Jailbreak resistance tester for LLM agents: DAN, role-play, encoding bypass probes.
Wellknown found it in public sources; nobody has proven control of it yet. Claiming takes one click if the repository is under your GitHub account, or a small file on your domain otherwise. Verified owners get the badge, 15-minute checks, status alerts, edits that outrank crawled data, and a ranking boost.
Agents can do it too: POST https://wellknown.network/api/v1/claims with {"agent":"jailbreak-resistance-tester","method":"well_known_file"} — machine-readable steps at claim.json, guide at /docs/claim.
Everything here was measured by our prober or read from a registry. Nothing is self-reported.
Not distributed through a package registry we index.
| when | check | result | http | latency | detail |
|---|---|---|---|---|---|
| 7 h ago | mcp initialize | failed | 530 | 39 ms | http 530 |
| 23 h ago | mcp initialize | failed | 503 | 437 ms | http 503 |
| 31 h ago | mcp initialize | ok | 200 | 1494 ms | server llm-red-team-scanner 1.0 · protocol 2025-06-18 · 5 tools |
| 2 d ago | mcp initialize | ok | 200 | 1431 ms | server llm-red-team-scanner 1.0 · protocol 2025-06-18 · 5 tools |
| 3 d ago | mcp initialize | failed | 530 | 19 ms | http 530 |
| 3 d ago | mcp initialize | failed | 530 | 35 ms | http 530 |
No hidden text found in the tool definitions.
Text a person reading a tool definition does not see and a model does. Invisible characters can be pasted in by accident; this says what is there, not why. Covers tools only, not prompts or resources.
Attributed to the source that supplied each field. Treated as claims, not facts.
No long description beyond the summary.
Mapped onto the structured taxonomy from declared text and observed tool names. Confidence shown for derived entries.
Every source is kept verbatim. Field changes are logged as events.