feat(skills): add skill-creator scripts for benchmarking, optimization, and scaffolding

[JeremyDev87]

Summary

• Add aggregate_benchmark.py: aggregates grading.json + timing.json from eval directories into benchmark.json and benchmark.md reports
• Add run_loop.py: interactive description optimization loop with 60/40 train/test split, precision/recall/F1 metrics, and per-iteration scoring log
• Add init_skill.sh: skill directory scaffolding with SKILL.md template (Core principle, When to Use structure), kebab-case validation, and overwrite prevention

All scripts use Python 3.8+ standard library only (no external dependencies) and support --help.

Test plan

• aggregate_benchmark.py --help displays usage
• aggregate_benchmark.py with mock iteration data produces valid benchmark.json (schema-conforming) and benchmark.md
• run_loop.py --help displays usage
• run_loop.py with missing file returns proper error
• init_skill.sh --help displays usage
• init_skill.sh creates SKILL.md + references/ + examples/ + scripts/ directories
• init_skill.sh rejects invalid names and prevents overwriting existing directories
• ajv agent schema validation passes
• markdownlint passes

Closes #742

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
codingbuddy-landing	Ready	Preview, Comment	Mar 21, 2026 1:54pm