Skip to content

cheatsheet.md

Latest commit

 

History

History
201 lines (167 loc) · 7.86 KB

cheatsheet.md

File metadata and controls

201 lines (167 loc) · 7.86 KB
Raw

tinycode Cheat Sheet

Quick reference for the most common keyboard shortcuts, agents, and commands.

Keyboard Shortcuts

Leader key = Ctrl+X by default (500ms timeout for follow-up key)

Key Action
Ctrl+C Clear prompt / quit (if empty)
Ctrl+D Quit tinycode
Ctrl+P Command palette (commands, agents, sessions, skills)
Enter Submit prompt
Shift+Enter / Alt+Enter Insert newline in prompt
Escape Interrupt current session
Tab / Shift+Tab Cycle to next/previous agent
F2 / Shift+F2 Cycle to next/previous recent model
PgUp / PgDown Scroll chat history
↑ / ↓ Navigate prompt history (in composer)
Ctrl+R Open prompt history browser (pick a past prompt)
Ctrl+S Stash current draft (save) / restore it (press again when empty)
<leader>b Toggle session tree sidebar
<leader>o List all sessions
<leader>n Create a new session
<leader>m List available models
<leader>a List available agents
f Fork session at selected turn (in /rewind dialog)

Agents

Press Tab to cycle, or <leader>a to list. Use /ask <agent> <prompt> to invoke any agent inline.

Tab cycles build, general, ops, plan, architect, and code-reviewer. <leader>a lists the rest. code-simplifier, qa-tester, and scientist ship disabled.

Agent Use when...
build Doing ordinary coding. Small work stays here; the rest is delegated
general Asking a direct question
ops Working on a cluster, a host, or an ssh target. Read-only first
plan Writing a work plan. Edits stay in plans/ and drafts/
analyst Turning decided scope into acceptance criteria
architect Choosing a design. Known failures go to debugger
code-reviewer Reviewing a code change
critic Reviewing a plan for gaps
debugger Finding one root cause in application code
designer Building UI
document-specialist Reading external SDK docs or a changelog
executor Applying a scoped code change
explore Finding a file or symbol in this repo
git-master Committing, rebasing, or reading history
scout Reading upstream dependency source
security-reviewer Looking for exposure in source
test-engineer Writing tests
tracer Ranking competing explanations
verifier Proving a change works
writer Writing technical documentation

Skills (Slash Commands)

Type / to autocomplete. Use before or after your prompt.

Command Purpose
/debug Ask debugger for one root cause
/trace Ask tracer to rank competing explanations
/plan Ask the plan agent for a work plan
/verify Ask verifier for proof a change works
/test Ask test-engineer to write tests
/review Ask code-reviewer to review a change
/incident Triage a live system failure: impact, evidence, one next command
/change Plan one cluster or host change: the command, the check, and the undo
/host Inspect this machine, or another host over ssh
/doctor Diagnose the tinycode environment (also: tinycode doctor)
/mcp-setup Configure MCP servers
/remember Triage findings to memory surfaces
/deepinit Generate per-directory AGENTS.md files

CLI commands:

Command Purpose
tinycode run "prompt" Run a prompt non-interactively and exit
tinycode run --format json NDJSON event output (for programmatic consumers)
tinycode run --multi-turn Multi-turn mode: loop on stdin after initial prompt
tinycode run --max-iterations N Cap processor iterations per prompt
tinycode run --permissions json Programmatic permission handling via stdin/stdout
tinycode run --dangerously-skip-permissions Auto-approve all tool permissions
tinycode doctor Headless health check (config, DB, providers, agents, plugins, skills)
tinycode mcp list List configured MCP servers + connection status
tinycode mcp add NAME -- CMD… Add stdio MCP server (use --project for project config)
tinycode mcp add --transport http NAME URL Add remote streamable-http MCP server
tinycode mcp auth NAME --env VAR Set Bearer {env:VAR} (or --token)
tinycode mcp logout NAME Clear Authorization header
tinycode mcp debug NAME Handshake / tools diagnostics
tinycode plugin list List installed / available plugins
tinycode plugin install <name> Install a plugin binary (--from for a local path)
tinycode plugin uninstall <name> Remove an installed plugin

Common Workflows

Start a new conversation

<leader>n          # Create session
Type your prompt
Return             # Submit

Switch models

<leader>m          # List models
Select one

Use an agent for a specific task

/ask architect write a summary of the auth flow
/ask test-engineer write unit tests for this component

Review and refactor code

<leader>a          # Switch to code-reviewer
Paste or reference the code
Return

Debug a failing test

/ask debugger why is internal/session/processor_test.go failing?
/debug              # (if the agent found a likely cause)

Generate documentation

/ask writer        # Switch agent
Write guide for new developers on session architecture

Session navigation (vim-style hjkl)

<leader>h          # Go to previous sibling session
<leader>j          # Go to first child session
<leader>k          # Go to parent session
<leader>l          # Go to next sibling session

Scripted / CI usage (run mode)

# Single prompt, auto-approve tools
tinycode run --dangerously-skip-permissions -m ollama/qwen3:8b "fix lint errors"

# Pipe input, JSON output
echo "explain main.go" | tinycode run --format json -m ollama/qwen3:8b

# Multi-turn with iteration cap
tinycode run --multi-turn --max-iterations 20 -m ollama/qwen3:8b

Configuration

Edit ~/.config/tinycode/config.json:

{
  "model": "ollama/qwen2.5:latest"
}

LSP is enabled by default; set "lsp": false to disable.

Set leader key to something else (e.g., space):

{
  "keybinds": {
    "leader": "space"
  }
}

Model Shortcuts

After listing models (<leader>m):

  • Ctrl+F — Toggle favorite status
  • Ctrl+A — Show provider list
  • Return — Select

Tips

  • Press ? in diff viewer for more navigation shortcuts
  • Sessions form a tree — child sessions inherit context from parent
  • Use <leader>c or /compact to compact a session (runs real summarization via Processor.Compact) before exporting
  • Export to HTML with tinycode export --format html <session-id>
  • Type @filename to reference a file in your prompt
  • Use <leader>; to collapse code blocks and focus on analysis
  • Session tree shows hierarchy with <leader>b
  • Prompt stash: Ctrl+S parks your current draft (clearing the composer) so you can run a side command or wait on a permission prompt; press Ctrl+S again (with an empty composer) to restore it. The stash is in-memory only — session-local, not persisted to disk, and lost on restart — but it does survive the composer being cleared after a submit
  • Prompt history browser: Ctrl+R opens a scrollable list of past prompts (newest first) instead of relying on bare ↑/↓; a status hint below the composer also shows the entry count once you have history
  • Model warmup: On startup, tinycode probes the Ollama model to verify tool-call support and pre-load it into GPU memory. Look for "qwen3.5:9b ready — tool calling supported" in the toast/footer
  • Tool-call warnings: If you see "Multiple tool call failures detected," switch to a larger model via <leader>m — tinycode auto-repairs common JSON issues, but very small models may not support tool calling at all