Repository navigation

feat(aws-devops-agent): add the experiment-orchestration skill and agent prompt - #230

Merged

cchu-ld merged 21 commits into

mainfrom

cchu/aws-devops-agent

Oct 29, 2026

Merged

feat(aws-devops-agent): add the experiment-orchestration skill and agent prompt#230
cchu-ld merged 21 commits into
mainfrom
cchu/aws-devops-agent

Commits on Oct 28, 2026
- feat(aws-devops-agent): add the experiment-orchestration skill and agent prompt
  
  cchu-ld
  and
  claude
  committed
  bbdf962
- docs(aws-devops-agent): fix the skill's description and add catalog metadata
  
  cchu-ld
  and
  claude
  committed
  77fb5d4
- fix(aws-devops-agent): handle missing GitHub and guarded rollouts in the skill
  
  cchu-ld
  and
  claude
  committed
  b72f6dd
- fix(aws-devops-agent): make the skill's handoffs and fallback consistent
  
  cchu-ld
  and
  claude
  committed
  bfd924a
- docs(aws-devops-agent): clarify the skill's wording from review
  
  cchu-ld
  and
  claude
  committed
  d6e5e50
- docs(aws-devops-agent): say ldcli is only needed for setup
  
  cchu-ld
  and
  claude
  committed
  a0d0efc
- fix(aws-devops-agent): guard the experiment against peeking and small samples
  
  cchu-ld
  and
  claude
  committed
  16aa985
- fix(aws-devops-agent): spell out the 10% experiment setup
  
  cchu-ld
  and
  claude
  committed
  d97d6d4
- docs(aws-devops-agent): name the steps that numbered lists refer to
  
  cchu-ld
  and
  claude
  committed
  d705754
- fix(aws-devops-agent): find or create the experiment's metrics
  
  cchu-ld
  and
  claude
  committed
  e520323
- fix(aws-devops-agent): request approval directly once an entity needs it
  
  cchu-ld
  and
  claude
  committed
  32dea21
Commits on Oct 29, 2026
- feat(evals): per-test mock overrides, tool exclusion, and inlined skills
  
  cchu-ld
  and
  claude
  committed
  6446e86
- test(evals): add the experiment-orchestration suite
  
  cchu-ld
  and
  claude
  committed
  eefc28c
- fix(aws-devops-agent): never apply a pending approval or act on an unanswered goal
  
  cchu-ld
  and
  claude
  committed
  01caf7b
- test(evals): let a test dismiss ask-question
  
  cchu-ld
  and
  claude
  committed
  7ac0421
- fix(evals): keep new mock tools and create-experiment rules from changing other suites
  
  cchu-ld
  and
  claude
  committed
  ea419bf
- feat(aws-devops-agent): support multivariate experiments and blocked tools
  
  cchu-ld
  and
  claude
  committed
  53f88a6
- refactor(evals): keep this suite's tools and mocks out of the shared files
  
  cchu-ld
  and
  claude
  committed
  74e599c
- test(evals): cover multivariate flags in experiment-orchestration
  
  cchu-ld
  and
  claude
  committed
  4ef63aa
- fix(experiment-orchestration): judge treatments one at a time; send all metrics to the rollout
  
  cchu-ld
  and
  claude
  committed
  50d2d59
- fix(experiment-orchestration): exact hand-off steps for blocked tools; create-flag defaults fallback
  
  cchu-ld
  and
  claude
  committed
  dd04f43

Commits

Commits on Oct 28, 2026

Commits on Oct 29, 2026