🚀 Added
- OpenAPI 3.2 support.
- Mix schema examples into fuzzing and stateful data generation.
- Syntax-level fuzzing for JSON bodies in negative mode.
- NDJSON report format (
--report=ndjson) for exporting all engine events as newline-delimited JSON. - Capture primitive identifiers from POST/PUT responses returning bare strings or integers.
- Normalize schema names by stripping common suffixes (
-Output,-Input,Out,In,DTO). - Discover sub-resources inside array
items(e.g.,BackupFilefromAllBackups.imports[]). - Recognize
_nameand-nameparameter suffixes for resource inference (e.g.,file_name->File). - Match parameters to resources with matching suffixes or prefixes when exact match not found (e.g.,
file_name->BackupFile.name,group_slug->GroupSummary.slug).
🔧 Changed
- Show all affected operations in CLI warnings instead of truncating to 3.
- Deprioritize successfully deleted resources in variant selection to reduce wasted requests on non-existent resources.
- Generate clean HTTP headers (ASCII without control characters) 95% of the time to reduce invalid request errors.
🐛 Fixed
- CLI headers (
-H) not being passed to schema loading requests. #3440 - PCRE Unicode property escapes (
\p{L},\p{N}) in response schemas causing crashes during dependency analysis. - OpenAPI 3.1 response schemas with conflicting
anyOfarrays (e.g., differentconstvalues) causing crashes during dependency analysis. - Negative testing error descriptions displaying internal bundled refs like
#/x-bundled/schema1instead of original refs like#/components/schemas/Item. - Resource pool lookups failing due to naming mismatch between producer ($ref-based) and consumer (path-based) sides.
- "Path parameter 'X' is not defined" errors when using captured resource values for endpoints with parameters that have no resource requirements.