AgentSkillsCN

incremental-deployment

在部署时,运用此技能确保仅部署经过修改且受影响的组件,并明确输入版本/标签信息,同时保持环境的可追溯性。

SKILL.md
--- frontmatter
name: incremental-deployment
description: >
  Use this skill when deploying to ensure only modified and impacted components
  are deployed, with explicit version/tag inputs and environment traceability
  preserved.

Incremental Deployment Planning

Intent

Deploy only what changed (and what is impacted), while preserving strict traceability via immutable versions and tags.


When to Use

  • Planning deployments in a monorepo with multiple components
  • Preparing environment promotion
  • Reviewing deployment scope for safety and efficiency

Precondition Failure Signal

  • Broad deployments include unrelated components
  • Deployments are triggered from branches without tags
  • Environment state cannot be explained in component version terms
  • Incremental deployments miss required dependent components

Postcondition Success Signal

  • Deployment scope matches impacted components (and required dependencies)
  • Deployments reference explicit component versions/tags
  • Environment state is traceable to deployed component versions/tags
  • Deployment plan is reproducible and reviewable

Process

  1. Source Review: Inspect the impact set produced by incremental-change-impact and the current state of the target environment.
  2. Implementation: Execute the deployment only for the identified impacted components using their specific versions/tags.
  3. Verification: Verify that only the intended components were deployed and that their versions/tags correctly reflect the source code.
  4. Documentation: Update the environment traceability records with the new deployment details.
  5. Review: Release Manager/SRE and Platform Engineer review the deployment plan and execution results.

Example Test / Validation

  • Given a change set, show only impacted components are deployed
  • Validate dependent components are included when dependency changes require it
  • Confirm environment records can list component versions/tags after deployment

Common Red Flags / Guardrail Violations

  • “Just redeploy everything to be safe”
  • Deploying from main or HEAD without a tag
  • Allowing “implicit latest” component selection
  • Changing deployment scope without updating impact assessment

Recommended Review Personas

  • Platform/DevOps Engineer – validates deployment scope and traceability
  • Tech Lead – validates correctness of impacted set and risk discipline
  • Domain Expert – validates runtime dependency assumptions

Skill Priority

P1 – Quality & Correctness
(Escalate to P0 where traceability/immutability constraints are threatened.)


Conflict Resolution Rules

  • Must consume incremental-change-impact and release-tagging-plan outputs
  • If uncertainty exists, broaden scope via impact justification, not via bypass of tagging/traceability
  • Overrides Delivery/Optimisation shortcuts that weaken traceability

Conceptual Dependencies

  • incremental-change-impact
  • release-tagging-plan

Classification

Operational
Governance


Notes

Incremental deployment is only safe when traceability is strict.