Skip to content

AIXCL v1.1.69

Choose a tag to compare

View all tags
@github-actions github-actions released this 26 Aug 20:36
· 17 commits to main since this release
v1.1.69
72a24b7
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

AIXCL v1.1.69

Maintenance release: recurring OpenCode default-model EOL fixes and a new model-liveness check, a reverted stale-branch config regression, routine stack service image updates, and an Ollama runtime-core bump -- all verified live.

What's New in v1.1.69

Added

  • ✅ Delegate skill shared server, default-model delegation, bounded parallelism (closes #2018): shared opencode serve process with --attach, no hardcoded model, 2-tier fallback, up to 3 concurrent delegations.
  • ✅ Dynamic OpenCode model liveness check (closes #2026): ./aixcl checks models derives the model list from opencode.json's own provider block and probes the configured model/small_model defaults through the shared server, failing only if a default is dead.

Changed

  • ✅ Bump docker/setup-buildx-action from 4.2.0 to 4.3.0 (closes #2031)
  • ✅ Bump github/codeql-action from 4 to 4.37.4 (closes #2021)
  • ✅ Update 10 stack service images to latest patch/minor versions (closes #2036): Open WebUI, Grafana, Loki, Vault, PostgreSQL, Prometheus, Alertmanager, Node Exporter, JSON Exporter, NVIDIA GPU Exporter. Verified live at 21/21 services healthy.
  • ✅ Update Ollama from v0.31.2 to v0.33.0 (closes #2037): runtime-core bump. Verified live: inference tested against an existing model, OpenAI-compatible endpoint (/v1/chat/completions) confirmed responding correctly, 21/21 services healthy.

Fixed

  • ✅ Test-03-stack-status no longer fails on stopped-stack exit code (closes #2020)
  • ✅ Default OpenCode to NVIDIA model, local as last resort (closes #2016)
  • ✅ Dead OpenCode default model after NVIDIA deepseek-v4 EOL (closes #2024)
  • ✅ Dead safety agent model replaced with live nemotron-safety-guard (closes #2028): NVIDIA retired nemotron-3-content-safety; replaced with nvidia/llama-3.1-nemotron-safety-guard-8b-v3.
  • ✅ Reverted PR #2030's stale opencode.json.example regression on main (closes #2033): a month-stale branch had been merged directly to main, clobbering dev's permission block, provider schema, and live model catalog; restored dev's current state.
  • ✅ Dead OpenCode default model nvidia/z-ai/glm-5.2 (closes #2035): NVIDIA retired z-ai/glm-5.2; replaced with live-candidate-probed nvidia/moonshotai/kimi-k3 after confirming tool-call support.

Documentation

Full Changelog: v1.1.68...v1.1.69

Assets 2
Loading