← ClaudeAtlas

frame-audit-scopelisted

Define a bounded, evidence-aware audit scope for automations, AI agents, workflows, or connected business systems. Use when starting an audit preparation, deciding what local evidence may be inspected, recording exclusions, or narrowing an open-ended assessment request.
kotnisofiane-bit/dubsar-agent-skills · ★ 0 · AI & Automation · score 68
Install: claude install-skill kotnisofiane-bit/dubsar-agent-skills
# Frame Audit Scope ## Objective Create an attributable, reviewable scope before collecting new evidence or drawing conclusions. ## Inputs - the business question and candidate systems, automations, agents, and processes; - proposed local evidence sources and how the user supplied them; - exclusions, time window, limitations, and completion criteria; - an explicit output path; - approval attribution only when the user actually approves the scope. ## Workflow 1. Restate the business objective in one sentence. 2. List the systems, automations, agents, and business processes explicitly in scope. Do not add systems merely because they probably exist. 3. Record the approved evidence sources and how each source was provided. 4. Record excluded systems, data, actions, and time periods. 5. Mark every uncertain statement as `unknown` or `reported`, never `observed`. 6. Define what would make the preparation complete enough for human review. 7. Ask the user to approve the scope before suggesting evidence collection. Record a role or pseudonymous identifier, an ISO 8601 UTC timestamp, and a local approval reference. Do not invent these values. ## Output Produce `audit-scope.json` with this stable shape: ```json { "format": "dubsar.audit-scope/1", "case_id": "user-chosen-or-generated-local-id", "objective": "bounded business question", "in_scope": [], "approved_evidence": [], "excluded": [], "time_window": null, "completion_criteria": [], "limitation