Project context carries over
Loki can retain project details and preferences between sessions, so the next session can start with context you have already provided.
The agent that evolves with you
curl -fsSL https://loki.computer/install.sh | bashhttps://github.com/wundercorp/loki
It can keep details that matter later and save recurring procedures as skills. When you return to the same project or task, you do not have to rebuild that context from scratch.
Loki can retain project details and preferences between sessions, so the next session can start with context you have already provided.
The desktop app, CLI, TUI, and supported messaging integrations share the same agent core. You can move between them without setting up a separate agent for each surface.
Loki can inspect files, run commands, use a browser, call connected tools, and hand independent parts of a job to subagents.
Loki can keep project details for later and turn recurring procedures into skills. When the same kind of work comes back, the agent can use the context and procedure you already established.
Desktop, CLI, TUI, and supported messaging integrations use the same agent core. You can move between them without creating a separate setup for each interface.
The desktop app puts chat beside files, terminals, generated artifacts, subagents, git review, memory, and voice.
Use Loki from the terminal while you work with the repository, scripts, servers, and data already on that machine.
Supported messaging integrations can send tasks to Loki while the agent runs on another machine or remote environment.
For a larger job, Loki can read project instructions, inspect files, run commands, use the browser, split independent work across subagents, and gather the results.
Checks passed, release notes updated, two issues fixed, and the final diff is ready for review.
Loki can run recurring jobs and send the results through a configured channel. Use schedules for morning briefs, PR queues, health checks, reports, maintenance, or any task you want Loki to run at a set time.
See automation guidesLoki can run locally, over SSH, in containers, or on supported remote compute. You can connect supported model providers and add skills, plugins, or MCP servers for the systems your work depends on.
Run Loki on your machine, in Docker, over SSH, or through a supported remote execution backend.
Connect supported model providers and select one based on the task, cost, or deployment requirements.
Add skills, plugins, and MCP servers to give Loki access to procedures and external systems used by your team.
Saved sessions, project context, and checkpoints let you inspect long-running work and continue it later.
Get started or clone the repo today. Become a Loki 𖤍
curl -fsSL https://loki.computer/install.sh | bash