Add prompt-engineer to Development & Code Tools - #1853

Adds one entry to Development & Code Tools, inserted alphabetically.

What it is
prompt-engineer treats prompt engineering as a procedure rather than a document. Before writing anything it decides which artifact the instruction belongs in — a hook, a skill, a subagent or a prompt — because placement is what decides whether an instruction takes effect at all, and it is the most expensive thing to get wrong in an agent harness. It then sizes an eval set to the actual volume of the task and ships the tests alongside the prompt.

Why it is not a duplicate of the existing prompt-engineering entry
That entry teaches known techniques and patterns. This one is a working toolchain: four modes (create, revise, migrate, execute), a regex linter for the mechanically detectable anti-patterns, an offline-testable eval runner, and a flattener that turns the skill into a single system prompt for platforms without a skill mechanism.

Against the checklist in CONTRIBUTING.md

• Real use case: built for and used in daily agent work, not a hypothetical.
• Tested: scripts/lint_all.sh reports all skill files clean and evals/run_evals.py --self-test passes offline with no API calls. The skill passes its own linter.
• Safe: read-and-write of prompt files only, no destructive operations.
• Portable: Claude Code skill, with the flattener covering platforms that have no skill mechanism.
• Licence: MIT.

🤖 Generated with Claude Code