Repository files navigation

Patchwork automates development gruntwork like PR reviews, bug fixing, security patching, and more using a self-hosted CLI agent and your preferred LLMs. Try the hosted versionhere.

• Steps: Reusable atomic actions like create PR, commit changes or call an LLM.
• Prompt Templates: Customizable LLM prompts optimized for a chore like library updates, code generation, issue analysis or vulnerability remediation.
• Patchflows: LLM-assisted automations such as PR reviews, code fixing, documentation etc. built by combining steps and prompts.

Patchflows can be run locally in your CLI and IDE, or as part of your CI/CD pipeline. There areseveral patchflows available out of the box, and you can alwayscreate your own.

Patchwork is available on PyPI and can be installed using pip:

pip install'patchwork-cli[all]' --upgrade

The following optional dependency groups are available.

• security: Installssemgrep anddepscan withpip install 'patchwork-cli[security]' and is required forAutoFix andDependencyUpgrade patchflows.
• rag: Installschromadb withpip install 'patchwork-cli[rag]' and is required for theResolveIssue patchflow.
• notifications: Used by steps sending notifications, e.g. slack messages.
• all: installs everything.
• Not specifying any dependency group (pip install patchwork-cli) will install a core set of dependencies that are sufficient to run theGenerateDocstring,PRReview andGenerateREADME patchflows.

If you'd like to build from source using poetry, please see detailed documentationhere .

The CLI runs Patchflows, as follows:

patchwork <PatchFlow> <?Arguments>

Where

• Arguments: Allow for overriding default/optional attributes of the Patchflow in the format ofkey=value. Ifkey does not have any value, it is considered a booleanTrue flag.

For an AutoFix patchflow which patches vulnerabilities based on a scan using Semgrep:

patchwork AutoFix openai_api_key=<YOUR_OPENAI_API_KEY> github_api_key=<YOUR_GITHUB_TOKEN>

The above command defaults to patching code in the current directory by running Semgrep to identify the vulnerabilities. You can view thedefault.ymlfile for the list of configurations you can set to manage the AutoFix patchflow. For more details on how you can use a personal access token from GitHub on CLI, can readthis.

You can replace the OpenAI key with a key from our managed serviceby signing in athttps://app.patched.codes/signin and generating an API key from the integrations tab. You can then call the patchflow with the key as follows:

patchwork AutoFix patched_api_key=<YOUR_PATCHED_API_KEY> github_api_key=<YOUR_GITHUB_TOKEN>

To use Google's models you can set thegoogle_api_key andmodel, this is useful if you want to work with large contexts as thegemini-pro-1.5 model supports an input context length of 1 million tokens.

Thepatchwork-template repository contains the default configuration and prompts for all the patchflows. You can clone that repo and pass it as a flag to the CLI:

patchwork AutoFix --config /path/to/patchwork-configs/patchflows

Patchwork supports any OpenAI compatible endpoint, allowing use of any LLM from various providers like Groq, Together AI, or Hugging Face.

E.g. to use Llama 3.1 405B from Groq.com run:

patchwork AutoFix client_base_url=https://api.groq.com/openai/v1 openai_api_key=your_groq_key model=llama-3.1-405b-reasoning

You can also use a config file to do the same. To use Llama 3.1 405B from Hugging Face, create a config.yml file:

openai_api_key:your_hf_tokenclient_base_url:https://api-inference.huggingface.co/models/meta-llama/Meta-Llama-3.1-405B-Instruct-FP8/v1model:Meta-Llama-3.1-405B-Instruct-FP8

And run as:

patchwork AutoFix --config=/path/to/config.yml

This allows you to run local models viallama.cpp,ollama,vllm ortgi. For instance, you can run Llama 3.1 8B locally usingllama_cpp.server:

python -m llama_cpp.server --hf_model_repo_id bullerwins/Meta-Llama-3.1-8B-Instruct-GGUF --model 'Meta-Llama-3.1-8B-Instruct-Q4_K_M.gguf' --chat_format chatml

Then run your patchflow:

patchwork AutoFix client_base_url=http://localhost:8080/v1 openai_api_key=no_key_local_model

Patchwork comes with predefined patchflows, with more added over time. Sample patchflows include:

• GenerateDocstring: Generate docstrings for methods in your code.
• AutoFix: Generate and apply fixes to code vulnerabilities in a repository.
• PRReview: On PR creation, extract code diff, summarize changes, and comment on PR.
• GenerateREADME: Create a README markdown file for a given folder, to add documentation to your repository.
• DependencyUpgrade: Update your dependencies from vulnerable to fixed versions.
• ResolveIssue: Identify the files in your repository that need to be updated to resolve an issue (or bug) and create a PR to fix it.

Prompt templates are used by patchflows and passed as queries to LLMs. Templates contain prompts with placeholder variables enclosed by{{}} which are replaced by the data from the steps or inputs on every run.

Below is a sample prompt template:

{"id":"diffreview_summary","prompts": [      {"role":"user","content":"Summarize the following code change descriptions in 1 paragraph. {{diffreviews}}"      }    ]}

Each patchflow comes with an optimized default prompt template. But you can specify your own using theprompt_template_file=/path/to/prompt/template/file option.

Contributions for new patchflows and steps, or to the core framework are welcome. Please look at open issues for details.

• To create a new patchflow, followthese instructions.
• To create a new step, followthese instructions.

We also provide a chat assistant to help you create new steps and patchflows easily.

• Patchwork Assistant on HuggingChat (based on Llama-3.1)

• Expand patchflow library and integration options
• Patchflow debugger and validation module
• Bug fixing and performance improvements
• Refactor code and documentation

• Support large-scale code embeddings in patchflows
• Support parallelization and branching
• Fine-tuned models that can be self-hosted
• Open-source GUI

Patchwork is licensed underAGPL-3.0 terms. However, custom patchflows and steps can be created and shared using thepatchwork template repository which is licensed underApache-2.0 terms.