Chhlafiu4312/promptwall ↗★ 2

dsh-promptwall

Local prompt-injection and secret-exfiltration firewall for DeepSeek Harness 适合需要在本地对输入输出进行安全扫描、拦截可疑注入和泄露的用户。

パッケージ
dsh-promptwall
互換性
未検証
Harness ピア範囲
>=0.1.0-rc.2 <0.2.0
Cordis ピア範囲
>=4.0.1 <5
バージョン
0.1.6
ライセンス
BSD-3-Clause
最終更新
2026/09/14

インストール

$npx -p @deepseek-ai/dsh dsh plugin --profile web add github:Chhlafiu4312/promptwall

ドキュメント

README 全文を読む ↗

Configuration

FieldDefaultPurpose
enabledtrueRegister the tool and policy hooks.
injectionActionsanitizemonitor, sanitize, or block suspicious output. Dangerous and truncated output still fails closed.
suspiciousThreshold30Score that produces a suspicious verdict.
dangerousThreshold70Score that produces a dangerous verdict.
maxScanChars250000Maximum UTF-16 code units inspected per prompt or credential-bearing string; incomplete inspection fails closed.
maxJsonDepth256Maximum canonical tool-result nesting depth; exceeding it fails closed.
maxJsonNodes100000Maximum canonical JSON values inspected per tool result; exceeding it fails closed.
inspectToolOutputstrueInspect post-execution output automatically.
trustedToolspromptwall_scanExact tool names exempt from automatic reinspection.
egressActionaskoff, ask, or deny for secret-like egress arguments.
egressToolPatternscommon network namesCase-insensitive patterns identifying egress-capable tools.
rules[]Additional deterministic injection rules.
secretPatterns[]Additional credential patterns.

The complete default composition is in cordis.patch.yml. Custom rules are JavaScript regular-expression sources and should be reviewed like code.