| name | dispute-game-investigator |
| description | Investigate an OP Stack fault dispute game during (or after) an incident — answer the user's questions and diagnose: why a challenger is disagreeing, doing a lot of moves, or contradicting itself; whether a proposal is valid; which op-node is at fault; and (after the fact) who wins the bonds. Read-only. |
Dispute Game Investigator
When to Use
- During an incident, to answer questions and diagnose: a challenger disagreeing
with a proposal or another challenger, performing many moves, or contradicting itself;
whether the in-progress proposals are valid; which op-node is responsible.
- After the fact, for specifics like the bond outcome.
How to use it
Answer the user's actual question using only the relevant parts of the guide — you do
not need to run every step. Triage first (which game(s)? whose node? valid or not?) and pull
in only what bears on the question; some parts are situational (e.g. the bond outcome in §7 is
usually a post-incident question, not live triage). For a full end-to-end written analysis,
use the dispute-game-investigator agent instead.
Read-only: investigate and explain; never make moves, resolve, fund, or restart a challenger.
Guide
@docs/ai/dispute-game-investigation.md
Tooling
op-challenger list-games/list-claims --format json — enumerate games/claims.
op-chain-ops/cmd/check-output-root and check-super-root — canonical output root (EL-only, work against public nodes).
op-challenger game-proposal-outputs — per-game outputAtBlock +
safeHeadAtL1Block from a node (spots an incomplete safe-head DB).
op-challenger/scripts/check-game-block-hashes.sh — block-hash cross-check, node vs reference.