This guide is the topic map for the koder CLI. Start here when you want to know which part of the product to use, then follow the deeper guide for the exact commands and setup.
| Topic | Guide | Use It For |
|---|---|---|
| First run | Getting Started | Install Koder, configure a model, start a session, and run a safe first task. |
| Terminal usage | Interactive TUI | Prompt controls, slash commands, mentions, shell mode, history, status output, and voice input. |
| Models and config | Configuration Guide | API keys, OAuth providers, base URLs, reasoning effort, settings bundles, MCP config, and provider examples. |
| Sessions and memory | Sessions and Memory | Named sessions, goals, scheduled loops, resume, clear, compact, rewind, memory files, thinkback, and AutoDream. |
| Agents and teams | Agents and Teams | Project agents, background subagents, task_delegate, /fork, /peers, teammate modes, and shared team memory. |
| Engineering workflows | Workflows | Reviews, security review, planning, Git readiness, PR comments, GitHub Actions setup, and release notes. |
| Extension system | Skills, Plugins, and MCP | Skills, verifier skills, plugins, MCP servers, channels, and Magic Docs. |
| Permissions and data | Permissions and Privacy | Tool approvals, sandbox policy, managed settings, workspace directories, local paths, and privacy boundaries. |
| Shell sandboxing | Sandbox Guide | Backend selection, real backend status, shell isolation boundaries, backend setup, and troubleshooting. |
| Voice input | Voice Mode | Dictation controls and provider-specific speech-to-text configuration. |
| Full command list | Command Reference | Complete slash-command catalog with examples. |
| Mode | Command | When To Use It |
|---|---|---|
| Interactive | koder |
Daily coding work with live output, slash completion, file mentions, shell mode, and persistent context. |
| Single prompt | koder "fix this test" |
One task from your shell while still recording the turn in the selected session. |
| Print mode | koder --print "summarize" |
Script-friendly output. Pair with --output-format json or stream-json for automation. |
| Named session | koder -s my-project |
Keep a project or topic in a durable conversation thread. |
| Resume | koder --resume or koder --continue |
Continue a previous local session. |
| Bare mode | koder --bare |
Start without hooks, skills, plugins, MCP, memory, or project AGENTS.md context. |
Koder is organized around a few everyday jobs:
- Talk to models from OpenAI, Anthropic, Google, GitHub Copilot, OAuth-backed subscriptions, Azure, OpenRouter, and other LiteLLM providers.
- Work in a terminal TUI with streaming output, command completion, shell mode, reverse search, file mentions, status output, usage tracking, and optional voice dictation.
- Keep durable local context through SQLite-backed sessions, named sessions, session goals, exports, summaries, memory files, compaction, rewind, and thinkback.
- Track long-running work with
/goaland run recurring local prompts with/loopand/schedule. - Edit and inspect repositories with file tools, search tools, shell tools, Git helpers, review commands, and project instructions from
AGENTS.md. - Delegate work to local background agents and teams while keeping teammate state, task records, and memory under Koder-owned local paths.
- Extend the runtime with project skills, user skills, plugins, MCP servers, session channels, and Magic Docs.
- Control risk with explicit permission rules, sandbox policy, managed settings, allowed workspace roots, and local privacy diagnostics.
| I want to... | Start with |
|---|---|
| Try Koder safely in a new repository | Getting Started, then /onboarding, /status, /permissions, and a read-only prompt. |
| Understand the terminal UI | Interactive TUI, then /help, /keybindings, /theme, /voice status, and shell mode. |
| Configure a model or endpoint | Configuration Guide, then /model, /effort, /reasoning, and koder config show. |
| Keep project context across sessions | Sessions and Memory, then /session, /goal, /resume, /compact, /memory, and /thinkback. |
| Review or prepare a branch | Workflows, then /diff, /review, /security-review, /passes, and /commit. |
| Run recurring checks | Sessions and Memory, then /loop, /schedule, and /tasks. |
| Add local domain knowledge | Skills, Plugins, and MCP, then /skills, /init-verifiers, /plugin, and /mcp. |
| Run parallel investigations | Agents and Teams, then /fork, /agents, /peers, and /tasks. |
| Audit data access and command risk | Permissions and Privacy and Sandbox Guide. |
Use these commands when you are exploring a new workspace:
koder /onboarding
koder /status
koder /model
koder /files
koder /permissions
koder /skills
koder /agents
koder /goalUse these commands when you are already in a coding flow:
koder /diff
koder /review
koder /security-review
koder /commit
koder /usage
koder /memory
koder /peers
koder /loopUse these commands when you are setting up the runtime:
koder config show
koder auth login google
koder auth login claude
koder mcp list
koder plugin list
koder /voice status