Claude-skill-registry agent-exec

Use the agent-exec CLI to run Codex/Claude/Cursor and manage skills for this repo.

install

source · Clone the upstream repo

git clone https://github.com/majiayu000/claude-skill-registry

Claude Code · Install into ~/.claude/skills/

T=$(mktemp -d) && git clone --depth=1 https://github.com/majiayu000/claude-skill-registry "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/data/agent-exec" ~/.claude/skills/majiayu000-claude-skill-registry-agent-exec && rm -rf "$T"

manifest: skills/data/agent-exec/SKILL.md

agent-exec

Use this skill when you need to run the local agent-exec CLI, select an agent, or install skills.

When to use

Run Codex/Claude/Cursor via a single wrapper command.
Install or update skills using the open agent skills ecosystem.
Capture a JSON summary of git changes after an agent runs.

Instructions

Run agent-exec from the repo root unless a different
```
--dir
```
is required.

For normal tasks, run:

```
npx agent-exec "<prompt>"
```

If a specific agent is required:

npx agent-exec "<prompt>" --agent codex|claude|cursor

To pass through agent-specific flags:

npx agent-exec "<prompt>" -- --model ...

For skills installation, run:
- ```
npx agent-exec skills add <owner>/<repo>
```
- Add
```
-g
```
  for global installs and
```
-y
```
  to skip prompts.
JSON output uses headless defaults (codex
```
exec
```
, claude
```
-p
```
, cursor
```
--print
```
). Override with
```
AGENT_EXEC_*_ARGS
```
(set to empty to disable defaults).
Default output is JSON. Use
```
--format text
```
only if human-readable logs are explicitly needed.
Codex has an experimental "run terminal in background" capability. If it is enabled, prefer that for long-running commands so the agent can continue while the job runs.