Changelog

0.3.0

Breaking changes

Prompt templates are replaced by skills throughout the SDK, matching the RevelryAI API (where prompt_template_id has been deprecated in favor of skill_id). Function signatures also changed so that path/query values are positional arguments and the params map is exactly the request body.

• RevelryAI.PromptTemplate is removed. Use RevelryAI.Skill.list/2, which calls the current /skills endpoint and returns skills instead of prompt templates.

• RevelryAI.Artifact.create/2 is now RevelryAI.Artifact.create/4 and takes skill_id instead of prompt_template_id:

  # before (0.2.x)
  RevelryAI.Artifact.create(%{
    prompt_template_id: 2,
    artifact_slug: "story",
    project_id: 1,
    inputs: [%{name: "Context", value: "test"}]
  })
  
  # after (0.3.0) — the skill_id replaces the prompt_template_id
  RevelryAI.Artifact.create("story", 1, %{
    skill_id: 2,
    inputs: [%{name: "Context", value: "test"}]
  })

  Skill IDs are not interchangeable with prompt template IDs — look up the new ID with RevelryAI.Skill.list/2.

• RevelryAI.Artifact.stream_create_artifact/2 is now stream_create_artifact/4 with the same signature change as create/4.

Added

• V2 API support (async-only endpoints):
  • RevelryAI.V2.Artifact.create/3 — create an artifact asynchronously; completion is delivered via webhook.
  • RevelryAI.V2.Skill.run/4 — run a skill without creating an artifact.
  • RevelryAI.V2.ApiJob.get/2 and RevelryAI.V2.ApiJob.await/2 — poll async job results.
• V1 Artifact.create/4 now passes through optional name and model_configuration_id params.

Fixed

• RevelryAI.Client treated only HTTP 200 as success; any 2xx (e.g. the 202 returned by v2 endpoints) is now handled as success.

0.2.0 and earlier

See the GitHub releases.