返回 Skill 列表
extension
分类: 开发与工程无需 API Key

cto-audit

进行深入的、专家级的代码库和架构审核,以识别技术优势、劣势、风险和机会。当用户请求评估代码库的结构、质量或扩展准备情况时使用。提供基于工程最佳实践的具体、可操作且优先排序的建议。

person作者: jakexiaohubgithub

CTO Audit

Overview

Produce a rigorous, CTO-level audit of a software system, focusing on architecture, implementation quality, performance, security, and operational readiness. Deliver evidence-backed findings with clear priorities and pragmatic improvements.

Audit Workflow

  1. Clarify scope and constraints

    • Identify repositories, services, environments, and time constraints.
    • Capture business goals, scale targets, SLAs/SLOs, and regulatory constraints.
    • Ask for missing inputs (architecture docs, infra diagrams, deployment details).
  2. Map the system

    • Inventory components: services, data stores, queues, third-party dependencies.
    • Trace critical flows: request path, data lifecycle, and failure modes.
    • Identify system boundaries and ownership.
  3. Inspect architecture and design

    • Evaluate modularity, separation of concerns, and dependency direction.
    • Identify overengineering, premature abstraction, or tight coupling.
  4. Assess implementation quality

    • Verify correctness in critical paths; check error handling and edge cases.
    • Identify incomplete features, TODOs, stubs, or silent failure points.
    • Note code smells that degrade maintainability.
  5. Evaluate performance and efficiency

    • Identify hot paths, heavy I/O, N+1 patterns, and memory pressure.
    • Assess scalability limits (CPU, database, cache, network).
    • Recommend targeted optimizations with expected impact.
  6. Review security and reliability

    • Check input validation, authn/authz, secrets handling, and data exposure.
    • Evaluate resiliency: retries, timeouts, backoff, circuit breaking, idempotency.
    • Note incident response readiness and observability coverage.
  7. Review tooling, testing, and ops

    • Inspect test coverage and critical gap areas.
    • Check CI/CD, build reproducibility, dependency hygiene, and release safety.
    • Assess runtime monitoring, alerting, and runbooks.
  8. Synthesize and prioritize

    • Group findings by severity and theme.
    • Highlight what works well and why.
    • Produce a prioritized roadmap of fixes and upgrades.

Evaluation Dimensions

  • Architecture and system design
  • Implementation quality and correctness
  • Performance and efficiency
  • Code quality and maintainability
  • Security and reliability
  • Tooling, testing, and DevOps
  • Elegance and simplicity

Evidence Standards

  • Cite concrete evidence: file paths, line numbers, configs, or logs.
  • Separate confirmed issues from hypotheses; state assumptions clearly.
  • Avoid ungrounded claims; request missing data when needed.

Output Format

Summary:

  • State purpose and scope.
  • Summarize the system overview (1-2 sentences).
  • List key strengths (2-4 bullets).
  • List top risks (2-4 bullets).

Findings (ordered by severity):

  • Use severity buckets: Critical, High, Medium, Low.

For each finding, include impact, evidence, and a crisp fix direction.

Recommendations (prioritized):

  1. List immediate fixes (safety and correctness).
  2. List near-term improvements (quality, performance, maintainability).
  3. List strategic investments (architecture, platform, scale).

Overall Assessment:

  • State readiness for scale and operational risk.
  • Provide a short rationale and confidence level.