v0.1.7

What's Changed

• Make jailbreak guardrail multi-turn by @steven10a in #42
• Add screenshot to readme by @steven10a in #44
• Dev/steven/parallel eval by @steven10a in #43
• Multi-turn Jailbreak eval results by @steven10a in #45
• Support Unicode characters in keywords check by @steven10a in #46
• Enhance URL allow list matching by @steven10a in #49
• Pass Agent level guardrails conversation history by @steven10a in #48
• Version to v0.1.7 by @gabor-openai in #50

Full Changelog: v0.1.6...v0.1.7

v0.1.6

What's Changed

• Dev/steven/pii update by @steven10a in #35
• Agent parse conversation history by @steven10a in #36
• Update SWIFT PII detection by @steven10a in #37
• Bump version to v0.1.6 by @gabor-openai in #38

Setting version to v0.1.6 to align with https://github.com/openai/openai-guardrails-python

Full Changelog: v0.1.4...v0.1.6

v0.1.4

What's Changed

• Fix custom and off topic to use llm_base by @steven10a in #31
• More robust handling of checking guardrailLlm client type by @steven10a in #32
• Remove unsupported models from docs by @steven10a in #34
• Bump version to v0.1.4 by @gabor-openai in #33

Full Changelog: v0.1.3...v0.1.4

v0.1.3

What's Changed

• Convo history examples by @steven10a in #20
• Add NSFW docs by @steven10a in #22
• Updating prompt injection sys prompt and evals by @steven10a in #24
• Support using prompt param with GuardrailAgent by @steven10a in #23
• Adding Korean RRN PII detection by @steven10a in #25
• Updating safety_identifier usage by @steven10a in #27
• Fix: Correctly use context model for moderation by @steven10a in #26
• Bump version to v0.1.3 by @gabor-openai in #28

Full Changelog: v0.1.2...v0.1.3

v0.1.2

What's Changed

• Fixing typing warnings and errors and enabling npm run lint by @cosmiccrisp in #17
• Cleaner support of conversation history by @steven10a in #18
• Add a GH Actions job for trusted publishing by @seratch in #14
• version bump v0.1.2 by @steven10a in #19

Full Changelog: v0.1.1...v0.1.2

v0.1.1

• Update PI sys prompt and new eval by @steven10a in #16
• Add safety header to LLM calls by @steven10a in #15