<h1 align="center">Smart Tool Creator</h1>
<p align="center">
<a href="https://microsoft.github.io/amplifier-smart-tool-creator/">Website</a> |
<a href="https://microsoft.github.io/amplifier-smart-tools/">Smart Tools</a>
</p>
<p align="center">Turn what you know into a tool any agent can use. It builds the tool and checks it against the spec.</p>
## Quickstart
Ask your agent:
> Go to https://github.com/microsoft/amplifier-smart-tool-creator and install it.
Then describe the tool you want to build:
> Create a research notebook smart tool. Give it a topic, URLs, papers, or datasets to gather and organize in a searchable local collection. Track sources and changes over time, and let me ask questions across everything I have collected, with citations back to the original material.
For manual setup, see [Installation](#installation).
## About
Smart Tool Creator builds [smart tools](https://github.com/microsoft/amplifier-smart-tools). It scaffolds the structure the spec requires, checks a tool against the spec's conformance kit, reviews it against the parts of the spec the kit cannot decide, and adds model-backed capabilities to it.
We support [GitHub Copilot SDK](https://github.com/github/copilot-sdk), [Amplifier Agent](https://github.com/microsoft/amplifier-agent), the [OpenAI Codex SDK](https://github.com/openai/codex/tree/main/sdk/python), and the [Claude Agent SDK](https://github.com/anthropics/claude-agent-sdk-python).
## Installation
Prerequisites:
- [uv](https://docs.astral.sh/uv/getting-started/installation/) 0.9.17 or newer
- For the model-backed capabilities, one of:
- `copilot` agent provider: [GitHub CLI](https://cli.github.com/) signed in to an account with a [GitHub Copilot subscription](https://github.com/github/copilot-cli#prerequisites).
- `amplifier-agent` agent provider: the model provider's credentials, for instance `OPENAI_API_KEY` for the default `openai/...` models. See [providers](https://github.com/microsoft/amplifier-agent/blob/main/docs/providers.md).
- `codex` agent provider: the [Codex CLI](https://github.com/openai/codex) signed in with ChatGPT or an API key. See [authentication](https://developers.openai.com/codex/auth).
- `claude` agent provider: `ANTHROPIC_API_KEY`, or any of its other supported modes of [authentication](https://code.claude.com/docs/en/agent-sdk/quickstart).
- (Optional) Install the Smart Tools skill so your agent knows about Smart Tools: `npx skills add microsoft/amplifier-smart-tools`
Install the agent skill and the tool:
```bash
# Install the Agent Skill so your agent knows about it.
npx skills add microsoft/amplifier-smart-tool-creator
# Install the tool, choosing the provider(s) you want
uv tool install "amplifier-smart-tool-creator[all | copilot | amplifier-agent | codex | claude] @ git+https://github.com/microsoft/amplifier-smart-tool-creator"
# Check the install; needs no credentials
smart-tool-creator manifest
```
To use it as a library:
```bash
uv add "amplifier-smart-tool-creator[all] @ git+https://github.com/microsoft/amplifier-smart-tool-creator"
```
To update:
```bash
npx skills update smart-tool-creator # add -g for global installs
uv tool upgrade amplifier-smart-tool-creator
```
To uninstall:
```bash
npx skills remove smart-tool-creator # add -g for global installs
uv tool uninstall amplifier-smart-tool-creator
```
## Interface
```bash
# Print the tool's manifest as JSON
smart-tool-creator manifest
# Scaffold a new smart tool into ./incident-postmortem, optionally with an Agent Skill and the repository it will be pushed to. Specify a specific directory with --directory
smart-tool-creator init incident-postmortem --description "Writes, reviews, and tracks blameless postmortems from your incident platform's records" --skill --repository https://github.com/org/incident-postmortem
# Run the conformance kit against a smart tool
smart-tool-creator check-conformance --directory ./incident-postmortem
# Review a smart tool against the parts of the spec the kit cannot decide, and get suggestions (model-backed)
smart-tool-creator check-spec-adherence --directory ./incident-postmortem
# The same review through Amplifier Agent on another model
smart-tool-creator check-spec-adherence --directory ./incident-postmortem --agent-provider amplifier-agent --model anthropic/claude-opus-5-5
# Add one model-backed capability to an existing smart tool, verified against that tool's own checks (model-backed)
smart-tool-creator add-smart-capability "Given an incident id, fetch its chat transcript and alert timeline from the incident platform and draft a blameless postmortem: summary, impact, contributing factors, and action items with owners" --directory ./incident-postmortem
```
See the [CLI reference](docs/02-cli.md) for every flag and the [library reference](docs/01-library.md) for the Python surface.
## Contributing
See [docs/DEVELOPMENT.md](docs/DEVELOPMENT.md) for details on how to set up your development environment.
> [!NOTE]
> This project is not currently accepting external contributions, but we're actively working toward opening this up. We value community input and look forward to collaborating in the future. For now, feel free to fork and experiment!
Most contributions require you to agree to a
Contributor License Agreement (CLA) declaring that you have the right to, and actually do, grant us
the rights to use your contribution. For details, visit [Contributor License Agreements](https://cla.opensource.microsoft.com).
When you submit a pull request, a CLA bot will automatically determine whether you need to provide
a CLA and decorate the PR appropriately (e.g., status check, comment). Simply follow the instructions
provided by the bot. You will only need to do this once across all repos using our CLA.
This project has adopted the [Microsoft Open Source Code of Conduct](https://opensource.microsoft.com/codeofconduct/).
For more information see the [Code of Conduct FAQ](https://opensource.microsoft.com/codeofconduct/faq/) or
contact [opencode@microsoft.com](mailto:opencode@microsoft.com) with any additional questions or comments.
## Trademarks
This project may contain trademarks or logos for projects, products, or services. Authorized use of Microsoft
trademarks or logos is subject to and must follow
[Microsoft's Trademark & Brand Guidelines](https://www.microsoft.com/legal/intellectualproperty/trademarks/usage/general).
Use of Microsoft trademarks or logos in modified versions of this project must not cause confusion or imply Microsoft sponsorship.
Any use of third-party trademarks or logos are subject to those third-party's policies.