- 7b91dc7: Add HTTP logs to the server
- 7b91dc7: echo: fix an issue which prevented the service from running
- 11e8cea: fix empty yaml key handling after streaming order reversal
- 5d387ca: fix model use in rag
- workflow_chat: Fix streaming sending "null" as YAML changes event for informational responses (#427)
- chat agents: Reverse the order of text and attachments in streaming responses
- Update adaptor API dependency, fixing an issue in the global agent
- b4215f5: pass global agent api key to subagents
-
da1c1c9: add global_chat service for one-shot generation
Includes planner agent and router
- Dependency updates
- de90b22: remove adaptor docs from job code rag
- fd43595: prefill doc query answers
- 7d42848: lint code
- 8f39895: update protobuf
- 203175d: add error handling to job_chat
- Fix doc loading in production environments
- bd2c51b: workflow_chat: don't refer to a seperate AI assistant when asked to write job code
- af07795: Update adaptor apis package
- 27fda5b: Ensure db tables are checked before adaptors are uploaded
- Update claude model version for RAG
-
9a7ede4: Support mixed-history sessions in chat services
job_chat: add navigation prefixes and rag refresh
- 133a38e: workflow_chat: improve the model's understanding of edge branching behaviour
- ec31a56: workflow_chat: support read_only mode flag
- ec31a56: workflow_chat: prevent identical job names being created in generated workflows
- 66d3839: Fix anthropic healthcheck
- Fix server dependencies
Fix an issue where the server fails to start
- 942cf4f: - Add
load_adaptor_docsandsearch_adaptor_docsservices, which use postgres to cache and search docs data- Use new adaptor docs lookup in
job_chat, for faster, leaner and better docs lookup
- Use new adaptor docs lookup in
- 3095a37: add refresh rag flag
- fc9bc49: workflow_chat: fix an issue where ids may not be perserved properly
- fc9bc49: job_chat: improved performance when reading logs from a run
- 02f2a8e: Add barebones support for postgresdb, with status check
- Increase server timeout for streaming
- Add support for HTTP streaming to job_chat and workflow_chat
- 3c1b932: update to claude 4
-
- Truncate adaptor docs for legacy job_chat
- Report apollo errors to sentry
- Update error message when the prompt is too long
- Add caching to adaptor docs generation
- 79bc424: Make job_chat json output more robust
- ff7a9bc: add sanitisation to edges in yaml
- 4cb99d1: add sentry-sdk dependency (fixes an error when calling job chat)
- d4ffb0a: job_chat: add suggested_code
- ed32f7f: job_chat: truncate adaptor docs to 40k characters
- ed32f7f: job_chat: Fix an issue where the model would ignore user logs
- job_chat: Fix an issue where the model would ignore user logs
- 7b48834: workflow_chat: preserve UUIDs during workflow generation
- b98bfab: add updated adaptor fetching
- 4f5a5f2: allow ai edits of existing workflows
- 2776d4d: update token limit for job chat
- bdc0daf: improve prompt for workflow generation
- a3119a3: Override workflow generation yaml body key to be empty
- 1159eeb: fix workflow gen variable scope
- workflow_chat: update adaptor list
- Ensure the user's API key is used for rag search
- bae48d3: Add workflow_chat service
- 6f88248: Integrate docsite rag for the AI assistant
- aef476d: Update claude model from 3.5 to 3.7
- job_chat: update Claude to 3.7
- c0e391c: Add new docsite rag
- 14c149d: Add a
/statusendpoint which returns the status of server-managed API keys
- a03bbea: vocab mapper: Add batching
- 1d824ba: job_chat: Update the AI Assistant prompt to be more concise, secure, and context-aware
- 960dffb: job_chat: add basic prompt support for collections
- ob_chat: Fix dependencies
- vocab_mapper: Fix production key loading
- Bump dependencies
- ed5ff26: Add vocab mapper service
- ef9f9d2: Add version number to the service listing and healthcheck
- 032f387: Added embeddings service
- job-chat: return standardised and detailed error messages
- Improve job_chat prompt to be less strict about talking about non-job-related stuff (eg, it's a lot more willing to discuss adaptors and integrations"
- Temporarily disable embeddings in build (search is disabled anyway)
- 73096a0: Added project generator
- d3da3cb: Add job generator service
- 013506f: Added search service (RAG)
- job chat: fix formatting issue
- 4c8c7eb: use Claude for job_chat
- chat: de-emphasise each in the prompt
- job_chat: tweak prompt for better performance
- 5fa089b: - Support Typescript services
- Add describe_adaptor service
- Add job_chat service
- Use child processes to run python scripts
- add websocket connection with streaming logs