Require fresh, scope-appropriate evidence before any completion claim. Distinguishes implemented, locally verified, repository-green, PR-green, deployed, and observed-stable states.
Persistent HTTPS URL for local Gradio/Streamlit/FastAPI/webhook services without deploying. Use when the user wants a stable permanent URL on a domain they own (not a throwaway URL, not managed hosting).
When the user asks for a dev JWT, Bearer token, or to test / hit the linked Vercel deployment API with auth — run the smoke command and report status (not the token)
Release & deploy gate for @kadam-net/mcp-server — what the agent may do autonomously vs. what requires explicit human approval. Applies to all branch/MR/release/deploy work in this repo.
Full expert review protocol. 8-expert, 2-wave deliberation for comprehensive codebase and documentation review. Invoke before major milestones, after shipping feature clusters, or before staging/production deployment.
Full expert review protocol. 8-expert, 2-wave deliberation for comprehensive codebase and documentation review. Invoke before major milestones, after shipping feature clusters, or before staging/production deployment.
Helm chart engineering: chart/subchart layout, values schema validation, hardened Deployment templates (probes, resources, securityContext, PDB), and release lifecycle with rollback. Use when authoring or reviewing a Helm chart, templating Kubernetes manifests, or debugging a failed or stuck Helm release.
DORA metrics instrumentation: deployment frequency, lead time for changes, change failure rate, and time to restore, with PromQL dashboards, benchmark bands, and anti-gaming guidance. Use when measuring delivery performance, building an engineering-metrics dashboard, or diagnosing why throughput or stability is poor.
Decouple deploy from release with feature flags — ship code dark, turn it on deliberately, and delete the flag once it has served its purpose. Invoke when rolling out a risky or incomplete change, running an experiment, or gating access. Skip for a small change with no rollout risk, where a flag is just indirection.
GCP platform engineering: organization, folder and project hierarchy, Organization Policy constraints, Shared VPC, VPC Service Controls, IAM Workload Identity Federation, and BigQuery operations. Use when designing a Google Cloud resource hierarchy or network, letting GitHub Actions or another external CI deploy to GCP without a service account key, or building a data perimeter.
Database DevOps: expand/contract zero-downtime schema migration, migration-as-code with Flyway, Liquibase and Atlas, connection pooling, replication lag, and rollback strategy. Use when adding, renaming or dropping a column on a large Postgres or MySQL table without downtime, running migrations from a deploy pipeline, or diagnosing read replicas lagging behind the primary and serving stale data.
CI/CD pipeline architecture for GitHub Actions and GitLab CI: matrix testing, dependency caching, OIDC keyless cloud authentication, security gating, and concurrency control. Use when a pipeline is slow because every job reinstalls dependencies, when long-lived AWS or cloud access keys stored as CI secrets must be removed, or when building, gating and speeding up a build-test-deploy workflow.