Potpie CLI
Potpie builds a context graph from code, history and team decisions and wires it into Codex or Claude Code through a skill bundle
- 5,797 ★ on the repository
- Last commit
- Apache-2.0
Extended access
Why this level
- Indexes project code and notes into local storage
- Starts a local daemon and changes the agent configuration
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/potpieDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Indexes project code and notes into local storage
- Starts a local daemon and changes the agent configuration
This is third-party code. Review the repository files before installing.
What it does
Potpie is a CLI that turns a repository and the development process into a context graph for agents. It indexes code, structure, decisions, history and team knowledge so an agent can answer questions, plan changes and debug failures with project context. The `potpie setup` command creates local config, storage, a daemon, a default graph and skills for the chosen harness. The bundle includes skills for change timelines, a repository baseline, infrastructure, project preferences and source ingestion. The `potpie ui` command opens the graph explorer.
Who it is for. Teams whose agent loses context in a large repository and knows nothing of past decisions, infrastructure or agreements.
Good fit when
- The agent proposes an edit that contradicts a decision made six months ago.
- Before a refactor you need the change timeline of a module and the reasons behind each step.
- A new developer joins and the agent has to explain structure and infrastructure.
- A failure is investigated with past incidents and accumulated notes in view.
- Requirements and documents live in Markdown, PDF and spreadsheets and the agent needs access to them.
Not a fit when
- The project is small and its whole history fits in the agent's normal context.
- Code and notes cannot be indexed even locally.
- You want a hosted service with no daemon on your own machine.
Example request
Use Potpie for this repository: show the change timeline of the payment module and list the decisions that must not be broken.Limitations
Setup starts a local daemon and storage, so persistent processes and data appear on the machine. The `potpie setup` wizard picks integrations and the harness from your answers: check which sources you connected. Answer quality depends on what is indexed. Installation needs Python and `uv` or `pip`. Version and agent compatibility are covered in the project documentation. Answers still come from the agent, so verify them against the code. Daemon behavior and supported harnesses may change between releases.
How to disable. Stop the daemon and remove the package with `uv tool uninstall potpie`. Delete the skills from the agent's skills directory, the local config and the graph storage by hand.
FAQ
How do I install Potpie?
Run `uv tool install potpie` or `python3 -m pip install --user potpie`, then `potpie setup`. The wizard creates config, a daemon, a default graph and skills for the chosen harness.
Do I need to start ingestion manually?
No. The CLI registers sources and the configured agent ingests or updates context when the task requires it.
How do I view the graph?
The `potpie ui` command opens a graph explorer in the browser.
Which skills are installed?
The bundle includes `potpie-change-timeline`, `potpie-repo-baseline`, `potpie-infra-architecture`, `potpie-project-preferences`, `potpie-debug-memory`, `potpie-source-ingestion` and skills for Markdown, PDF and spreadsheet resources.
What if the wizard picked the wrong harness?
Run `potpie setup` again and choose the right agent harness.
Discussion
No questions or reviews yet.
Related
CLI toolsCodexPotter loops Codex with a clean context each round until the code matches the goal, as a replacement for the built-in /goal.
A process kit for agents: a skill router, project memory in a DEV folder and result verification before saying done.
Knowledge-graph memory for agents: Claude Code and Codex plugins, an MCP server and a Python library
Personal Pi agent setup with four commands for plan, todos, execution and review, passing state through Markdown session files.
A memory and coordination layer for coding agents: specs and issues live in git while agents work in parallel
Compozy runs Claude Code, Codex and Cursor sessions: a task board, scheduled check-ins and a call to you only when needed.
README in short
Two-step setup: the PyPI package and the `potpie setup` wizard. Afterwards the chosen harness already knows about Potpie.