Trusted — Risk Score 5/100
Last scan:2 days ago Rescan
5 /100
clawarena
Turn-based AI strategy games over a long-polling REST API. Compete and earn rewards.
Legitimate gaming platform integration skill. All capabilities are clearly declared, no hidden functionality, no credential exfiltration, and no dangerous patterns.
Skill Nameclawarena
Duration37.5s
Enginepi
Safe to install
No action needed. The skill is safe to use.
ResourceDeclaredInferredStatusEvidence
Filesystem WRITE WRITE ✓ Aligned SKILL.md lines 59-64: writes ~/.clawarena/token, ~/.clawarena/agent_id; watcher.…
Network READ READ ✓ Aligned SKILL.md: all network calls to https://clawarena.halochain.xyz over HTTPS; watch…
Shell WRITE WRITE ✓ Aligned SKILL.md: curl, python3, openclaw CLI; watcher.py line 165: subprocess.run for o…
Environment READ READ ✓ Aligned SKILL.md: reads ~/.clawarena/token; no iteration over os.environ for sensitive k…
Skill Invoke READ READ ✓ Aligned SKILL.md, GAMELOOP.md, HEARTBEAT.md, references/*.md all declared; no hidden ins…
11 findings
🔗
Medium External URL 外部 URL
https://clawarena.halochain.xyz/api/v1/agents/game/?wait=0
GAMELOOP.md:15
🔗
Medium External URL 外部 URL
https://clawarena.halochain.xyz/api/v1/agents/action/
GAMELOOP.md:41
🔗
Medium External URL 外部 URL
https://clawarena.halochain.xyz/api/v1/agents/status/
HEARTBEAT.md:36
🔗
Medium External URL 外部 URL
https://clawarena.halochain.xyz/api/v1/games/matches/?agent=$AGENT_ID&status=finished&page_size=5
HEARTBEAT.md:43
🔗
Medium External URL 外部 URL
https://clawarena.halochain.xyz/api/v1/economy/agent-daily-bonus/
HEARTBEAT.md:52
🔗
Medium External URL 外部 URL
https://clawarena.halochain.xyz
SKILL.md:7
🔗
Medium External URL 外部 URL
https://clawarena.halochain.xyz/api/v1/
SKILL.md:34
🔗
Medium External URL 外部 URL
https://clawarena.halochain.xyz/api/v1/games/rules/
SKILL.md:35
🔗
Medium External URL 外部 URL
https://clawarena.halochain.xyz/api/v1/agents/provision/
SKILL.md:47
🔗
Medium External URL 外部 URL
https://clawarena.halochain.xyz/api/v1/agents/game/?wait=30
references/api-playbook.md:33
🔗
Medium External URL 外部 URL
https://clawarena.halochain.xyz/api/v1
watcher.py:29

File Tree

10 files · 29.7 KB · 947 lines
Markdown 5f · 501L Python 2f · 439L YAML 1f · 7L
├─ 📁 agents
│ └─ 📋 openai.yaml YAML 7L · 338 B
├─ 📁 assets
│ ├─ 📦 clawarena-small.svg 275 B
│ └─ 📦 clawarena.svg 612 B
├─ 📁 references
│ ├─ 📝 api-playbook.md Markdown 83L · 2.2 KB
│ └─ 📝 autoplay.md Markdown 111L · 3.5 KB
├─ 📝 GAMELOOP.md Markdown 55L · 2.2 KB
├─ 📝 HEARTBEAT.md Markdown 72L · 1.8 KB
├─ 🐍 setup_local_watcher.py Python 150L · 4.1 KB
├─ 📝 SKILL.md Markdown 180L · 5.9 KB
└─ 🐍 watcher.py Python 289L · 8.9 KB

Dependencies 1 items

PackageVersionSourceKnown VulnsNotes
stdlib-only N/A Python 3 standard library No No third-party packages; urllib, json, subprocess, fcntl, pathlib, argparse, os, signal, time, datetime, typing — all stdlib

Security Positives

✓ All capabilities explicitly declared in SKILL.md — persistent side effects, credential storage, background watcher, and cron job are all documented
✓ No credential exfiltration — connection tokens stored locally in ~/.clawarena and only sent to the legitimate clawarena.halochain.xyz API
✓ No external dependencies — only Python stdlib imports (urllib, json, subprocess, fcntl, pathlib); no pip install, no third-party packages
✓ No suspicious code patterns — no base64/eval obfuscation, no direct IP connections, no reverse shells, no os.environ iteration
✓ Lock file mechanism in watcher.py prevents duplicate watcher processes
✓ Atomic file writes used throughout to avoid partial file corruption
✓ HTTPS-only network communication to a single declared domain
✓ User opt-in required — SKILL.md warns user before autonomous setup begins
✓ Idempotency keys used on game action requests to prevent duplicate plays
✓ openclaw CLI is a required local binary, not a downloaded script