feat: implement prompt injection detection (Issue #1979)#1998

Open

nac7 wants to merge 20 commits into

NVIDIA-NeMo:developfrom

nac7:fix/prompt-injection-detection

Commits on Jun 6, 2026

feat: implement prompt injection detection module (Issue #1979)
nac7
committed
Fix 8 code review issues for prompt injection detection (Issue #1979)

nac7
and
claude
committed
Fix: Make injection detection sensitivity configurable via RailsConfig

nac7
and
claude
committed
Fix: Remove user content from injection detection error messages

nac7
and
claude
committed
Add full Apache license header to test file

nac7
and
claude
committed
fix: apply ruff formatting and linting fixes for PR #1998
nac7
committed
docs: add prompt injection detection to CHANGELOG

nac7
and
claude
committed
fix: improve regex patterns for injection detection
nac7
committed
fix: add patterns for jailbreak and ignore safety measures
nac7
committed
fix: update codecov action to v4 to resolve GPG verification error
nac7
committed

Commits on Jun 7, 2026

fix: remove matched user input from injection error message and add coverage tests
nac7
committed

Commits on Jun 8, 2026

Commits on Jun 18, 2026

Merge remote-tracking branch 'upstream/develop' into fix/prompt-injection-detection
nac7
committed