← ClaudeAtlas

dubsar-audit-readinesslisted

Run the complete local audit-readiness workflow for automations or AI agents: scope, evidence index, inventory, sensitive actions, evidence gaps, validation, and deterministic export. Use when the user requests end-to-end preparation for human review; use a specialized skill for only one phase.
kotnisofiane-bit/dubsar-agent-skills · ★ 0 · AI & Automation · score 68
Install: claude install-skill kotnisofiane-bit/dubsar-agent-skills
# DUBSAR Audit Readiness ## Objective Prepare a coherent local evidence workspace for human review without claiming that the audited system passed. For a single step, prefer `frame-audit-scope`, `inventory-automations`, `map-sensitive-actions`, `review-evidence-gaps`, or `export-audit-bundle`. ## Inputs - an explicit local workspace path and, if export is requested, a separate output path; - user-approved local evidence and its source context; - the business question, scope, exclusions, time window, and completion criteria; - human decisions needed for scope approval and sensitive-action review. Do not invent approval attribution. Accept only a user-provided role or pseudonymous identifier, UTC timestamp, and local reference. ## Workflow 1. Read [the data contracts](references/data-contracts.md). 2. Initialize the workspace or inspect the five existing case files. 3. Frame the objective, approved evidence, exclusions, time window, completion criteria, limitations, and attributable scope approval in `audit-scope.json`. 4. Index approved regular local files and their SHA-256 values in `evidence-index.json`. 5. Inventory only evidence-supported automations and agents in `automation-inventory.json`. 6. Map material effects and proposed review points in `sensitive-actions.json`; require human review even when the map is empty. 7. Separate observations, reports, contradictions, gaps, and limitations in `evidence-review.json`; link every supported obser