key Features
- Headless browser interaction via Playwright
- Automated form navigation and locator selection
- DOM snapshot reduction for small model context windows
- Zero external API calls
02 / Standalone OS Harness Installers, CLI & Python SDK
In SLMAgents 2.0, SLM Web Agent executes inside the 10-Plane Pure-CPU Agent Harness (slm_harness). Download a self-extracting Individual Agent + Harness OS Installer for macOS, Windows, or Linux (bundling slm_web_agent + slm_harness + the ./slm CLI), install the full 26-agent desktop suite, or install via Python pip:
1. Individual SLM Web Agent + 10-Plane Harness Installers (All OSes)
slm_web_agent + slm_harness + ./slm CLIslm_web_agent + slm_harnessslm_web_agent + slm_harness + ./slm CLI2. Full 26-Agent Studio & Desktop OS Installers
3. 10-Plane Harness CLI (./slm) & Python pip Installation
# Option A: Install via PyPI or editable source with the 10-Plane Harness $ pip install slm-web-agent $ pip install -e ./slm_harness -e ./slm_web_agent # Option B: Run directly through the Unified 10-Plane Pure-CPU Harness CLI (./slm) $ ./slm ask "Navigate DOM elements to search for CPU quantization benchmarks and extract top results" $ ./slm chat # Interactive Rich TUI with 18 slash commands (/trace, /models, /arena, /undo) $ ./slm serve # 100% Offline Web AI Studio + OpenAI /v1 API at http://127.0.0.1:7860
03 / 10-Plane Pure-CPU Harness Integration (slm_harness)
Rather than executing as an isolated prompt wrapper, SLM Web Agent is natively orchestrated by the SLMAgents 2.0 10-Plane Pure-CPU Harness (via slm_harness.runtime.legacy_bridge and CascadeRouter). Every call benefits from hybrid llama.cpp GGUF (Q4_K_M) + Quantized INT4/INT8 CPU execution, watermarked LRU memory governance, grammar-constrained decoding, and deterministic verification at $0 API cost:
| Harness Plane | Subsystem | How It Powers SLM Web Agent on Pure CPU |
|---|---|---|
| Plane 6 | CascadeRouter (<35ms) |
Stage 2 (DOM Action / Browser Navigation Intent) & Stage 3 (Embedding Margin) |
| Plane 2 & 5 | MemoryArena & Hybrid CPU Runtime |
Specialist Role: primary (Qwen3-4B GGUF Q4_K_M / Llama-3.2-3B INT4). Managed inside Watermark LRU MemoryArena (lite 2.6 GB / balanced 4.7 GB / power 8.2 GB) with zero-copy session sharing and automatic OOM tier step-down (8B -> 4B -> 3B -> 0.8B). |
| Plane 7 & 8 | Deterministic Verifier & Governance | Plane 2 GBNF Action-Schema Constrained Decoding (CLICK, TYPE, SCROLL, FINISH) + Plane 8 L3_NETWORK Governance Gate |
| Plane 1, 3 & 4 | Kernel, Artifact Scratchpad & 4-Tier Memory | SQLite WAL CheckpointStore (/fork, /rewind, /undo), content-addressed artifact://<sha12> handles to prevent prompt bloat, and hybrid SQLite FTS5 BM25 + HNSW vector memory. |
04 / Source Checkout (Git Sparse-Checkout)
Clone this agent alongside the 10-Plane Pure-CPU Harness (slm_harness) from the monorepo using Git sparse-checkout — no need to download all 26 agents:
Option 1 — Sparse Checkout with 10-Plane Harness (Recommended)
# 1. Create and enter a new directory $ mkdir slm_web_agent_workspace && cd slm_web_agent_workspace # 2. Initialise empty git repo and add remote $ git init $ git remote add origin https://github.com/t00114218-stack/SLMAgents.git # 3. Enable sparse-checkout for both slm_harness and slm_web_agent $ git sparse-checkout init --cone $ git sparse-checkout set slm_harness slm_web_agent # 4. Pull the 10-Plane Harness and agent source $ git pull origin main
Option 2 — Full Repository Clone
$ git clone https://github.com/t00114218-stack/SLMAgents.git $ cd SLMAgents $ ./slm doctor
Note: After checkout, install the harness and agent in editable mode with pip install -e ./slm_harness -e ./slm_web_agent or download the pre-packaged OS Harness Installer above.
05 / Configuration & API Reference
Constructor Parameters
Instantiate SLMWebAgent with performance and runtime options:
| Parameter | Type / Default | Description |
|---|---|---|
headless | bool | True | Run browser in headless background mode. Default: True. |
temperature | float | 0.0 | Action planning temperature. Default: 0.0. |
top_p | float | 0.9 | Nucleus sampling cutoff. Default: 0.9. |
max_tokens | int | 256 | Max token output limit. Default: 256. |
n_threads | int | 4 | CPU threads for model execution. Default: 4. |
Methods
| Method Signature | Return Type | Description |
|---|---|---|
browse(goal, start_url, ...) | dict | Executes multi-step Playwright web browsing tasks autonomously. |
Execution Parameters
Complete list of execution parameters accepted by the primary agent method:
| Parameter | Type / Default | Description |
|---|---|---|
goal | str | Target browser automation goal description. |
start_url | str | Initial Web URL to navigate to. |
system_prompt | str | None | Custom system instruction prompt override. Default: None. |
user_input | str | None | Optional additional user form fill values. Default: None. |
temperature | float | 0.0 | Sampling temperature. Default: 0.0. |
top_p | float | 0.9 | Nucleus sampling probability. Default: 0.9. |
max_tokens | int | 256 | Maximum token limit. Default: 256. |
Quick Start
from slm_web_agent import SLMWebAgent
web = SLMWebAgent(headless=True, n_threads=4)
res = web.browse(
goal="Navigate to portal signup and enter test email",
start_url="https://portal.slmagents.ai/signup",
system_prompt="Wait 2 seconds after submit",
user_input="Email: dev@slmagents.ai",
temperature=0.0,
top_p=0.9,
max_tokens=256
)
print(res)
06 / Verified Execution Logs
Diagnostic execution console output running locally on CPU:
→ INPUT:
Goal: Navigate to portal signup and enter test email
URL: https://portal.slmagents.ai/signup
← OUTPUT:
{
'status': '200 OK',
'goal': 'Navigate to portal signup and enter test email',
'success': True,
'steps_taken': 3
}