| |
|
|
|
|
|
|
| |
I converted 16 user-invoked skills to commands. Skills cost ~150-300 tokens each per session for descriptions the model uses to auto-route. Commands cost nothing until you type the slash. These 16 are workflows I always trigger deliberately. The auto-routing wasn't earning its keep. This reclaims ~4-5k tokens per session.
Nine skills stayed where auto-routing genuinely helps: debug, root-cause-trace, five-whys, add-tests, frontend-design, humanizer, playwright-js, playwright-py, and pairwise-tests. Pairwise-tests stays a skill because its helper files don't fit a single-file command shape.
For arch-decide, I preserved the upstream MIT LICENSE alongside the command at .claude/commands/arch-decide.LICENSE so attribution stays intact.
|
|
|
Scans a file for cj: annotations (Craig's in-line instructions and questions) and handles each one with subagent-delegated accuracy. Classifies each comment as instruction or question, spawns subagents per item, applies changes from the main thread, and produces a structured summary. Covers the voice and attribution discipline for any public-facing writing the skill produces, a long-summary-to-file pattern that writes to /tmp/ and opens in emacsclient when the output runs long, and a requirement that every file path in the summary is a clickable org-mode link.
|
