CLI-Anything for Reasonix
An adapter that brings the CLI-Anything harness methodology to the Reasonix coding agent.
CLI-Anything is a methodology for making any GUI software agent-native by generating stateful CLI harnesses with REPL mode, JSON output, undo/redo, and comprehensive tests. This adapter lets Reasonix act as the builder agent.
Installation
macOS / Linux:
git clone https://github.com/HKUDS/CLI-Anything.git
cd CLI-Anything/reasonix-skill
./scripts/install.sh
Windows (PowerShell):
git clone https://github.com/HKUDS/CLI-Anything.git
cd CLI-Anything\reasonix-skill
.\scripts\install.ps1
Both scripts copy the skill to Reasonix's global skill directory at ~/.reasonix/skills/cli-anything.
Configuration Note
A full CLI-Anything build (architecture analysis, 10+ file generation, installation, and multiple test runs) typically requires 25–40 tool-call rounds. Reasonix's subagent step budget is derived from the agent.max_steps setting in reasonix.toml:
- The default
max_steps = 0means unlimited, which is recommended for this workflow. - If you have set
max_stepsto a finite value, the subagent receives half that budget (minimum 5), which may cause a build to be truncated before completion.
To ensure builds complete successfully, set agent.max_steps = 0 in your reasonix.toml, or use a sufficiently high value such as 64 or more.
Usage
After installation, use slash invocation to inline the skill into the parent Reasonix loop:
/cli-anything https://github.com/GNOME/gimp
Use run_skill when you want isolated subagent execution. The same isolation applies when Reasonix invokes the skill through its dedicated subagent path:
run_skill({ name: "cli-anything", arguments: "/path/to/software" })
What It Does
The skill follows the CLI-Anything 7-phase methodology:
- Codebase Analysis — Surveys the target software's architecture, data model, and API surface
- CLI Architecture Design — Designs command groups, state model, and output formats
- Implementation — Generates a Click-based Python CLI with REPL, JSON output, and session state
- Test Planning — Creates TEST.md with comprehensive test plan
- Test Implementation — Writes unit tests and E2E tests with real backend invocation
- Test Documentation — Runs all tests and documents results
- PyPI Packaging — Creates setup.py and verifies
pip install -e .
Example
# Build a CLI for GIMP from local source
/cli-anything /home/user/gimp
# Build from a GitHub repo
/cli-anything https://github.com/blender/blender
# Refine an existing harness
/cli-anything /home/user/gimp "Refine the existing harness for batch processing and Script-Fu filters"
Reasonix Tool Mapping
| CLI-Anything Need | Reasonix Tool |
|---|---|
| Shell execution | bash |
| File generation | write_file |
| Code editing | edit_file / multi_edit |
| Source reading | read_file |
| Codebase search | grep / glob / ls |
| Architecture analysis | mcp__codegraph__search / mcp__codegraph__context when CodeGraph is enabled |
| Documentation fetch | web_fetch |
File Structure
reasonix-skill/
├── SKILL.md # Main skill document (methodology + tool bindings)
├── README.md # This file
├── agents/
│ └── reasonix.yaml # Agent interface metadata
└── scripts/
├── install.sh # macOS/Linux installer
└── install.ps1 # Windows installer
How It Compares to Other Agent Adapters
| Feature | Claude Code | Codex | Hermes | Reasonix |
|---|---|---|---|---|
| Skill format | .claude-plugin/ |
SKILL.md + YAML |
SKILL.md + YAML |
SKILL.md + YAML |
| Shell | bash |
terminal |
terminal |
bash |
| File ops | native | execute_code |
execute_code / write_file |
write_file / edit_file / multi_edit |
| Code analysis | file reads | file reads | file reads | Optional mcp__codegraph__search / mcp__codegraph__context symbol graph |
| Install method | /plugin install |
install.sh → $CODEX_HOME/skills/ |
install.sh → $HERMES_HOME/skills/ |
install.sh → ~/.reasonix/skills/ |
Contributing
This adapter follows the same contribution pattern as the Codex and Hermes adapters. See CONTRIBUTING.md in the CLI-Anything repository.
License
Apache License 2.0 — same as CLI-Anything.