Load the Use Cases section of .github/instructions/metadata.instructions.md
into the --describe prompt so generated use-cases match the repository
style, and make useCase an array of phrases per usage group.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Q2D9GYbnSV56MAVVhHPRhH
Retry 408 responses (and other transient errors) with backoff, only waiting
for the rate-limit reset on 403/429. Requests that still fail are skipped
with a warning instead of aborting the run, and the OpenAI client retries
too. Each icon is now described as soon as its searches finish and
--output is rewritten after every icon, so an interrupted run keeps its
results.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Q2D9GYbnSV56MAVVhHPRhH
Add --context and --describe to the icon usage script. --context extracts
how each matched icon is used (parent elements, labels, handlers, links,
snippet) and summarizes it per icon. --describe groups those usages into
use-case descriptions with an OpenAI model. Also adds --max-contexts and
--icons-file, accepts export names like Trash2, and allows a single icon.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Q2D9GYbnSV56MAVVhHPRhH
Claude