Shared AI foundation for TYPO3. Configure LLM providers once — every AI extension uses them. Supports OpenAI, Anthropic, Google Gemini, Ollama, and more. Includes services for chat, translation, vision, and embeddings with encrypted API keys and full admin control. - by Netresearch
13 LTS 14 LTS
13 LTS 14 LTS
- chore: release 0.17.0 (#355)
- chore: release 0.17.0
- feat: configuration presets declared by consuming extensions (ADR-056) (#354)
- feat: configuration presets declared by consuming extensions (ADR-056)
- feat: embedding configuration path and model dimensions metadata (ADR-055) (#353)
- feat: embedding configuration path and model dimensions metadata (ADR-055)
- feat: typed tool-turn messages on ChatMessage (ADR-054) (#352)
- fix(ci): run functional suites without the PHP JIT (#351)
- feat: typed tool-turn messages on ChatMessage (ADR-054)
- fix(ci): run functional suites without the PHP JIT
- feat: NrLlmExceptionInterface marker for all public-surface exceptions (ADR-053) (#350)
- feat: NrLlmExceptionInterface marker for all public-surface exceptions (ADR-053)
- fix: usage attribution honours the caller-supplied beUserUid (ADR-052) (#349)
- fix: usage attribution honours the caller-supplied beUserUid (ADR-052)
- feat: add ToolCallingService feature-service pair (ADR-051) (#348)
- feat: add ToolCallingService feature-service pair (ADR-051)
- docs(adr-050): decide the deferred question — no nr_llm-owned persistent index (#344)
- docs(adr-050): decide the deferred question — no nr_llm-owned persistent index
- docs: ADR-050 — retrieval/embedding scope boundary with nr_ai_search (#343)
- docs: match ADR-050 title rule length, tidy open-question phrasing
- docs: ADR-050 retrieval/embedding scope boundary with nr_ai_search
- fix: surface Ollama's native message.thinking in completion responses (#342)
- test: cover plain chat path and whitespace normalisation for message.thinking
- fix: surface Ollama's native message.thinking in completion responses
- feat: reasoning toggle for hybrid-thinking models (Ollama think) (#341)
- feat: reasoning toggle for hybrid-thinking models (Ollama think)
For details see https://github.com/netresearch/t3x-nr-llm/releases
Filter for TYPO3 versions
- chore: release 0.17.0
- feat: configuration presets declared by consuming extensions (ADR-056) (#354)
- feat: configuration presets declared by consuming extensions (ADR-056)
- feat: embedding configuration path and model dimensions metadata (ADR-055) (#353)
- feat: embedding configuration path and model dimensions metadata (ADR-055)
- feat: typed tool-turn messages on ChatMessage (ADR-054) (#352)
- fix(ci): run functional suites without the PHP JIT (#351)
- feat: typed tool-turn messages on ChatMessage (ADR-054)
- fix(ci): run functional suites without the PHP JIT
- feat: NrLlmExceptionInterface marker for all public-surface exceptions (ADR-053) (#350)
- feat: NrLlmExceptionInterface marker for all public-surface exceptions (ADR-053)
- fix: usage attribution honours the caller-supplied beUserUid (ADR-052) (#349)
- fix: usage attribution honours the caller-supplied beUserUid (ADR-052)
- feat: add ToolCallingService feature-service pair (ADR-051) (#348)
- feat: add ToolCallingService feature-service pair (ADR-051)
- docs(adr-050): decide the deferred question — no nr_llm-owned persistent index (#344)
- docs(adr-050): decide the deferred question — no nr_llm-owned persistent index
- docs: ADR-050 — retrieval/embedding scope boundary with nr_ai_search (#343)
- docs: match ADR-050 title rule length, tidy open-question phrasing
- docs: ADR-050 retrieval/embedding scope boundary with nr_ai_search
- fix: surface Ollama's native message.thinking in completion responses (#342)
- test: cover plain chat path and whitespace normalisation for message.thinking
- fix: surface Ollama's native message.thinking in completion responses
- feat: reasoning toggle for hybrid-thinking models (Ollama think) (#341)
- feat: reasoning toggle for hybrid-thinking models (Ollama think)
For details see https://github.com/netresearch/t3x-nr-llm/releases