rulesets/.claude/commands/prompt-engineering.md, branch main

docs(commands): fix prompt-engineering citation and add an eval harness

2026-05-22T19:46:02+00:00

Two audit fixes. The Meincke citation had the wrong title and was used to imply persuasion framing improves prompt quality. It now reads as the safety caution it is: applying the principles raised an LLM's compliance with objectionable requests from ~33% to ~72%, a reason for care, not a recipe. The correct title ("Call Me A Jerk...") and SSRN id are fixed in all three spots. Critique mode also gains an eval-harness step: for fragile or production prompts, run 3-5 adversarial examples against the old and new prompt and record the delta, so quality is verified rather than asserted.

chore(commands): mark user-invoked commands disable-model-invocation

2026-05-07T02:59:52+00:00

Add disable-model-invocation: true to the user-triggered slash commands so the harness drops their descriptions from the model's preloaded skill listing while keeping / routing intact. Skills meant for model recommendation (add-tests, debug, five-whys, frontend-design, humanizer, pairwise-tests, playwright-js, playwright-py, root-cause-trace) are unchanged.

fix(commands): drop leftover name: frontmatter from converted commands

2026-05-06T11:38:24+00:00

The conversion commit ran sed -i to strip the name: line, but the staged version from git mv was captured before the sed ran. Result: the commit shipped the original frontmatter, and the sed cleanup was sitting in the working tree. This commit lands the cleanup that should have been in aa69245.

refactor(skills): convert 16 user-invoked skills to commands

2026-05-06T11:17:08+00:00

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.