generative-computing/mellea: v0.8.0

Abstract

<!-- Release notes generated using configuration in .github/release.yml at release/v0.8 --> What's Changed New Features feat: surface container artifacts via executionresult artifacts by @AngeloDanducci in https://github.com/generative-computing/mellea/pull/1384 refactor: make json_util a private module by @AngeloDanducci in https://github.com/generative-computing/mellea/pull/1399 feat: audio text to text support by @markstur in https://github.com/generative-computing/mellea/pull/1396 feat: carry reasoning across turns with per-backend replay policy by @AngeloDanducci in https://github.com/generative-computing/mellea/pull/1374 feat: imageblock and imgurblock ergonomics by @AngeloDanducci in https://github.com/generative-computing/mellea/pull/1377 feat: add license header remediation to pre commit by @AngeloDanducci in https://github.com/generative-computing/mellea/pull/1419 refactor(intrinsics): promote _DictContract to shared adapters module; collapse guardian.py's factuality contracts by @planetf1 in https://github.com/generative-computing/mellea/pull/1428 feat(telemetry): emit execute_tool spans for tool calls by @ajbozarth in https://github.com/generative-computing/mellea/pull/1430 refactor: extract shared Ollama eviction helper from conftest files by @AngeloDanducci in https://github.com/generative-computing/mellea/pull/1434 refactor: consolidate duplicate import_optional into optional.py by @AngeloDanducci in https://github.com/generative-computing/mellea/pull/1438 refactor(backends)!: narrow AdapterMixin verbs, centralize option resolution, add AdapterFunctionMetricsPlugin skeleton by @planetf1 in https://github.com/generative-computing/mellea/pull/1422 feat: add support for audio-text-to-text with m serve CLI and include examples by @markstur in https://github.com/generative-computing/mellea/pull/1443 refactor(backends): unify mot.raw.response shape across streaming and non-streaming by @cptnm3 in https://github.com/generative-computing/mellea/pull/1449 feat(telemetry): add sampling and validator tracing spans via hooks by @ajbozarth in https://github.com/generative-computing/mellea/pull/1488 refactor: enforce act/aact without strategy by @AngeloDanducci in https://github.com/generative-computing/mellea/pull/1468 refactor: introduce alias for component cblock modeloutputthunk union by @AngeloDanducci in https://github.com/generative-computing/mellea/pull/1458 feat!: emit per-chunk streaming span events on backend spans by @ajbozarth in https://github.com/generative-computing/mellea/pull/1496 feat: add mellea version dunder by @jakelorocco in https://github.com/generative-computing/mellea/pull/1524 feat(cli): allow model passed by OpenAI client to be used in the served Mellea program by @markstur in https://github.com/generative-computing/mellea/pull/1512 feat(backends): add default_extra_body to OpenAIBackend by @planetf1 in https://github.com/generative-computing/mellea/pull/1536 feat: remove m async fix command (and other improvements) by @jakelorocco in https://github.com/generative-computing/mellea/pull/1537 feat: add nemotron model ids by @jakelorocco in https://github.com/generative-computing/mellea/pull/1526 feat: allow components to declare their chat role via format_for_llm by @AngeloDanducci in https://github.com/generative-computing/mellea/pull/1542 feat: add zizmor audit job to CI quality workflow by @planetf1 in https://github.com/generative-computing/mellea/pull/1530 feat(backends): LocalFileBinding implements verbs (PEFT/aLoRA path) + from_catalog() (Epic #929 Phase 2) by @planetf1 in https://github.com/generative-computing/mellea/pull/1454 feat: make call_tools public by @akihikokuroda in https://github.com/generative-computing/mellea/pull/1544 feat(telemetry): align spans and metrics with OTel GenAI semantic conventions by @ajbozarth in https://github.com/generative-computing/mellea/pull/1551 refactor(streaming)!: replace stream_with_chunking with single-task stream() by @ajbozarth in https://github.com/generative-computing/mellea/pull/1543 refactor: improve message serialization by @AngeloDanducci in https://github.com/generative-computing/mellea/pull/1564 refactor(intrinsics): resolve the adapter's output contract from the adapter, not a parallel argument by @planetf1 in https://github.com/generative-computing/mellea/pull/1556 feat(backends): EmbeddedBinding implements apply_activation (Granite Switch path); remove render_controls + set_request_adapter (Epic #929 Phase 2) by @planetf1 in https://github.com/generative-computing/mellea/pull/1559 feat: add author-pluggable extension point for provider-specific message fields by @AngeloDanducci in https://github.com/generative-computing/mellea/pull/1586 feat(backends): populate mot.raw.response on the HF generate_from_raw by @cptnm3 in https://github.com/generative-computing/mellea/pull/1518 feat: add Granite 4.2 model defaults by @planetf1 in https://github.com/generative-computing/mellea/pull/1587 feat(backends): support embedded adapters on LocalHFBackend by @planetf1 in https://github.com/generative-computing/mellea/pull/1593 feat(streaming): add EventStreamer for iterating stream events by @ajbozarth in https://github.com/generative-computing/mellea/pull/1567 feat(backends): accept composed Adapter directly (Epic #929, issue #1144, PR1) by @planetf1 in https://github.com/generative-computing/mellea/pull/1619 feat: add per-requirement chunking for streaming validation by @ajbozarth in https://github.com/generative-computing/mellea/pull/1630 feat: preserve Context subtype and enforce input==output context type by @AngeloDanducci in https://github.com/generative-computing/mellea/pull/1582 feat(core): audio file/URL constructors and representation-aware capability guards by @jakelorocco in https://github.com/generative-computing/mellea/pull/1601 feat(backends): support Ollama adapter functions by @planetf1 in https://github.com/generative-computing/mellea/pull/1634 Bug Fixes fix(ci): serialize ollama requests to prevent llama-server segfault cascade by @ajbozarth in https://github.com/generative-computing/mellea/pull/1381 fix: port telemetry test-order fix + broken aLora example skip to main (#1378) by @ajbozarth in https://github.com/generative-computing/mellea/pull/1382 fix(ci): generate API/CLI docs before snapshotting release version by @ajbozarth in https://github.com/generative-computing/mellea/pull/1386 fix: preserve tool_calls in multi-turn history and skip malformed tool-arg JSON for OpenAI-compatible backends by @mvanhorn in https://github.com/generative-computing/mellea/pull/1373 fix(ci): revert OLLAMA_NUM_PARALLEL serialization (#1381) by @ajbozarth in https://github.com/generative-computing/mellea/pull/1387 fix: test kv flake with rephrase and chatcontext by @AngeloDanducci in https://github.com/generative-computing/mellea/pull/1403 fix: refactor bash validation by @akihikokuroda in https://github.com/generative-computing/mellea/pull/1400 fix: force cpu accelerator for docling on apple silicon mps by @AngeloDanducci in https://github.com/generative-computing/mellea/pull/1417 fix: ollama releases prior to the pre release today ran into issues with amx runners by @AngeloDanducci in https://github.com/generative-computing/mellea/pull/1404 fix: pypi build files by @jakelorocco in https://github.com/generative-computing/mellea/pull/1414 fix: updated assert to fix flakeyness of sampling example by @rphadnis in https://github.com/generative-computing/mellea/pull/1420 fix: bump pinned Ollama version to 0.32.2 (rc0 tag removed) by @planetf1 in https://github.com/generative-computing/mellea/pull/1425 fix: raise ValueError for image/audio inputs in LocalHFBackend by @markstur in https://github.com/generative-computing/mellea/pull/1410 fix: Context.last_turn by @akihikokuroda in https://github.com/generative-computing/mellea/pull/1423 fix: add recursive nested-model $ref resolution in Pydantic tool parameter schemas by @akihikokuroda in https://github.com/generative-computing/mellea/pull/1416 fix: allow act to operate on CBlock and ModelOutputThunk by @AngeloDanducci in https://github.com/generative-computing/mellea/pull/1429 fix: bump langchain-core and litellm floors to clear cve by @AngeloDanducci in https://github.com/generative-computing/mellea/pull/1447 fix(hf): stop generation workers on stream timeout by @dk3yyyy in https://github.com/generative-computing/mellea/pull/1469 fix(backends): merge intrinsic extra_body once by @binggao1230 in https://github.com/generative-computing/mellea/pull/1300 fix: port analytics review improvements from website PR by @ajbozarth in https://github.com/generative-computing/mellea/pull/1463 fix: doc autogen test assertions by @AngeloDanducci in https://github.com/generative-computing/mellea/pull/1480 fix: change tool_calls from dict to list by @akihikokuroda in https://github.com/generative-computing/mellea/pull/1435 fix(core): replace bare assert with ValueError in ImageBlock/AudioBlock validation by @markstur in https://github.com/generative-computing/mellea/pull/1490 fix: assign the result of Context.add in transform/atransform by @planetf1 in https://github.com/generative-computing/mellea/pull/1477 fix: do not share session state across requests in m_serve examples by @markstur in https://github.com/generative-computing/mellea/pull/1481 fix: converge litellm to a single version across python versions by @AngeloDanducci in https://github.com/generative-computing/mellea/pull/1494 fix(stdlib): resolve postponed annotations in generative stub signature by @planetf1 in https://github.com/generative-computing/mellea/pull/1504 fix: surface HF auth failures when loading switch adapters by @psschwei in https://github.com/generative-computing/mellea/pull/1501 fix: feed tool results to open ai compat providers as role tool with id by @AngeloDanducci in https://github.com/generative-computing/mellea/pull/1505 fix: raise clear