py-shiny: Skill for Claude Code

Memory Bank

The memory-bank/ directory at the repo root holds contributor-facing documentation for py-shiny: architecture, conventions, and step-by-step recipes. Each file covers one subsystem or workflow and starts with YAML frontmatter containing title and description.

This is distinct from shiny/.agents/skills/, which ships with the package and documents shiny's public API for app authors. See memory-bank/agentSkills.md for that boundary.

Finding and reading documents

Loading this skill is preparation, not completion of documentation discovery. Before exploring code, identify the relevant memory-bank docs and read their applicable sections.

List the catalog (titles and descriptions) with:

# posit-dev/assistant uses a list_memory_bank.mjs script for this catalog and
# for per-file heading outlines with line ranges and byte sizes. With only a
# handful of small docs here, grep does the same job; port that script if the
# memory bank grows large enough that outlines save real context.
grep -A1 -m1 '^title:' memory-bank/*.md

• Read small, relevant documents directly. For larger ones, list their headings first (grep -n '^#' memory-bank/<file>.md) and read only the sections you need.
• If titles and descriptions don't point at anything, search contents with task terms, related symbols, and synonyms (grep -rni <term> memory-bank/) before concluding no guidance exists.
• Follow source-code references to verify current behavior; docs can lag the code.

Updating documents

Update a memory-bank file when:

• You discover an architectural pattern or pitfall worth recording
• You make a significant change to something a doc describes
• The user asks to update the memory bank

Keep the frontmatter title and description in sync with the content.

Creating new documents

1. Use camelCase naming (e.g., newFeatureSystem.md)
2. Add YAML frontmatter with title and description; quote values that contain :
3. If the doc should be read for a common task, add a pointer to it in AGENTS.md