ss-bs-discovering-architecture

To generate an observed architecture map, the skill scans repository structure, build configuration, entry points, runtime topology, data stores, and integration signals to produce a factual Markdown overview without ungrounded speculation.

Documenting data stores and integration discovery involves scanning dependency files, IaC, and integration signals, then using interactive single-question confirmations to resolve grouping and relationship cardinality without making assumptions.

Yes, you can extend existing architecture documentation safely; extend mode reads current architecture content and merges new findings without overwriting existing accurate content, writing the final artifact atomically.

Yes, architecture mapping supports audit mode by performing drift checks against existing documentation, identifying gaps and provenance markers while ensuring the evidence-bounded code scan matches reality.

The main limitation is that the output is an evidence-bounded architecture overview containing no diagrams and no ungrounded speculation; it strictly maps observed layout and runtime topology facts into a Markdown artifact.

You resolve human-knowledge gaps during repository scanning by answering guided, single-question-at-a-time confirmations that help the system accurately map boundaries and relationship cardinality without guessing.