fsi-architecture-assessment

Overview

This skill assesses a financial services customer's architecture diagram against the AWS Well-Architected Financial Services Industry (FSI) Lens (January 2026 update, including Gen AI and Agentic AI guidance). It ingests a diagram, contextualizes it for the stated FSI use case, rates maturity across the six Well-Architected pillars, and produces a prioritized, actionable improvement report. Use it for FSI Well-Architected reviews, compliance gap analysis, and architecture improvement planning.

Workflow

<Workflow - Assess Architecture description="Ingest a diagram, contextualize it for the FSI use case, assess against the FSI Lens pillars, and produce a prioritized report." tools=[get_current_time, file_read, file_read_image, file_read_pdf, file_write, web_search, url_fetch, open_in_session_tab] triggers=["assess my architecture against the FSI Lens", "run a Well-Architected FSI review", "check this diagram for FSI compliance gaps", "review my financial services architecture"]

1. [Agent] Ingest and interpret {{architecture_diagram}}. Use file_read_image for an image or file_read_pdf for a PDF. Extract a structured inventory: AWS services, network topology (VPCs, subnets, Availability Zones, Regions, on-premises links), data flows, security boundaries, high-availability patterns, any AI or machine learning components, external integrations, and monitoring components. Validate: The inventory covers at least compute, storage, networking, security boundaries, data flow directions, and any AI or machine learning components. If fails: If the diagram is unclear, low-resolution, or too high-level, ask the user for a higher-quality version or a text description of the architecture.

2. [Agent] Contextualize for the FSI use case. Read references/fsi-regulatory-map.md and map {{use_case}} to its regulatory regime, data sensitivity classification, RTO and RPO expectations, and business criticality. Validate: The context names a regulatory regime, a data classification, RTO/RPO expectations, and a business criticality level. If fails: If the use case is ambiguous or spans multiple jurisdictions, ask the user to clarify the specific regulatory environment and data sensitivity.

3. [Agent] Assess against the FSI Lens pillars named in {{focus_pillars}} (all six if "all" or unset). Read references/fsi-lens-pillars.md and evaluate the inventory against each pillar's best-practice questions. Optionally use web_search and url_fetch to confirm current FSI Lens guidance, citing any source consulted. Validate: Each assessed pillar has a maturity rating, identified strengths, identified gaps, and a named FSI Lens best-practice reference per gap. If fails: If a pillar cannot be assessed from the available information, mark it Insufficient Information and list the additional details needed.

4. [Agent] Call get_current_time, then generate the report. Read assets/report-template.md and fill it with the executive summary, architecture overview, per-pillar scorecards, prioritized roadmap, regulatory notes, and references. Use the maturity labels Well-Aligned, Partially Aligned, Gaps Identified, or Insufficient Information. Validate: The filled report contains every template section and the roadmap holds no more than 10 items. If fails: Rebuild the missing sections from the Step 3 findings before continuing.

5. [Decide] Where should the report be saved?

   • {{output_directory}} provided -> write the report there as a Markdown file with a dated filename.
   • Not provided -> [Ask user] Ask for a destination directory, then write it there. Validate: The report file is written and non-empty. If fails: If the write fails, present the full report content directly in chat.

6. [Agent] Open the report with open_in_session_tab and summarize the key findings in chat. Offer follow-up actions: deep-dive into one pillar, draft a remediation architecture description, build a compliance mapping for specific regulations, or export the report to another document format. Validate: The report opens in a session tab, or the summary is presented in chat. If fails: Present the full report content directly in chat.

</Workflow - Assess Architecture>