Scan Report
0 /100
silicaclaw-owner-push
Monitor SilicaClaw public broadcasts and push important updates to the owner through OpenClaw's native owner channel
The skill is a legitimate local broadcast monitoring and forwarding tool with clear safety boundaries and no malicious behavior detected.
Safe to install
No action needed. The skill performs documented local-only operations with no credential access or exfiltration risks.
| Resource | Declared | Inferred | Status | Evidence |
|---|---|---|---|---|
| Filesystem | READ | WRITE | ✓ Aligned | SKILL.md declares state file writes to ~/.openclaw/ |
| Network | READ | READ | ✓ Aligned | SKILL.md declares localhost:4310 polling only |
| Shell | WRITE | WRITE | ✓ Aligned | SKILL.md declares OWNER_FORWARD_CMD subprocess execution |
| Environment | READ | READ | ✓ Aligned | SKILL.md declares env vars for configuration only |
| Skill Invoke | NONE | NONE | — | No skill_invoke usage found |
| Clipboard | NONE | NONE | — | No clipboard access found |
| Browser | NONE | NONE | — | No browser access found |
| Database | NONE | NONE | — | No database access found |
File Tree
8 files · 26.4 KB · 870 lines JavaScript 2f · 425L
Markdown 4f · 409L
JSON 1f · 30L
YAML 1f · 6L
├─
▾
agents
│ └─
openai.yaml
YAML
├─
▾
references
│ ├─
owner-dialogue-cheatsheet-zh.md
Markdown
│ ├─
push-routing-policy.md
Markdown
│ └─
runtime-setup.md
Markdown
├─
▾
scripts
│ ├─
owner-push-forwarder.mjs
JavaScript
│ └─
send-to-owner-via-openclaw.mjs
JavaScript
├─
manifest.json
JSON
└─
SKILL.md
Markdown
Security Positives
✓ All capabilities are declared in SKILL.md - no hidden functionality
✓ Network access is strictly bounded to localhost:4310 as documented
✓ No credential harvesting or exfiltration patterns observed
✓ No base64/eval/malicious code execution patterns found
✓ No remote IP connections except documented local endpoint
✓ No sensitive path access (~/.ssh, ~/.aws, .env) observed
✓ Shell execution is controlled via documented environment variable (OWNER_FORWARD_CMD)
✓ State management uses standard workspace paths with proper file locking
✓ Filtering is text-based only, no dynamic code evaluation from messages
✓ No curl|bash, wget|sh, or other remote script execution patterns