Custom Agent Remediation Plan

Workflow

Use only after verified audit findings identify remediation gaps.

1. Group findings into content, product knowledge, crawler access, policy, integration, and workflow buckets.
2. Decide what the custom agent can improve directly versus what needs storefront/platform changes.
3. Define required sources of truth: catalog, policies, support inbox, order system, CMS, analytics exports when provided.
4. Produce a setup checklist with acceptance tests.
5. Define how score lift or readiness improvement will be verified.

Run the checks in references/checks.md against the audit findings and proposed plan before presenting it.

Classify every remediation item as agent, storefront, or shared delivery. Also classify it as action_capable or read_only and assign a low, moderate, or high risk level. Name an accountable owner, source of truth, observable acceptance test, baseline check, and post-change check for every remediation item.

For every action_capable item, and for any high-risk item, define operational controls covering: