aceteam-nodes 0.7.2

Workflow node types for aceteam-workflow-engine

aceteam-nodes

A bundle of workflow node types for aceteam-workflow-engine: LLM text generation, HTTP requests, authenticated browser fetch, and XPath extraction.

This is a node-source package, not a runner. It ships no CLI and no execution engine — it advertises its nodes through the aceteam_workflow_engine.nodes entry-point group, and an operator mounts them into an engine with wengine install aceteam-nodes. The engine then executes them.

Install into a workflow engine

Nodes are discovered through entry points and mounted by the engine operator with the wengine CLI (which drives uv under the hood). From an engine project:

# Create an engine project if you don't have one yet
wengine init  # creates engine.yaml + a pyproject.toml to install into

# Mount every node this package exposes
wengine install aceteam-nodes

# ...or mount only specific nodes
wengine install aceteam-nodes --only LLM --only APICall

wengine install runs uv add aceteam-nodes and records each node in the operator-controlled name map (engine.yaml). It reads the entry-point table from package metadata — listed below — without importing the modules. Nodes whose entry declares an extra (e.g. LLM [llm]) pull that optional dependency automatically, so wengine install aceteam-nodes --only LLM installs aceteam-nodes[llm].

Workflows then reference these nodes by their type discriminator (LLM, APICall, …). See the engine's node-distribution docs for the full install/uninstall surface and the trust model.

Direct install

To add the package (and its optional deps) to an environment directly:

pip install aceteam-nodes                  # core nodes only
pip install "aceteam-nodes[llm]"           # + LLM (aceteam-aep)
pip install "aceteam-nodes[browser-fetch]" # + BrowserFetch (Playwright)
pip install "aceteam-nodes[xpath-extract]" # + XPathExtract (lxml)

Using uv

uv add aceteam-nodes                  # core nodes only
uv add "aceteam-nodes[llm]"           # + LLM (aceteam-aep)
uv add "aceteam-nodes[browser-fetch]" # + BrowserFetch (Playwright)
uv add "aceteam-nodes[xpath-extract]" # + XPathExtract (lxml)

Available Nodes

Node	Extra	Description
LLM	llm	AI text generation via aceteam-aep (OpenAI, Anthropic, Google, xAI, Ollama)
APICall	—	HTTP requests with Jinja templating
Shell	—	Run shell commands and capture stdout/stderr (trusted engines only)
BrowserFetch	browser-fetch	Authenticated web fetch via Playwright
DiscordBotInfo	discord-bot-info	Return the authenticated Discord bot's ID and username
DiscordListChannels	discord-list-channels	List channels in a Discord server
DiscordReadMessages	discord-read-messages	Fetch recent messages from a channel (Message Content intent required)
DiscordSendMessage	discord-send	Send a message to a Discord channel
SlackListChannels	slack-list-channels	List workspace channels via conversations.list
SlackReadMessages	slack-read-messages	Fetch channel history via conversations.history
SlackSearchMessages	slack-search-messages	Search workspace messages (requires SLACK_USER_TOKEN)
SlackSendMessage	slack-send	Send a message via Slack Web API
TelegramBotInfo	telegram-bot-info	Return the authenticated Telegram bot's ID and username
TelegramListChats	telegram-list-chats	Look up metadata for a single chat via getChat
TelegramReadMessages	telegram-read-messages	Fetch recent updates via getUpdates (incompatible with webhooks)
TelegramSendMessage	telegram-send	Send a message via Telegram Bot API
XPathExtract	xpath-extract	XPath extraction over HTML/XML

Each node needing a heavy/optional dependency has one extra named after the node (llm, browser-fetch, xpath-extract), so aceteam-nodes[browser-fetch] reads as "BrowserFetch is installed." The APICall node needs no extra.

Browser profile setup

BrowserFetch drives a real Chromium profile so it can fetch authenticated pages. After installing the browser-fetch extra and Playwright's browser binaries, create the profile interactively:

# After: pip install "aceteam-nodes[browser-fetch]" && playwright install chromium
ace-browser-setup
# equivalent: uv run python scripts/browser_setup.py

Configuration

LLM provider credentials are read from the environment:

export OPENAI_API_KEY=sk-...
# or
export ANTHROPIC_API_KEY=sk-ant-...
export DISCORD_BOT_TOKEN=...   # Discord send/read nodes
export SLACK_BOT_TOKEN=...     # Slack send/read/list nodes
export SLACK_USER_TOKEN=...    # Slack search node (user token, xoxp-...)
export TELEGRAM_BOT_TOKEN=...  # Telegram send/read nodes

Development

# Setup (installs project + dev deps: pyright, pytest, ruff, aceteam-aep, playwright, lxml)
uv sync

# Test
uv run pytest
uv run pytest -x                   # stop on first failure

# Lint & format
uv run ruff check && uv run ruff format

# Type check
uv run pyright

# Build
uv build

# Release
scripts/release.sh -v vX.Y.Z --dry-run   # preview
scripts/release.sh -v vX.Y.Z -y          # publish to PyPI + GitHub

Adding a node

Add three things in lockstep:

1. The node class in src/aceteam_nodes/nodes/ (with its TYPE_INFO).
2. The export in nodes/__init__.py.
3. The export in __init__.py.
4. The entry-point line under [project.entry-points."aceteam_workflow_engine.nodes"] in pyproject.toml (append [extra_dependency_name] if it needs an optional dependency).

Related

• Workflow Engine — the DAG execution engine that discovers and runs these nodes
• aceteam-aep — multi-provider LLM abstraction used by the LLM node

License

MIT