promptfoo Files

0.120.5 (2025-12-16)

Features

• cli: support multiple --env-file flags (#6622) (015f2df)
• esm: add resolvePackageEntryPoint for ESM-only packages (#6586) (fbc0eca)
• providers: add gpt-image-1-mini support to OpenAI image provider (#6603) (bedc616)
• providers: update OpenAI Codex SDK to v0.65.0 with new SDK options (#6563) (fc19872)
• redteam: ability to change plugin for strategy test case generation (#6549) (ed8b268)
• redteam: add OWASP API Security Top 10 example (#6615) (8d383b5)
• redteam: add tiered severity grading for SSRF plugin (#6444) (1556df3)
• types: add pluginId and strategyId to EvaluateTableOutput (#6638) (9e2121b)
• webui: improve model audit navigation and test coverage (#5595) (55cb973)
• webui: redesign Table Settings modal with compact two-column layout (#6683) (4c0f3ef)

Bug Fixes

• assertions: handle variable declarations in single-line JavaScript assertions (#6609) (f00cf71)
• assertions: improve error messages for malformed assertion configs (#6600) (6afb4ca)
• ci: trigger code scan when PR moves from draft to ready (#6705) (790d6b5)
• cli: filter providers before instantiation (#6596) (2891658)
• cli: Increased the number of characters for displaying test variables (#5041) (70c4258)
• deps: update dependency @googleapis/sheets to v13 (#6682) (b53f2d5)
• deps: update dependency openai to ^6.10.0 (#6641) (3637b6c)
• deps: update github actions to v2 (major) (#6681) (a0434b5)
• evaluator: isolate _conversation state between scenarios (#6623) (6fb90c8)
• geval: enforce stricter output formatting in GEVAL prompts (#6632) (9432bf7)
• improving error logging for GOAT (#6486) (c0da862)
• logger: prevent "write after end" crash during shutdown (#6625) (516aa2a)
• mcp: prevent HTTP server from shutting down immediately (#6520) (f1f10c3)
• Presentation of strategy demo modal (#6675) (e1f5e68)
• providers: treat vertex MAX_TOKENS as success instead of error (#6697) (b6cf509)
• redteam: comprehensive token usage tracking across all providers (#6619) (babd51c)
• redteam: filter runtime vars and merge defaultTest.vars for accurate test matching (#6706) (783961b)
• redteam: fix infinite loop in updatePlugins by comparing output vs state (#6647) (28140c8)
• redteam: handle object output in TestCaseDialog (#6672) (33ec0ea)
• redteam: prevent infinite loop in Plugins page when selecting presets (#6626) (25f3ea6)
• restore TypeScript type checking in CI (#6652) (d0af689)
• server: avoid redundant httpServer.close warning on shutdown (#6646) (0975fc3)
• site: remove homepage code box scrollbars (#6668) (5ac8abc)
• test: make Python concurrency test environment-independent (#6660) (645b6e2)
• util: make transform.ts browser-compatible by splitting Node.js code (#6642) (8fc1f12)
• webui: include grading rubric in storedGraderResult for iterative, hydra, and custom strategies (#6696) (933a752)