원클릭으로
fase-completa
Full phase close protocol — tests, clippy, fmt, benchmarks, docs, memory update, commit, push
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
메뉴
Full phase close protocol — tests, clippy, fmt, benchmarks, docs, memory update, commit, push
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Verify and fix documented gaps one at a time — reproduce, root cause, minimal fix, wire-test regression, close
Discover undocumented gaps by comparing AxiomDB against MySQL/PostgreSQL — build inventory, run tests, classify, hand off to hunt-gap
Run Criterion micro-benchmarks and 3-Docker comparison benchmarks, verify no regression against MySQL/PostgreSQL
Explore approaches before proposing — read context, ask questions, present 2+ options with trade-offs, write sprint with dependencies
Save session context to a checkpoint file so the next session can resume without losing state
Systematic debug protocol — reproduce with minimal test, 2+ hypotheses, fix root cause, add regression test
| name | fase-completa |
| description | Full phase close protocol — tests, clippy, fmt, benchmarks, docs, memory update, commit, push |
Run this COMPLETE protocol when finishing each phase. No exceptions.
Important: the commands below are the final gate. Do not run the full workspace suite after every edit during implementation; use targeted crate tests first and reserve the full sweep for this close step.
# Final tests
cargo test --workspace
# If it fails: DO NOT continue until they pass
# Final linting
cargo clippy --workspace -- -D warnings
# If there are warnings: DO NOT continue until resolved
# Final format check
cargo fmt --check
# If there are differences: run cargo fmt and add to the commit
# Unsafe documentation
grep -r "unsafe" crates/ --include="*.rs" | grep -v "// SAFETY:"
# If there are unsafe blocks without SAFETY: add the comment
cargo bench --workspace 2>&1 | tail -30
# Compare against budget in CLAUDE.md
# If any critical operation regressed >5%: investigate before continuing
Create docs/fase-N.md with this template:
# Phase N — [Name]
Completed: [date]
Estimated weeks: N-M | Actual weeks: X
## What was built
[description of what was implemented]
## Crates created/modified
- `crates/axiomdb-X` — [what it does]
## Decisions made
- [decision] → [reason]
## Tests written
- [test] — [what it verifies]
## How to continue from here
[exact instruction for the next session]
## Next phase
Phase N+1 — [name]. See `specs/fase-N+1/` and `db.md`.
memory/project_state.md:
- Move phase N from "pending" to "completed"
- Update "Current phase" to N+1
memory/architecture.md:
- Add the actually implemented crates and structs
- Update the directory tree
memory/decisions.md:
- Add technical decisions made during the phase
memory/lessons.md (if there were learnings):
- ### [Phase N] Learning title
- Problem, cause, solution, when to apply
git add -A
git commit -m "feat(fase-N): [concise description]
- [detail 1]
- [detail 2]
Phase N/34 completed. See docs/fase-N.md
Spec: specs/fase-N/ | Tests: X passing"
git push origin main
Report:
✅ Phase N completed
Tests: X/X passing
Clippy: 0 warnings
Benchmarks: [within/outside] budget
Documented in: docs/fase-N.md
Memory: updated
Commit: [hash]
Next phase: N+1 — [name]
To continue: /brainstorm on Phase N+1