v0.9.0

What's Changed

New Features

• feat: add llama.cpp launcher script and gated example by @psschwei in #1656
• refactor: consolidate extra_body merge logic into ModelOption.merge_extra_body by @cptnm3 in #1695
• feat(telemetry): emit llm.prompt_template.* on backend spans by @ajbozarth in #1672
• feat: llmon formatter by @AngeloDanducci in #1816

Bug Fixes

• fix: harden nightly Ollama and vision example coverage by @planetf1 in #1671
• fix(sampling): sampling span missing for strategies that override sam… by @cptnm3 in #1584
• fix: disable thinking in generate_with_context example by @simoncraf in #1686
• fix: add OpenAI adapter deregistration by @BrUn3y in #1676
• fix: pin create-github-app-token and scope app token permissions by @simoncraf in #1668
• fix: anchor coverage ellipsis exclude pattern to stub bodies by @planetf1 in #1682
• fix(ollama): case-fold model name comparison in is_model_available by @planetf1 in #1691
• fix(backends): activate aLoRA adapters on the LocalHFBackend generate path by @planetf1 in #1685
• fix(adapters): raise or fall back when an aLoRA cannot activate by @planetf1 in #1684
• fix: event loop handler teardown bug by @jakelorocco in #1704
• fix(backends): parse Granite XML tool calls on the local HF path by @planetf1 in #1692
• fix(ci): route template expressions through env in dispatch-to-contribs by @simoncraf in #1717
• fix(tools): carry element types and constraints into tool schemas by @planetf1 in #1694
• fix(backends): reject empty user content before sending chat request by @simonyang08 in #1632
• fix(backends): raise the documented ValueError when intrinsic output is not JSON by @planetf1 in #1725
• fix(backends): omit reasoning_effort=none for real OpenAI, not just the bool-False case by @planetf1 in #1709
• fix(core): await async validation functions in Requirement.validate so judge validation actually runs by @planetf1 in #1729
• fix(ci): bump gh-action-pypi-publish to v1.14.3 for GHSA-5xjx-3w67-w9x9 (backport #1779) by @ajbozarth in #1806
• fix: tool call result iteration (backport #1796) by @ajbozarth in #1803
• fix(ci): snapshot release docs from release tag (backport #1802) by @ajbozarth in #1808

Documentation

• docs: sync changelog for v0.8.0 by @mellea-auto-release[bot] in #1657
• docs(examples): link the example files in README by @spzala in #1711

Other Changes

• chore(adapters): bump granitelib-core pin for fixed requirement-check io.yaml by @planetf1 in #1706
• ci: pin action comments to exact releases by @planetf1 in #1713
• test(backends): raise lock-test timeout to tolerate shared GPU nodes by @planetf1 in #1726

New Contributors

• @simoncraf made their first contribution in #1686
• @BrUn3y made their first contribution in #1676
• @simonyang08 made their first contribution in #1632

Full Changelog: v0.8.0...v0.9.0