Problem you're trying to solve
edit_document accepts "target": { "find": "Acme Corp" } and resolves it against live content. When the text matches more than once, FindTargetResolver refuses the operation and lists the candidates so the agent can resend with "match": <index>.
Right now agent is not able distinguish between matches from different part of document.
Proposed shape
Include the location in each candidate when the hit has one, and leave the line unchanged when it does not.
Problem you're trying to solve
edit_documentaccepts"target": { "find": "Acme Corp" }and resolves it against live content. When the text matches more than once,FindTargetResolverrefuses the operation and lists the candidates so the agent can resend with"match": <index>.Right now agent is not able distinguish between matches from different part of document.
Proposed shape
Include the location in each candidate when the hit has one, and leave the line unchanged when it does not.