Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.wolffi.sh/llms.txt

Use this file to discover all available pages before exploring further.

​
Built-in Capabilities

These capabilities are included in the default workspace when Wolffish creates ~/.wolffish/workspace/ on first launch.

​
shell

Type: Plugin Wraps Node.js child_process to execute shell commands. Provides the shell_exec tool with a command parameter (string) and an optional background parameter (boolean) for detached processes. Danger patterns block destructive commands like rm -rf /, sudo rm, and chmod 777. Confirm patterns require approval for operations like npm install, git push, and docker rm.

​
filesystem

Type: Plugin Wraps Node.js fs for file operations. Provides three tools:
  • file_read — Read file contents
  • file_write — Write content to a file (creates or overwrites)
  • file_patch — Apply targeted edits to an existing file

​
git

Type: Pure Skill (no plugin) Provides git instructions to the LLM via SKILL.md. The LLM uses shell_exec to run git commands. Instructions include using conventional commit format, running git status first, and showing diffs before committing.

​
introspect

Type: Plugin Self-awareness capability. Provides three tools:
  • wolffish_status — Current task counts, uptime, active providers
  • wolffish_performance — Success rates from basalganglia feedback
  • wolffish_memory — Recent episode summaries and memory stats
When you ask “how are you doing?” or “what have you been up to?”, the LLM calls these tools to read its own state.

​
Modifying Built-in Capabilities

All built-in capabilities are just folders in your workspace. You can edit their SKILL.md files to change behavior, add new danger patterns, or modify tool instructions. Changes take effect on the next message — no restart needed.
If you delete a built-in capability, Wolffish won’t recreate it automatically. Keep a backup or use git to version your workspace.