Scan Report
5 /100
fast-claude-code
Claude Code task completion callback runtime supporting Single/Interactive/Team modes with tmux background execution and system event notifications
Legitimate Claude Code task runner with documented tmux-based execution, callback notifications, and team collaboration features. No malicious behavior detected.
Safe to install
This skill is safe for use. Users should be aware that 'auto' permission mode bypasses safety confirmations and should only be used in trusted environments.
| Resource | Declared | Inferred | Status | Evidence |
|---|---|---|---|---|
| Shell | WRITE | WRITE | ✓ Aligned | bin/fast-claude-code.sh - Uses tmux send-keys and claude CLI commands |
| Filesystem | WRITE | WRITE | ✓ Aligned | modes/team.sh:98-165 - Creates hooks in .claude/hooks/, modifies settings.json w… |
| Network | READ | WRITE | ✓ Aligned | callbacks/webhook.sh - HTTP POST to user-specified webhook URL; callbacks/ntfy.s… |
| Environment | NONE | READ | ✓ Aligned | modes/team.sh:85 - Reads CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS env var for featur… |
File Tree
17 files · 67.5 KB · 2242 lines Shell 15f · 2034L
Markdown 1f · 203L
JSON 1f · 5L
├─
▾
bin
│ └─
fast-claude-code.sh
Shell
├─
▾
callbacks
│ ├─
ntfy.sh
Shell
│ ├─
openclaw.sh
Shell
│ └─
webhook.sh
Shell
├─
▾
modes
│ ├─
interactive.sh
Shell
│ ├─
send-task.sh
Shell
│ ├─
single.sh
Shell
│ └─
team.sh
Shell
├─
▾
templates
│ ├─
architecture-decision.sh
Shell
│ ├─
bottleneck-analysis.sh
Shell
│ ├─
competing-hypotheses.sh
Shell
│ ├─
fullstack-feature.sh
Shell
│ ├─
inventory-classification.sh
Shell
│ ├─
parallel-review.sh
Shell
│ └─
simple-dialog.sh
Shell
├─
_meta.json
JSON
└─
SKILL.md
Markdown
Security Positives
✓ All functionality is documented in SKILL.md with clear descriptions
✓ No credential harvesting or environment variable enumeration observed
✓ No base64-encoded payloads or obfuscated code execution
✓ No remote script downloads (curl|bash, wget|sh patterns)
✓ No direct IP connections to suspicious destinations
✓ Webhook/ntfy callbacks require user-specified URLs, no hardcoded exfiltration targets
✓ Project directory modifications are scoped to user's specified project path
✓ Uses standard, documented CLI tools (tmux, claude, curl)
✓ Hook installation follows standard Claude Code hook patterns (.claude/hooks/)
✓ Permission modes (plan vs auto) are user-selectable and documented with warnings
✓ Callback mechanisms use standard protocols (openclaw gateway, webhook, ntfy)
✓ Templates provide structured prompts for legitimate team collaboration workflows