Skip to content

Latest commit

 

History

History
408 lines (220 loc) · 6.41 KB

File metadata and controls

408 lines (220 loc) · 6.41 KB

apollo

0.21.2

Patch Changes

  • 7b91dc7: Add HTTP logs to the server
  • 7b91dc7: echo: fix an issue which prevented the service from running

0.21.1

Patch Changes

  • 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)

0.21.0

Minor Changes

  • chat agents: Reverse the order of text and attachments in streaming responses

0.20.2

Patch Changes

  • Update adaptor API dependency, fixing an issue in the global agent

0.20.1

Patch Changes

  • b4215f5: pass global agent api key to subagents

0.20.0

Minor Changes

  • da1c1c9: add global_chat service for one-shot generation

    Includes planner agent and router

Patch Changes

  • Dependency updates

0.19.8

Patch Changes

  • de90b22: remove adaptor docs from job code rag

0.19.7

Patch Changes

  • fd43595: prefill doc query answers

0.19.6

Patch Changes

  • 7d42848: lint code
  • 8f39895: update protobuf
  • 203175d: add error handling to job_chat

0.19.5

Patch Changes

  • Fix doc loading in production environments

0.19.4

Patch Changes

  • bd2c51b: workflow_chat: don't refer to a seperate AI assistant when asked to write job code

0.19.3

Patch Changes

  • af07795: Update adaptor apis package

0.19.2

Patch Changes

  • 27fda5b: Ensure db tables are checked before adaptors are uploaded

0.19.1

Patch Changes

  • Update claude model version for RAG

0.19.0

Minor Changes

  • 9a7ede4: Support mixed-history sessions in chat services

    job_chat: add navigation prefixes and rag refresh

Patch Changes

  • 133a38e: workflow_chat: improve the model's understanding of edge branching behaviour

0.18.0

Minor Changes

  • ec31a56: workflow_chat: support read_only mode flag

Patch Changes

  • ec31a56: workflow_chat: prevent identical job names being created in generated workflows
  • 66d3839: Fix anthropic healthcheck

0.17.2

Patch Changes

  • Fix server dependencies

0.17.1

Patch Changes

Fix an issue where the server fails to start

0.17.0

Minor Changes

  • 942cf4f: - Add load_adaptor_docs and search_adaptor_docs services, which use postgres to cache and search docs data
    • Use new adaptor docs lookup in job_chat, for faster, leaner and better docs lookup

Patch Changes

  • 3095a37: add refresh rag flag

0.16.3

Patch Changes

  • fc9bc49: workflow_chat: fix an issue where ids may not be perserved properly
  • fc9bc49: job_chat: improved performance when reading logs from a run

0.16.2

Patch Changes

  • 02f2a8e: Add barebones support for postgresdb, with status check

0.16.1

Patch Changes

  • Increase server timeout for streaming

0.16.0

Minor Changes

  • Add support for HTTP streaming to job_chat and workflow_chat

0.15.0

Patch Changes

  • 3c1b932: update to claude 4

0.14.3

Patch Changes

    • Truncate adaptor docs for legacy job_chat
    • Report apollo errors to sentry
    • Update error message when the prompt is too long

0.14.2

Patch Changes

  • Add caching to adaptor docs generation
  • 79bc424: Make job_chat json output more robust
  • ff7a9bc: add sanitisation to edges in yaml

0.14.1

Patch Changes

  • 4cb99d1: add sentry-sdk dependency (fixes an error when calling job chat)

0.14.0

Minor Changes

  • d4ffb0a: job_chat: add suggested_code

0.13.3

Patch Changes

  • ed32f7f: job_chat: truncate adaptor docs to 40k characters

0.13.2

Patch Changes

  • ed32f7f: job_chat: Fix an issue where the model would ignore user logs

Patch Changes

0.13.1

Patch Changes

  • job_chat: Fix an issue where the model would ignore user logs
  • 7b48834: workflow_chat: preserve UUIDs during workflow generation

0.13.0

Minor Changes

  • b98bfab: add updated adaptor fetching

0.12.0

Minor Changes

  • 4f5a5f2: allow ai edits of existing workflows

Patch Changes

  • 2776d4d: update token limit for job chat

0.11.4

Patch Changes

  • bdc0daf: improve prompt for workflow generation

0.11.3

Patch Changes

  • a3119a3: Override workflow generation yaml body key to be empty
  • 1159eeb: fix workflow gen variable scope

0.11.2

Patch Changes

  • workflow_chat: update adaptor list

0.11.1

Patch Changes

  • Ensure the user's API key is used for rag search

0.11.0

Minor Changes

  • bae48d3: Add workflow_chat service

0.10.0

Minor Changes

  • 6f88248: Integrate docsite rag for the AI assistant

0.9.2

Patch Changes

  • aef476d: Update claude model from 3.5 to 3.7

0.9.1

Patch Changes

  • job_chat: update Claude to 3.7

0.9.0

Minor Changes

  • c0e391c: Add new docsite rag
  • 14c149d: Add a /status endpoint which returns the status of server-managed API keys

0.8.0

Minor Changes

  • a03bbea: vocab mapper: Add batching

Patch Changes

  • 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

0.7.2

Patch Changes

  • ob_chat: Fix dependencies
  • vocab_mapper: Fix production key loading

0.7.1

Patch Changes

  • Bump dependencies

0.7.0

Minor Changes

  • ed5ff26: Add vocab mapper service

Patch Changes

  • ef9f9d2: Add version number to the service listing and healthcheck

0.6.0

Minor Changes

  • 032f387: Added embeddings service

0.5.3

Patch Changes

  • job-chat: return standardised and detailed error messages

0.5.2

Patch Changes

  • 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"

0.5.1

Patch Changes

  • Temporarily disable embeddings in build (search is disabled anyway)

0.5.0

Minor Changes

  • 73096a0: Added project generator
  • d3da3cb: Add job generator service
  • 013506f: Added search service (RAG)

0.4.1

Patch Changes

  • job chat: fix formatting issue

0.4.0

Minor Changes

  • 4c8c7eb: use Claude for job_chat

0.3.2

Patch Changes

  • chat: de-emphasise each in the prompt

0.3.1

Patch Changes

  • job_chat: tweak prompt for better performance

0.3.0

Minor Changes

  • 5fa089b: - Support Typescript services
    • Add describe_adaptor service
    • Add job_chat service

0.2.0

Minor Changes

  • Use child processes to run python scripts
  • add websocket connection with streaming logs