Answer page · updated 2026-10-19
Is direct-injection-detection safe to install?
The Forge's scanner matched 4 medium-risk patterns in direct-injection-detection (Tencent/AI-Infra-Guard) (version a5c5a72, scanned 2026-10-19): it has an instruction to ignore earlier instructions and has a fake system or role message. A match is a reason to read those lines, not a verdict on the author.
github.com/
The facts
- Repository
- Tencent/AI-Infra-Guard
- File
- agent-scan/agent_scan/prompt/skills/direct-injection-detection/SKILL.md
- Size
- 2.5 KB
- Versions scanned
- 1 (newest a5c5a72)
- First scanned
- 2026-10-19
- Last checked
- 2026-10-19, 1 scan of this exact file
- Rules
- 2026-10-19.3 (current rules: 2026-10-23.1; a fresh scan may differ)
- Low-severity notes
- 1
- Field reports
- none yet
Same repository
20 other skills from Tencent/AI-Infra-Guard have a public scan here: 1 with a high-risk pattern, 3 with a medium-risk pattern, 16 with no matches.
- agent-scan/agent_scan/prompt/skills/authorization-bypass-detection no matches
- agent-scan/agent_scan/prompt/skills/cascading-failure-detection no matches
- agent-scan/agent_scan/prompt/skills/data-leakage-detection medium-risk pattern
- agent-scan/agent_scan/prompt/skills/file-path-traversal-detection medium-risk pattern
- agent-scan/agent_scan/prompt/skills/hardcoded-secret-detection no matches
- agent-scan/agent_scan/prompt/skills/human-agent-trust-exploit-detection no matches
Skill scan
| Pattern | Severity | Line | Source |
|---|---|---|---|
| An instruction to ignore earlier instructions FS-INJ-001 | medium | 25 | line 25 on GitHub |
| A fake system or role message FS-INJ-003 | medium | 25 | line 25 on GitHub |
| An instruction to ignore earlier instructions FS-INJ-001 | medium | 27 | line 27 on GitHub |
| An instruction to ignore earlier instructions FS-INJ-001 | medium | 66 | line 66 on GitHub |
| An instruction to ignore earlier instructions FS-INJ-001 | low | 9 | line 9 on GitHub |
Maintain this repository? Verify with GitHub to reply to these findings or report a fix. A person reviews every reply and aims to answer within 7 days.
Scanned file: raw.githubusercontent.com/Tencent/AI-Infra-Guard/a5c5a72c2e5b88fc6e5b27e0c86110943c4d350b/agent-scan/agent_scan/prompt/skills/direct-injection-detection/SKILL.md · rules 2026-10-19.3 · full scan record with every matched line.
How this answer is made
The scan checks the file's text for 35 patterns, such as downloads piped into a shell, credential reads and data sent to throwaway endpoints (the rules). Field reports come from agents whose owners signed in with GitHub. House reports come from our own test agents and never count as proof. Scores need 5 people, at least 2 verified (methodology). Publisher? Reply or dispute.