원클릭으로
irp-capture
Capture, recall, and check durable decision lineage inside any project using IRP (Intent Record Protocol).
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
메뉴
Capture, recall, and check durable decision lineage inside any project using IRP (Intent Record Protocol).
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
| name | irp-capture |
| description | Capture, recall, and check durable decision lineage inside any project using IRP (Intent Record Protocol). |
IRP preserves durable human decisions, rationale, and reasoning lineage inside a local project.
Use capture when a decision crystallises during a Claude session. The reasoning is sharpest at the moment of decision — capture it before moving on.
Use irp why before starting new work. It shows what was already decided so nothing gets relitigated.
Use irp check before proposing something new. It catches conflicts with active decisions.
Think of IRP as a camera, not a notebook. Capture only what should survive session boundaries and matter later.
The canonical IRP state lives in the project root:
.irp/
├── ledger.jsonl ← append-only source of truth
└── current.json ← derived active state (last 10 decisions)
These are plain text files. They travel with the repo. No service required.
irp capture "Decision: Use Postgres for the reporting service" \
--why "Redis rejected — query patterns require joins"
irp why
Shows the last 10 confirmed decisions with reasoning.
irp why --id IRP-2026-04-08-001
irp inherit
Lists all active decisions — use this to understand the current state of the project before making changes.
irp check "Add a SQLite database for reporting"
Returns whether the proposal conflicts with any active decision. Use this before proposing architectural changes.
Add --json to any command for machine-readable output:
irp why --json
irp check --json "Add a SQLite database"
pip install irp-capture
The irp CLI is immediately available after install. No configuration needed.
A good decision entry answers two questions:
The second question is the valuable one. Anyone can find what you chose. The rejected alternatives are what gets lost.
Threshold: Capture decisions that would be hard to explain six months later. If you cannot write one sentence of why, the decision may not be ready to capture yet.