npx agentspec-cli install fc32abe1-28aa-485a-9467-0c2d8e64ab29 --framework ClaudeCode→ .claude/skills/redc-useage/SKILL.md
# Redc Skills - AI Operations Guide
Multi-cloud red team infrastructure automation tool built on Terraform.
## Quick Reference
**Config:** `~/redc/config.yaml` | **Templates:** `~/redc/redc-templates/` | **Docs:** https://github.com/wgpsec/redc
## Configuration
```yaml
providers:
aws:
AWS_ACCESS_KEY_ID: "KEY"
AWS_SECRET_ACCESS_KEY: "SECRET"
region: "us-east-1"
aliyun:
ALICLOUD_ACCESS_KEY: "KEY"
ALICLOUD_SECRET_KEY: "SECRET"
region: "cn-hangzhou"
tencentcloud:
TENCENTCLOUD_SECRET_ID: "ID"
TENCENTCLOUD_SECRET_KEY: "KEY"
region: "ap-guangzhou"
```
**Env vars:** Set `AWS_ACCESS_KEY_ID`, `ALICLOUD_ACCESS_KEY`, `TENCENTCLOUD_SECRET_ID`, etc. if config.yaml unavailable.
## Global Flags
| Flag | Default | Description |
|------|---------|-------------|
| `--config` | `~/redc/config.yaml` | Config file path |
| `-u, --user` | `system` | User identifier |
| `--project` | `default` | Project name |
| `--debug` | `false` | Debug mode |
## Commands
### Setup
```bash
redc init # Initialize templates
redc pull <image> # Download template (e.g., aliyun/ecs)
redc image ls # List local templates
```
### Deploy
```bash
redc run <template> -n <name> -e key=val # Plan & start (create infrastructure)
redc plan <template> -n <name> # Plan only (preview without creating)
redc start <case-id> # Start case (create infrastructure)
```
### Manage
```bash
redc ps # List all cases
redc status <case-id> # Check status
redc stop <case-id> # Stop infrastructure
redc kill <case-id> # Stop & remove
redc rm <case-id> # Remove case
```
### Execute
```bash
redc exec <case-id> <command> # Run command
redc exec -t <case-id> bash # Interactive shell
redc cp <src> <case-id>:<dest> # Upload file
redc cp <case-id>:<src> <dest> # Download file
```
## Workflows
**Quick Deploy:**
```bash
redc pull aliyun/ecs && redc init
redc run aliyun/ecs -n myserver -e password=Pass123
# Returns case-id (e.g., 8a57078ee856)
redc exec 8a57078ee856 whoami
```
**Controlled:**
```bash
redc plan aws/ec2 -n staging
# Review the planned resources, then:
redc start <case-id>
redc cp deploy.sh <case-id>:/root/
redc exec <case-id> "bash /root/deploy.sh"
```
**Cleanup:**
```bash
redc stop <case-id> && redc rm <case-id>
# Or: redc kill <case-id>
```
## Automation
```python
import subprocess, re
def redc_run(template, name, env=None):
cmd = ["redc", "run", template, "-n", name]
if env:
for k, v in env.items():
cmd.extend(["-e", f"{k}={v}"])
result = subprocess.run(cmd, capture_output=True, text=True, check=True)
match = re.search(r'[a-f0-9]{12,64}', result.stdout)
return match.group(0) if match else None
# Usage
case_id = redc_run("aliyun/ecs", "auto_deploy", {"password": "Secure123"})
```
## Output Patterns
- **Success:** `✅` in output
- **Error:** `❌` in output
- **Case ID:** `[a-f0-9]{64}` (use first 12 chars)
- **Status:** `running`, `stopped`, `created`, `error`
## Error Handling
| Error | Solution |
|-------|----------|
| Config not found | Create `~/redc/config.yaml` |
| Template not found | Run `redc pull <template>` |
| Case ID not found | Check `redc ps` |
| SSH timeout | Verify instance running, security groups |
| Init failed | Check network, configure Terraform mirror |
## JSON Schemas
**Case:**
```json
{
"id": "string[64]",
"name": "string",
"template": "string",
"status": "running|stopped|created|error",
"outputs": {"public_ip": "string"}
}
```
**Config:**
```json
{
"providers": {
"aws": {"AWS_ACCESS_KEY_ID": "string", "region": "string"}
}
}
```
## Best Practices
- Use short IDs (first 12 chars)
- Assign meaningful names: `<project>_<purpose>_<env>`
- Always cleanup: `redc stop` → `redc rm`
- Use `--debug` for troubleshooting
- Never commit config.yaml to version control
- Monitor costs with `redc ps`
## Resources
- Repo: https://github.com/wgpsec/redc
- Templates: https://github.com/wgpsec/redc-template
- Online: https://redc.wgpsec.org/Built on the Agent Skills standard
AWS Rules
Cursor IDE rules and conventions for AI-assisted development.
by AgentHub Crawler
Project TODO: Terraform Provisioning for Strata Cloud Manager (SCM) - Initial Scope
Based on PRD.md: Terraform Provisioning for Strata Cloud Manager (SCM), with a focus on Address Objects, Address Groups, External Dynamic Lists, and Security Rules for the initial phase.
by AgentHub Crawler
Teleport18 Dev Configuration
AI coding agent instructions for project conventions and guidelines.
by AgentHub Crawler
Sign in to leave a comment.