Fix descriptor registry — maps error codes to auto-fix descriptors.
Pure functions — no I/O, no side effects.

Fix descriptors describe WHAT to fix, not HOW. The shell layer
(boundary.devtools.shell.auto-fix) handles execution.

Pure functions for config diffing, dependency analysis, and formatting.
FC/IS: no I/O, no logging.

Pure documentation catalog for in-REPL exploration.
Provides (doc :topic) data for Boundary concepts with examples.
No I/O, no logging — pure data and transformations.

Classify exceptions into BND-xxx error codes.
Pure functions — catalog data loaded once at namespace init via boundary.devtools.error-codes.

Classification strategy (ordered, first match wins):
1. ex-data with :boundary/error-code — direct BND code
2. ex-data pattern matching — infer from :type, :schema, :malli/error
3. Message pattern — regex on .getMessage()
4. Exception type — SQLException, ConnectException, etc.
5. Unclassified — nil code

Enrich classified errors with stacktrace, suggestions, fix info, and URLs.
Pure functions — no I/O, no side effects.

Each sub-call is wrapped in try/catch for self-protection:
if any enrichment step fails, that field is omitted rather than
crashing the pipeline.

Rich error output formatting for development mode.
Pure functions — catalog data loaded once at namespace init via boundary.devtools.error-codes.

Pure functions for the guidance engine.
Renders contextual help messages, startup dashboards, and tips.
No I/O, no logging — all pure data transformations.

Pure formatting functions for REPL display of routes, config, and modules.
No I/O, no side effects — data in, formatted strings out.

Pure functions for mapping prototype specs to scaffolder contexts
and migration definitions.
Delegates to scaffolder's template functions for context building
to ensure generated output matches what generators expect.

Pure functions for recording session data structures.
No I/O, no atoms — just data transformations.

Pure functions for manipulating Reitit route data structures.
All functions take and return route trees (vectors of [path handler-map]).

Pure functions for Malli schema exploration: pretty-printing, diffing, and example generation.
No I/O, no side effects — schema in, formatted strings or data out.

Pure analysis of security configuration.
FC/IS: no I/O, no logging.

Stack trace filtering and reordering for development error output.
Pure functions — no I/O, no side effects.

Pure functions for analyzing project state.
Determines 'what should you do next?' based on project state data.
No I/O — accepts pre-collected state data as arguments.