Google ADK (Agent Development Kit) orchestration patterns — boundaries, agent composition, and tool seams. Trigger when designing or reviewing multi-agent systems built on ADK. Authoritative source: adk.dev.
中文简介正在准备中。当前显示原文说明。
Govern the SemVer release process — versioning discipline, tag-based GitHub releases, changelog hygiene, and the ADR that confirms the release strategy is agreed. Trigger when setting up a release process, before a first public release, or when release practice has become inconsistent. Owned by delivery-orchestrator.
中文简介正在准备中。当前显示原文说明。
安装前可以查看交给 AI 代理的指令内容。
A release without a confirmed process is a deployment. A deployment without a rollback plan is a gamble.
This skill governs the release process itself — not a specific deployment (that is
release-readiness) and not the CI infrastructure (that is github-manager). It ensures
the team has agreed, documented, and is consistently following a versioning and release
strategy, anchored by an ADR.
delivery-orchestrator identifies a release-process gap during R2/R3 classificationThree release-adjacent skills exist in this harness with distinct responsibilities:
| Skill | Responsibility |
|---|---|
github-manager | CI trigger configuration, runner cost, branch protection, tag-event wiring |
release-manager (this skill) | Process governance: SemVer discipline, changelog, ADR, deviation authority |
release-readiness | Go/no-go gate for a specific deployment: failure modes, rollback, monitoring |
Use all three in sequence for a new project. Use release-manager alone when auditing or
repairing an existing process. Check authority before any tag or build trigger. Hand off to release-readiness
before promotion; when a tag starts deployment, its workflow must hold promotion
until the artifact checks and authorised go/no-go decision are complete.
First resolve the project's approved release ADR and operating profile. The default below is a recommendation until adopted. An approved alternative governs the audit, checklist and pipeline too; it is not an exception to evidence or authority requirements. If the ADR and profile disagree, stop the affected release action and reconcile them.
Declare tag-ci or approved-alternative, build trigger/executor, promotion trigger,
release owner and decision reference in the profile. Read
the release contract
for the fields and offline structural check. The default, unless the ADR differs, is:
MAJOR.MINOR.PATCH
PATCH — backwards-compatible bug fixesMINOR — backwards-compatible new capabilityMAJOR — breaking changesv{MAJOR}.{MINOR}.{PATCH} tags on the default branch trigger release
builds in CI. Promotion is separately gated; the tag alone grants no authority.tag-ci mode. An approved
alternative must identify its controlled executor and retained provenance; a local
build does not qualify merely because it completed.CHANGELOG.md updated before every release; format follows
Keep a Changelog.v* tags to the default branch.An ADR must exist that records:
If no ADR exists, create one using the-architect. The ADR is the authority record —
process enforcement without one is informal and will drift.
Check the repository for evidence of adherence:
CHANGELOG.md up to date for every tagged release?Flag every gap between declared ADR and observed practice. Gaps are findings, not acceptable workarounds.
Confirm the following are in place (coordinate with github-manager for CI config):
v* → CI for the default mode).The release owner applies the declared mode, not an unconditional CI checklist:
Before promotion, obtain the authorised go/no-go decision for the actual artifact and target environment. If a tag triggers the pipeline, hold its deployment/publication stage until this gate completes. A structurally valid record is not release approval. After authorised delivery, verify the destination digest, observe the agreed signals, and reconcile the release record, changelog and status. Notify downstream consumers of breaking changes through the approved channel.
| Excuse | Counter |
|---|---|
| "We all know the release process" | Tribal knowledge drifts. An ADR does not. |
| "The changelog is a nice-to-have" | Every future debugging session starts there. Write it now. |
| "I'll build the release locally, it's faster" | Use only the approved executor and retain provenance; speed does not approve an alternative. |
| "We don't need an ADR for something this simple" | One page of ADR prevents months of inconsistency. Write it. |
| "The tag was already pushed, I'll do the changelog after" | The changelog belongs before the tag. Reversing this loses the discipline. |
还没有评价。欢迎分享你的使用感受。
无需登录即可发表。欢迎分享使用感受、提问,或说明在哪些环境下无法运行。
还没有评论。来写第一条吧。
Google ADK (Agent Development Kit) orchestration patterns — boundaries, agent composition, and tool seams. Trigger when designing or reviewing multi-agent systems built on ADK. Authoritative source: adk.dev.
中文简介正在准备中。当前显示原文说明。
JP's signature red-team pass — "how would I break this?" Argue against your own approach before proceeding. Trigger on any high-stakes decision, architecture choice, or before marking work complete.
中文简介正在准备中。当前显示原文说明。
Read-only SRE checkup of any GCP project: deterministic probes of the edge, Cloud Run services, 7-day error logs, Cloud Scheduler, alert policies and uptime checks, Secret Manager and IAM, the data stores and the machine's own scheduled jobs, audited into one fixed status table (LIVE / WARNING / RED / INCONCLUSIVE) with evidence, findings by severity, what could not be checked, and a single OVERALL line delivered as one notification. Parametrised by a per-project manifest, so the same routine runs on every project. Use when the operator says "cloud checkup", "SRE check", "is everything live", "what's healthy / warning / red", "any errors this week", "audit the infra", "weekly checkup", "set up the weekly checkup", before a deploy or demo, or after an incident. Cloud Run first; App Engine and GKE differ only in the serving probes.
中文简介正在准备中。当前显示原文说明。
Cloud guardrails for any vendor workload — Google Cloud (GCP, Vertex AI, GKE), AWS (IAM, EKS, Bedrock), Azure (Entra ID, Policy, AKS), Alibaba Cloud (RAM, mainland/international residency). Enforces identity least-privilege, mechanical policy, data boundaries, residency, cost caps, network egress and observability, with official-source validation before any claim. Trigger on any cloud infrastructure design, review, Terraform plan, or LLM/agent deployment; the-architect routes here.
中文简介正在准备中。当前显示原文说明。
LLM and cloud cost awareness — model tiering, token budgets, right-sizing, and when a cheaper model suffices. Trigger before finalising any architecture that calls LLMs, before scaling a workload, or when a cost estimate is needed.
中文简介正在准备中。当前显示原文说明。
Challenge vision alignment, value, cost, risk and operating ownership at material scope or release decisions.
中文简介正在准备中。当前显示原文说明。