mike-boost/java-man
Overview
This skill performs ops-grade, evidence-backed profiling for Java projects and produces two commercial deliverables: PROJECT_PROFILE.md and DEPLOYMENT_RUNBOOK.md. It is tool-agnostic, avoids hallucination, and prioritizes business flows, integrations, and Kubernetes readiness. The output is SRE-facing and explicitly cites evidence or marks Unknown when repo proof is missing.
How this skill works
The skill first absorbs two baselines—java-architect (architecture) and java-reviewer (Java quality signals)—then scans the repository in progressive passes to map tech stack, entrypoints, integrations, configurations, critical flows, observability, and deployment needs. Every non-trivial claim is anchored to file paths or code excerpts; if evidence is absent it is labeled Unknown or Hypothesis with the supporting chain. Finally it emits PROJECT_PROFILE.md and DEPLOYMENT_RUNBOOK.md (or prints them) inside analyze-report-YYYY-MM-DD.
When to use it
- Before onboarding a Java service to platform/Kubernetes for the first time.
- As a pre-production readiness check during the release pipeline.
- When documenting operational responsibilities for cross-team handoff.
- To triage integration risk before dependency upgrades or migrations.
- As an evidence-backed acceptance artifact for platform approvals.
Best practices
- Run absorptions of java-architect and java-reviewer first; mark Unknown if inaccessible.
- Treat tests as reference only; never infer production behavior solely from tests.
- Attach explicit evidence anchors for each claim (file path + symbol or excerpt).
- Label plausible but unproven items as Hypothesis (Evidence-based) and include the chain.
- Keep reports concise: surface business flows first, then ops details and clear Unknowns.
Example use cases
- Produce a platform-friendly runbook and readiness score before Kubernetes onboarding.
- Assess external dependencies and required connection checks for SRE playbooks.
- Create a deployment checklist and smoke tests tied to critical business flows.
- Trace top 3–5 business flows end-to-end to prioritize monitoring and SLIs.
- Generate a config inventory and secrets map for security review and vault onboarding.
FAQ
The skill marks the corresponding baseline as Unknown (skill not accessible) and proceeds using repository evidence only; the final report includes an explicit Unknowns section and the self-verification checklist entry.
How are uncertain findings reported?
Unproven but plausible items are labeled Hypothesis (Evidence-based) with the evidence chain; weak or indirect signals are labeled Unverified and not treated as fact.
