Skip to contents

Version 0.1

artgemini 0.1.0

  • Initial release with core Gemini API bindings

artgemini 0.1.5

  • Phase 3 complete: conversation continuation, tool support, health check
  • Phase 2 complete: artwork analysis functions aligned with artopenai

artgemini 0.1.6

Documentation

Naming Alignment with artopenai

  • Renamed artwork profile prompts to profile_* / profile_full_* (was art_profile_* / art_fullprofile_*) for cross-package parity.
  • Schema objects now follow artopenai naming: schema_profile, schema_profile_full (aliases retained for backward compatibility).
  • Introduced art_profile_full_ai() as the primary full-profile helper; art_fullprofile_ai() remains as a deprecated wrapper.

New Vignettes

  • Added Quickstart vignette covering core API usage: gemini_chat(), gemini_with_tools(), gemini_continue(), tools, model selection, temperature control, thinking mode, vision, and image encoding
  • Added Advanced Workflow vignette demonstrating production patterns: prompt system internals, structured JSON outputs, conversation history management, context caching, multi-candidate handling, and error handling

README Updates

  • Expanded “What is artgemini?” to reflect current capabilities (tools, continuation, caching)
  • Added “Key Features” section documenting: Thinking Mode (Gemini 3 only), Image Handling, Context Caching, and Multi-Candidate Responses
  • Updated “What endpoint is used?” to include tools in request payload
  • Consolidated Limitations sections
  • Added model table with use cases
  • Added ml parameter override example and validation note

Function Documentation

  • Enhanced gemini-cache docs with explicit cache size guard (4096 est. tokens), displayName behavior, and TTL defaults
  • Clarified max_think parameter: applies only to Gemini 3 models, ignored otherwise

Vignette Fixes

  • Fixed candidate-response-array.qmd build failure by converting to eval: false (was failing with HTTP 400 during live API calls)

artgemini 0.1.7

Vignette Standardization

  • Converted artgemini.Rmd to artgemini.qmd for format consistency
  • Standardized all vignettes to use eval: false for reliable CI builds
  • Fixed VignetteEngine specification: quarto::html (was causing pkgdown failures)

Housekeeping

  • Archived stale planning documents (TODO.md, TODO-REVIEW.md, PLANS.md) to .archived/
  • Created GitHub issues #22-#26 to track code review findings

artgemini 0.1.8

Documentation Quality Improvements

  • @param Documentation: Enhanced all exported function parameters to meet quality gate:

  • @description WHY: All function descriptions now explain WHY/WHEN to use each function, not just WHAT it does

  • README Enhancements:

    • Added “Quick Example” section with realistic code snippets
    • Added “Documentation” section with links to docs.artalytics.dev
    • Added “Development” section with AGENTS.md reference

artgemini 0.1.9

Reliability

  • Added ART_GEMINI_MAX_OUTPUT_TOKENS env var to control maxOutputTokens sent to Gemini generateContent API
    • When set, requests include maxOutputTokens in generationConfig
    • Default: not sent (uses Gemini API default); set env var to enable
    • Can also override per-request via genconf = list(maxOutputTokens = ...)
  • Added truncation detection for finishReason = "MAX_TOKENS"
    • Provides clear, actionable error message when output is truncated
    • Reports current cap value and how to increase it
  • Updated README with ART_GEMINI_MAX_OUTPUT_TOKENS documentation
  • Updated quickstart vignette Best Practices with max output tokens guidance