0.9.0

Features

• Add feature to return application/adapter configuration #297 (#298)
• Adjust user limits #278 (#279)
• Implement per request API keys through stateful layer #183 (#312)
• Implement rate limits for requests #277 (#282)
• Implement user limits #82 (#188)
• Support auto-sharing of folders #265 (#272)
• add default admin access configuration (#316)
• cache public rules (#281)
• extend publications response body with resourceTypes (#318)(#315)
• implement copy shared access operation (#283)(#271)
• implement move resource operation (#273)(#227)
• implement publication deletion workflow (#302)(#295)
• implement published resource listing (#299)(#294)
• implement version endpoint (#290)(#240)
• publication part 1 (#260)
• publications rules (#269)
• support custom claims check for publications (#309)(#306)
• supported folder_attachments feature in the listing (#304)
• supported metadata attachments (#314)

Fixes

• Core accumulates a huge number of http server metrics #308 (#311)
• Support OLTP metric configuration in Vertex #287 (#288)
• clean up shared record on discard (#285)(#280)
• cleanup publication api (#284)
• google cloud storage getBlob with incorrect encoding handling (#289)
• improve resource name validation (#303)
• run processing request body async in order to prevent event thread blocking (#274)
• validate resource path according to the azure blob storage restrictions (#267)

0.8.1

Fixes

• validate resource path according to the azure blob storage restrictions (#267)

0.8.0

Features

• Implement endpoint for getting API key spent limit information #218 (#219)
• Support available credentials providers for AWS blob storage #235 (#236)
• Support available credentials providers for Azure blob storage #251 (#255)
• Support available credentials providers for GCP blob storage #239 (#245)
• support aws-s3 EC2 instance metadata authentication (#215)
• support storage prefix (#208)
• supported url_attachments feature flag for models/applications (#261)

Fixes

• Application does not terminate if severe error occurs #185 (#211)
• Core doesn't check access to file which was auto shared previously (#243)
• Errors when trying to run Core locally from compose #237 (#238)
• Model and settings in message should not be required (#257)
• Model cost calculator fails with NPE for some adapters #201 (#252)
• NPE on revoke/discard shared resource (#212)
• caller must have access to attached files from all messages in chat completion request (#232)
• clean up shared access on resource deletion (#209)
• enforce json schema for prompts and conversations (#253)(#213)
• normalize URL for comparison (#230)(#224)
• removed USER directive and added entrypoint to allow volume setup (#214)
• support prefix for resource queue (#258)(#256)
• unintentional sharing; clean up invitations and revoke shared access (#241)(#221)
• updated README with core configs settings (#234)
• workaround for jclouds s3 when blob key was cut without notice (#263)

Other

• Update github workflow reference (#248)
• add dynamic settings table to README (#205)
• bump redisson version (#246)
• documented tokenizeModel core config parameter (#244)

0.7.4

Fixes

• caller must have access to attached files from all messages in chat completion request (#232)

0.6.3

Fixes:

• verify caller has access to attached files (#231)

0.7.3

Features

• support storage prefix (#208)

0.7.2

Features

• support aws-s3 EC2 instance metadata authentication (#215)

Fixes

• NPE on revoke/discard shared resource (#212)

0.6.2

Features:

• support local filesystem and aws-s3 ec2 instance metadata authentication (#217)

0.7.1

Experimental unstable version (with server-side persistence support). Use at your own risk.

Fixes

• clean up shared access on resource deletion (#209)

0.7.0

Experimental unstable version (with server-side persistence support). Use at your own risk.

• Invitations are not cleaned up/updated after resource deletion
• Crash of Redis cluster may lead to resource/data loss

Features

• Bring back demo regime of core working with IDP #150 (#153)
• Calculate cost for models with "char without space" #154 (#155)
• Calculate token usage statistics using call stack #117 (#145)
• Implement limits through stateful layer #59 (#184)
• Improve token usage stats #160 (#161)
• Print execution path as JSON array to prompt logs #168 (#169)
• added seed, tools, system_prompt features and display_version flag to the listing (#204)
• check resource key size (#194)
• resources (#164)
• share API implementation (#190)
• support local file system for storage (#165) (#151)
• support recursive resource listing (#196)
• support if-none-match=* header in resource PUT API (#191)

Fixes

• Core shouldn't check user roles in case authorization by API key… (#181)
• Dockerfile after cleanup (#189)
• Error on direct call to Proxy API #192 (#193)
• Model cost calculation fails if token usage stats is not found in the response #157 (#158)
• Model cost calculator fails with NPE for some adapters #201 (#202)
• NPE occurs on rate limit hit in PostDeploymentController #178 (#179)
• Revert "fix: bump redisson client version (#206)" (#207)
• bump redisson client version (#206)
• fix error handling in proxy (#195)
• listing for root folder (#174)
• prompt enum element name (#177)
• refactor Files metadata to generic resource metadata (#147)
• resource test (#170)
• resource test (#172)
• retry when client fails to send response header (#148)
• selector with strict patterns (#200)
• shared file access (#156)

Other

• Add logging for unsuccessful requests (status code != 200) to … (#149)
• GitHub workflow update (#187)
• Unify field names with realtime analytics (#166)
• clean up settings/config files (#186)
• disable OTEL exporters by default (#167)
• disable otl metrics/logs if they are not enable explicitly (#176)
• disable trace exporter if the endpoint is not provided explicitly (#173)
• logging not 200 responses (#199)
• use apache http client in resource tests (#175)