Dashboard
What the registry has found so far. Every chart counts published reports only, and the numbers refresh a few times an hour.
At a glance
Published reports
9
Repos and pastes tracked
38
Findings published
146
Scans today
0
Submissions over time
0 on 11/1URLs people asked us to check, per day, over the last 30 days.
Submissions
10/310/1711/1
Day (UTC)
Show the numbers
| Day | Submissions |
|---|---|
| 0 | |
| 0 | |
| 0 | |
| 0 | |
| 0 | |
| 0 | |
| 0 | |
| 0 | |
| 0 | |
| 1 | |
| 0 | |
| 0 | |
| 0 | |
| 0 | |
| 0 | |
| 0 | |
| 0 | |
| 0 | |
| 0 | |
| 0 | |
| 0 | |
| 0 | |
| 0 | |
| 0 | |
| 12 | |
| 23 | |
| 2 | |
| 0 | |
| 0 | |
| 0 | |
| Total | 38 |
How serious the findings are
Published reports grouped by the severity we scored them.
- Critical888.9%
- High111.1%
Show the numbers
| Severity | Reports | Share |
|---|---|---|
| Critical | 8 | 88.9% |
| High | 1 | 11.1% |
| Medium | 0 | 0% |
| Low | 0 | 0% |
| Info | 0 | 0% |
| Not scored yet | 0 | 0% |
| Total | 9 | 100% |
How the injections work
Findings grouped by the trick the hidden instructions use to get an assistant's attention.
Show the numbers
| Technique | Findings |
|---|---|
| Jailbreak / safety bypass | 56 |
| Role / persona hijack | 25 |
| Direct instruction override | 24 |
| Hidden content smuggling | 10 |
| Code execution lure | 7 |
| Encoding / obfuscation | 7 |
| Agent config poisoning | 4 |
| Data exfiltration lure | 4 |
| Everything else | 9 |
| Total | 146 |
Still up, or taken down
Whether the repo or paste behind each published report is still reachable.
- Still public9100%
- Taken down00%
- Not checked yet00%
Show the numbers
| Status | Reports | Share |
|---|---|---|
| Still public | 9 | 100% |
| Taken down | 0 | 0% |
| Not checked yet | 0 | 0% |
| Total | 9 | 100% |
Biggest projects we've reported on
Published reports, most-followed project first. Stars and forks are a snapshot from the last time we looked.
- goochbeater/spiritual-spell-red-teamingCriticalGitHubStars + forks 5,238
- greshake/llm-securityCriticalGitHubStars + forks 2,309
- taksec/prompt-injection-everywhereCriticalGitHubStars + forks 279
- invariantlabs-ai/mcp-injection-experimentsCriticalGitHubStars + forks 243
- rdwr-gabin/github-ipi-in-the-wildHighGitHubStars + forks 7
- just1njust1n/malprompt-exampleExampleCriticalGitHubStars + forks 1
- just1njust1n/agentwatchCriticalGitHubStars + forks 0
- just1njust1n/skillforgeCriticalGitHubStars + forks 0
- just1njust1n/token-trimmerCriticalGitHubStars + forks 0
| Repo or paste | Source | Severity | Stars + forks |
|---|---|---|---|
| goochbeater/spiritual-spell-red-teaming | GitHub | Critical | 5,238 |
| greshake/llm-security | GitHub | Critical | 2,309 |
| taksec/prompt-injection-everywhere | GitHub | Critical | 279 |
| invariantlabs-ai/mcp-injection-experiments | GitHub | Critical | 243 |
| rdwr-gabin/github-ipi-in-the-wild | GitHub | High | 7 |
| just1njust1n/malprompt-exampleExample | GitHub | Critical | 1 |
| just1njust1n/agentwatch | GitHub | Critical | 0 |
| just1njust1n/skillforge | GitHub | Critical | 0 |
| just1njust1n/token-trimmer | GitHub | Critical | 0 |
Which models did the reading
Tokens analysed per provider over the last 30 days. Scans and spend are in the table.
Show the numbers
| Provider | Scans | Tokens | Cost |
|---|---|---|---|
| anthropic | 29 | 10,981,555 | $18.54 |
| Total | 29 | 10,981,555 | $18.54 |