Skip to content

Commit 950774d

Browse files
authored
Merge pull request #20 from Soumyadas15/marketing
docs: add Claude Code skill install + first-run permission setup
2 parents ede7e6a + 118919f commit 950774d

2 files changed

Lines changed: 34 additions & 0 deletions

File tree

README.md

Lines changed: 25 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -68,6 +68,31 @@ The setup script installs Rust (if needed), builds the CLI, installs it to `~/.c
6868
cargo install --git https://github.com/kortix-ai/agent-computer-use --path cli
6969
```
7070

71+
## Use with Claude Code, Cursor, Codex, Copilot, …
72+
73+
agent-cu ships as a skill on [skills.sh](https://skills.sh). Install it and your AI agent will drive `agent-cu` directly whenever you ask it to operate a desktop app.
74+
75+
```bash
76+
npx skills add kortix-ai/agent-computer-use -a claude-code -g
77+
```
78+
79+
The `-a claude-code` flag installs for Claude Code; drop it to pick interactively from [40+ supported agents](https://skills.sh) (Cursor, Codex, Copilot, OpenCode, Cline, VS Code, etc.). `-g` installs globally — available in every project.
80+
81+
### First-run setup (optional)
82+
83+
The first time your agent runs an `agent-cu` command, Claude Code (or whichever agent you use) will ask for approval. You can:
84+
85+
- **Approve once**: pick _"Yes, and don't ask again for: `agent-cu _`"\* — saved to project settings
86+
- **Preconfigure globally**: add this to `~/.claude/settings.json`:
87+
88+
```json
89+
{
90+
"permissions": {
91+
"allow": ["Bash(agent-cu *)"]
92+
}
93+
}
94+
```
95+
7196
## Quick start
7297

7398
```bash

skills/agent-computer-use/SKILL.md

Lines changed: 9 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -12,6 +12,15 @@ metadata:
1212

1313
You have access to `agent-cu`, a CLI tool that controls desktop applications. You can click buttons, type text, read screens, scroll, drag files, move windows — all from the terminal.
1414

15+
## First-run setup (user action)
16+
17+
The first time you run an `agent-cu` command, Claude Code will prompt the user to approve it. Tell the user they have two options:
18+
19+
1. **Quick**: pick _"Yes, and don't ask again for: `agent-cu _`"\* — one-time approval, saved automatically
20+
2. **Preconfigure**: add to `~/.claude/settings.json``permissions.allow`: `"Bash(agent-cu *)"`
21+
22+
After that, all agent-cu subcommands run without prompts. Do not attempt to modify the user's settings file yourself — only the user should grant this permission.
23+
1524
## How to think
1625

1726
**Think like a human sitting at the computer.** Before you act, ask yourself: what would I see on screen? What would I click? What would I type?

0 commit comments

Comments
 (0)