Pi Coding Agent

Slash Commands

Command	Description
`/login`, `/logout`	Manage OAuth or API-key credentials
`/model`	Switch models
`/scoped-models`	Enable or disable models for cycling
`/settings`	Configure thinking level, theme, delivery mode, and transport
`/new`	Start a new session
`/session`	Show session file, ID, token usage, and cost
`/compact`	Compact older context, optionally with custom instructions
`/reload`	Reload keybindings, extensions, skills, prompts, and context files
`/hotkeys`	Show all keyboard shortcuts
`/quit`	Exit Pi

Skills are exposed as /skill:$NAME, and prompt templates are exposed as /$NAME.

Keyboard Shortcuts

Shortcut	Description
`Alt+Enter`	Queue a follow-up message
`Alt+Up`	Restore queued messages to the editor
`Esc`	Abort the current run
`Ctrl+P`	Cycle to the next model
`Shift+Tab`	Cycle thinking level

Keybindings can be customized in ~/.pi/agent/keybindings.json. After editing them, run /reload.

Installation

Install Pi with the official install script:

curl -fsSL https://pi.dev/install.sh | sh

Uninstall Pi:

npm uninstall -g @earendil-works/pi-coding-agent

Extensions

Pi supports extensions that can enhance functionality.

Here are some useful extensions:

pi-bar: status bar extension that shows useful information at the bottom of the interface.
pi-loadout: adds a /loadout command for choosing which tools and skills are active in the current Pi session.
pi-linter: deterministic linter that renders findings above the input bar in pi.
pi-mcp-adapter: use MCP servers with Pi.
@juicesharp/rpiv-web-tools: let the model search the web and read pages.
@juicesharp/rpiv-ask-user-question: ask clarifying questions instead of guessing.
@juicesharp/rpiv-advisor: ask a stronger model for a second opinion before it acts.
@gotgenes/pi-permission-system: provides permission gates over tool, bash, MCP, skill, and special operations.
@gotgenes/pi-subagents: gives pi a focused, in-process sub-agent core.
@gotgenes/pi-subagents-worktrees: git worktree isolation for @gotgenes/pi-subagents.
@kylebrodeur/pi-model-router: intelligent per-turn model router extension.
pi-lmstudio: integrating LM Studio with Pi, allowing you to use local LLMs.

Installation

After installation, you may need to run /reload to activate the extensions.

pi install npm:pi-bar
pi install npm:pi-loadout
pi install npm:pi-linter
pi install npm:pi-mcp-adapter
pi install npm:@juicesharp/rpiv-web-tools
pi install npm:@juicesharp/rpiv-ask-user-question
pi install npm:@juicesharp/rpiv-advisor
pi install npm:@gotgenes/pi-permission-system
pi install npm:@gotgenes/pi-subagents
pi install npm:@gotgenes/pi-subagents-worktrees
pi install npm:@kylebrodeur/pi-model-router
pi install npm:pi-lmstudio

Post Installation

Enable pi-linter rules (pi-linter):

/linter enable vague-opener
/linter enable pronoun-soup
/linter enable reactive-noop
/linter enable imperative-only
/linter enable scope-creep
/linter enable reversal
/linter enable unbounded-loop
/linter enable naked-review-paste
/linter enable review-drip

Check if all the rules have been enabled:

/linter status

Configure web searches (@juicesharp/rpiv-web-tools):

/web-tools

Configure the permission system extention (~/.pi/agent/extensions/pi-permission-system/config.json)(@gotgenes/pi-permission-system):

{
  "permission": {
    "*": "ask",
    "path": {
      "*": "allow",
      "*.env": "deny",
      "*.env.*": "deny",
      "*.env.example": "allow"
    },
    "read": "allow",
    "write": "deny",
    "edit": "deny",
    "web_search": "allow",
    "web_fetch": "ask",
    "bash": {
      "ls *": "allow",
      "ls": "allow",
      "pwd": "allow",
      "cat *": "allow",
      "head *": "allow",
      "tail *": "allow",
      "less *": "allow",
      "more *": "allow",
      "file *": "allow",
      "stat *": "allow",
      "wc *": "allow",
      "tree *": "allow",
      "tree": "allow",
      "realpath *": "allow",
      "basename *": "allow",
      "dirname *": "allow",
      "readlink *": "allow",
      "find *": "allow",
      "grep *": "allow",
      "rg *": "allow",
      "ag *": "allow",
      "fd *": "allow",
      "awk *": "allow",
      "sed *": "allow",
      "sort *": "allow",
      "uniq *": "allow",
      "cut *": "allow",
      "tr *": "allow",
      "diff *": "allow",
      "jq *": "allow",
      "yq *": "allow",
      "echo *": "allow",
      "printf *": "allow",
      "env": "allow",
      "printenv *": "allow",
      "printenv": "allow",
      "date *": "allow",
      "date": "allow",
      "whoami": "allow",
      "hostname": "allow",
      "uname *": "allow",
      "uname": "allow",
      "id": "allow",
      "uptime": "allow",
      "which *": "allow",
      "whereis *": "allow",
      "type *": "allow",
      "command -v *": "allow",
      "git *": "ask",
      "git status": "allow",
      "git status *": "allow",
      "git diff": "allow",
      "git diff *": "allow",
      "git log": "allow",
      "git log *": "allow",
      "git show": "allow",
      "git show *": "allow",
      "git branch": "allow",
      "git branch *": "allow",
      "git remote": "allow",
      "git remote *": "allow",
      "git rev-parse *": "allow",
      "git ls-files *": "allow",
      "git ls-files": "allow",
      "git ls-tree *": "allow",
      "git blame *": "allow",
      "git config --get *": "allow",
      "git config -l": "allow",
      "git tag": "allow",
      "git tag -l*": "allow",
      "git describe *": "allow",
      "git stash list": "allow",
      "git reflog": "allow",
      "git reflog *": "allow",
      "git shortlog *": "allow",
      "gh auth status": "allow",
      "gh repo view *": "allow",
      "gh pr view *": "allow",
      "gh pr list *": "allow",
      "gh issue view *": "allow",
      "gh issue list *": "allow",
      "sudo *": "ask",
      "curl * | sh": "ask",
      "curl * | bash": "ask",
      "wget * | sh": "ask",
      "wget * | bash": "ask",
      "rm -rf *": "deny",
      "rm -fr *": "deny",
      "chmod 777 *": "deny",
      ":(){ :|:& };:": "deny"
    },
    "mcp": { "mcp_status": "allow" },
    "skill": { "*": "ask" },
    "external_directory": "ask"
  }
}

Starting a Session

Start Pi in the current project directory:

pi

Run a one-shot prompt:

pi -p "Summarize this codebase"

Run a one-shot prompt with a specific model:

pi --model gemma-4-31b-qat:high -p "Commit to git. Follow conventional commit. Then push and open a PR on GitHub."

Pipe stdin into a one-shot prompt:

cat README.md | pi -p "Summarize this text"

Reference files directly on the command line:

pi @README.md "Summarize this"
pi @src/app.ts @src/app.test.ts "Review these together"

Context Files

Pi loads project instructions from:

~/.pi/agent/AGENTS.md
AGENTS.md in the current directory
AGENTS.md or CLAUDE.md in parent directories

Disable context file discovery with:

pi --no-context-files

System Prompt Files

Replace the default system prompt with:

.pi/SYSTEM.md for a project
~/.pi/agent/SYSTEM.md globally

Append to the default prompt with APPEND_SYSTEM.md in either location.

Prompt Templates

Pi loads prompt templates from:

~/.pi/agent/prompts/*.md
.pi/prompts/*.md
package prompts/ directories
--prompt-template <path>

Example .pi/prompts/review.md:

---
description: Review staged git changes
---

Review the staged changes (`git diff --cached`). Focus on:

- Bugs and logic errors
- Security issues
- Error handling gaps

This becomes the /review command inside Pi.

Skills

Pi loads skills from:

~/.pi/agent/skills/
~/.agents/skills/
.pi/skills/
.agents/skills/
--skill <path>

Skills are directories containing a SKILL.md file and are available as /skill:$NAME.

Example skill structure:

my-skill/
├── SKILL.md
├── scripts/
└── references/