COMcheckWeb API — Python Package

This repository contains the COMcheckWeb API Python package. It provides tools and scripts for working with the COMcheckWeb API and is intended to be used as a Python package maintained with uv for package management.

Requirements:

• Python: >=3.12

Package management:

• This project uses uv (configured in pyproject.toml) for dependency and workspace management. The tool.uv settings in pyproject.toml control workspace members and local sources.

Quickly useful notes:

• If you reference this package from other workspace packages, keep the tool.uv.sources entry that maps comcheck_api = { workspace = true } so uv resolves it to the local workspace copy.
• For a single-package repo you can also rely on members = ["."] and omit additional sources, but leaving the explicit source entry is harmless and makes intent clear.

Usage

1. Obtain an API Key

Get a Personal Access Token from the new COMcheck Web site:

1. Log in (or register a new account if you don't have one).
2. Click your username in the left-side navigation.
3. From the menu that appears, choose Settings.
4. Click Developer Setting to open the Personal Access Token page.
5. Click Generate, then immediately copy the token.

Important: the token is shown only once. Save it somewhere safe (a password manager, your .env, etc.) before leaving the page. If you lose it, generate a new one — the old one will stop working.

2. Configure the API Key

Create a .env file at the root of your project and add:

COM_API_KEY=<your-api-key-here>

The SDK does not auto-load this — you read it yourself and pass it to the client. With python-dotenv this is two lines:

import os
from dotenv import load_dotenv
from comcheck_api import COMcheckClient

load_dotenv()
client = COMcheckClient(api_key=os.environ["COM_API_KEY"])

(client.set_api_key(api_key) is the equivalent post-construction setter if you'd rather defer.)

For more detail, see the Getting Started guide.

3. Install the Package

pip install comcheck-api

4. Start Using the Package

• Simulation only: You can use the simulation features directly without creating a project on COMcheck Web.
• Updating a project: You must first create the project under your account on COMcheck Web before using this package to update it. Project creation is not yet supported through this package.

For detailed usage examples and API reference, see the documentation.

Introspection helpers

The package ships typed helpers for discovering what the SDK exposes and for validating project data — useful from notebooks, IDE plugins, and AI agents alike. All return Pydantic models; call .model_dump() when you need JSON.

import comcheck_api as cc

# What operation functions does the SDK ship?
for op in cc.list_operations():
    print(op.group, op.signature)

# What does the ComBuilding model look like?
schema = cc.lookup_type("ComBuilding")
for field in schema.fields:
    print(field.name, field.type, field.required)

# Does this dict satisfy the SDK schema?
result = cc.validate_project(project_dict)
if not result.ok:
    for err in result.errors:
        print(err.loc, err.msg)

See api/introspection in the docs for the full reference.

AI integration: the bundled Skill

A bundled Skill teaches AI coding agents how to use this SDK correctly — operation modules, default templates, the simulation polling loop, common pitfalls. The Skill folder lives at comcheck_api/ai/skill/ and ships in the wheel. It follows the open agent-skills standard (SKILL.md plus reference/ and scripts/), so the same folder works for both Claude Code and OpenAI Codex — only the install location differs.

Setup in your own repo

Run the installer once in the root of the project that consumes comcheck_api. By default it installs the Skill for both agents:

comcheck-api install-skill

This writes:

• .claude/skills/comcheck-api/ — Claude Code scans <project>/.claude/skills/ when a session opens against the repo.
• .agents/skills/comcheck-api/ — Codex scans .agents/skills from the working directory up to the repository root.

So the guidance kicks in only for projects that actually use this SDK, not on every session everywhere. Commit both folders; teammates get the same guidance the moment they open the repo, and the agent can pull in the reference docs and validate_code.py script on demand — not just the SKILL.md body.

To install for only one agent, pass --claude or --codex:

comcheck-api install-skill --claude   # Claude Code only
comcheck-api install-skill --codex    # Codex only

Re-run with --force after upgrading the package to refresh the Skill. Pass --global to install into the user-global skills dirs (~/.claude/skills/ and/or ~/.agents/skills/) for every session instead of per-project.

Development

Clone the repository and follow the commands below to set up developer tooling.

git clone https://github.com/pnnl/comcheckweb-api-python.git
cd comcheckweb-api-python
uv sync

Common commands

• Setup pre-commit (run once after cloning): uv run pre-commit install

• Fetch the latest COMcheck schema and regenerate types: ./tools/fetch_comcheck_schema.sh

• Regenerate types from existing schema (without fetching): uv run tools/generate_core_types.py

• Run a file from examples/: uv run examples/<script>

• Run tests: uv run pytest

• Format the repository with Black: uv run black comcheck_api tests examples

• Run type checking: uv run mypy comcheck_api

Running the docs locally

The documentation site is built with MkDocs (Material theme + mkdocstrings). The dependencies live in the optional docs group defined in pyproject.toml.

# Install the docs group (mkdocs, mkdocs-material, mkdocstrings).
uv sync --group docs

# Serve with live reload at http://127.0.0.1:8000
uv run mkdocs serve

# One-shot build into ./site/
uv run mkdocs build

# Fail on warnings (good before committing)
uv run mkdocs build --strict

Support

This is a publicly available library maintained by PNNL. While the code is open source and free to use, external contributions are not accepted at this time.

Reporting Issues

If you encounter bugs or have questions:

• Open an issue on GitHub for bug reports
• Check the examples/ directory for usage guidance
• Review the documentation for detailed information

Note: Issues are welcome, but pull requests from external contributors will not be accepted.

License

See the LICENSE file at the repository root for license details.