What's New
Added top-level # Execute decision-tree summaries to 6 core skill files. Each
block distills the skill's action sequence into a scannable 5-8 line summary,
allowing agents to locate the right action without reading through full
reference prose. Removed narrative duplication, motivational text, and
redundant diagrams that added cognitive load without improving decision quality.
Changes
| File | Before | After | Reduction |
|---|---|---|---|
| verification-before-completion | 237 lines | 68 lines | -71% |
| subagent-driven-development | 220 lines | 88 lines | -60% |
| writing-plans | 185 lines | 75 lines | -59% |
| systematic-debugging | 285 lines | 166 lines | -42% |
| test-driven-development | 372 lines | 235 lines | -37% |
| brainstorming | 153 lines | 99 lines | -35% |
| Total | 1,452 lines | 712 lines | -51% |
Preserved
- All rule semantics intact
- Dual-track governance (Repair Track + Retirement Track + Residual Risk)
unchanged - All compliance-check phrases passing across the full test suite
- Zero new dependencies, zero API changes
Verification
bash tests/e2e/layer1-fast-check.sh --host-profile none
bash tests/e2e/context-budget-check.sh
bash tests/e2e/boundary-compliance-check.sh
bash tests/e2e/governance-completion-contract-check.sh
bash tests/e2e/tdd-policy-check.sh
python tests/helpers/test_parse_codex_skills.py