# Investigate

> Investigate a failure with bounded hypotheses, cheap-first probes, and an evidence-bound conclusion.

- Skill: `operatorstack/investigate` (Agent Skill, multi-file: 4 files)
- Install (CLI): `npx skillmds@latest add operatorstack/investigate`
- Raw SKILL.md: https://api.skillmd.com/api/skills/operatorstack/investigate/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: operatorstack (https://skillmd.com/u/operatorstack)
- Updated: 2026-09-21
- Page: https://skillmd.com/skills/operatorstack/investigate

---


Run:

    yskill run .

Follow each returned operation exactly.

- `ask_user`: ask the user using the host's normal interface.
- `agent_task`: perform the task and return schema-valid JSON.
- `run_command`: yskill executes it itself; you will not see this kind.

Resume the run after each operation:

    yskill resume <run-id> --response response.json --skill .

Do not skip an operation or invent its response. The program bounds the
investigation: at least three hypotheses, cheapest-to-disprove first, at
most three failed attempts, and completion requires a causal chain.