feat: skill-aligned execution harness#5

Merged

byapparov merged 19 commits into

mainfrom

feat/skill-aligned-execution

Mar 7, 2026

Commits on Mar 7, 2026

feat: automatic skill alignment and lean headless context for one-shot harness
Bulat Skill7
committed
docs: add standard operating procedure for headless tasks in AGENTS.md
Bulat Skill7
committed
feat: add hush code review skill and agent powered by GLM-5
Bulat Skill7
committed
feat: add Hush AI code review workflow powered by GLM-5
Bulat Skill7
committed
ci: fix hush-review workflow paths, syntax, and diff calculation
Bulat Skill7
committed
chore: rename hush-review.yml to code-review.yml for clarity
Bulat Skill7
committed
ci: switch to glibc-based lite binary for github actions compatibility
Bulat Skill7
committed
ci: resolve agent discovery issues and robustify instruction fetching
Bulat Skill7
committed
ci: fix instruction fetch crash and improve diff calculation
Bulat Skill7
committed
ci: implement safeFetch for instructions to definitively fix undefined crash
Bulat Skill7
committed
ci: force clean build and ensure workspace dependencies are built
Bulat Skill7
committed
feat: implement smart OpenCode review with SHA tracking and change filtering
Bulat Skill7
committed
fix: use released opencode CLI and correct secret name for CI review

Bulat Skill7
and
claude
committed
fix: remove aictrl-specific tools for CI and handle workflow_dispatch

Bulat Skill7
and
claude
committed
fix: use zai-coding-plan/glm-5 endpoint matching working hush config

Bulat Skill7
and
claude
committed
fix: guard skill loading in prompt to prevent session init crash

Bulat Skill7
and
claude
committed
fix: log skill loading errors instead of silently swallowing

Bulat Skill7
and
claude
committed
feat: surface skill loading errors to session.error bus for --format json

Bulat Skill7
and
claude
committed
feat: enrich --format json output with full session lifecycle events

Bulat Skill7
and
claude
committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat: skill-aligned execution harness#5

feat: skill-aligned execution harness#5
byapparov merged 19 commits into
mainfrom
feat/skill-aligned-execution

Commits on Mar 7, 2026

feat: automatic skill alignment and lean headless context for one-shot harness

docs: add standard operating procedure for headless tasks in AGENTS.md

feat: add hush code review skill and agent powered by GLM-5

feat: add Hush AI code review workflow powered by GLM-5

ci: fix hush-review workflow paths, syntax, and diff calculation

chore: rename hush-review.yml to code-review.yml for clarity

ci: switch to glibc-based lite binary for github actions compatibility

ci: resolve agent discovery issues and robustify instruction fetching

ci: fix instruction fetch crash and improve diff calculation

ci: implement safeFetch for instructions to definitively fix undefined crash

ci: force clean build and ensure workspace dependencies are built

feat: implement smart OpenCode review with SHA tracking and change filtering

fix: use released opencode CLI and correct secret name for CI review

fix: remove aictrl-specific tools for CI and handle workflow_dispatch

fix: use zai-coding-plan/glm-5 endpoint matching working hush config

fix: guard skill loading in prompt to prevent session init crash

fix: log skill loading errors instead of silently swallowing

feat: surface skill loading errors to session.error bus for --format json

feat: enrich --format json output with full session lifecycle events