PPromptReports
Sign in
Use cases/SOC 2 and EU AI Act evidence from the record/OpenSpec plan

Plan: SOC 2 and EU AI Act evidence from the record

Plannedopenspec/changes/wf-audit-evidenceLDG-2289

Written for an AI agent to fetch and follow: it says what to connect, which steps to build with which product, what a person must approve, and how to check the result. Steps that depend on a part that is not built say so, and tell the agent to record the gap instead of guessing.

Raw URL: https://promptreports.ai/use-cases/audit-evidence/openspec.md

URL pattern: /use-cases/<id>/openspec.md, all plans at /use-cases/index.json. Checkboxes on this page are for reading only.

openspec/changes/wf-audit-evidence/proposal.md

Why

Auditors ask for logs of what AI systems did and proof that people oversaw them. Teams export them by hand, every year. The evidence collects itself as the work happens, so the audit is a review instead of a scramble.

What Changes

  • Add the audit-evidence workflow to the PromptReports workflow library, run by the workflow runner as an agent acting for a person.
  • PromptCompute: Read the signed approval ledger (today: Built).
  • RepoOps: Read tasks, events and outcomes (today: Built).
  • PromptReports: Map events to controls (today: Planned).
  • PromptCompute: A person approves the export (today: Built).
  • AI Source of Truth: Record the export's hash (today: Proposed).
  • Gate data.export.bulk through PromptCompute approvals; each gated step runs only after verifyResume passes, with final_args.
  • Write a signed record at the end of each run and count one workflow request.

Impact

  • Products: PromptCompute, RepoOps, AI Source of Truth, PromptReports.
  • Outside the family: Your auditor's evidence portal (optional).
  • State: Planned. In the plan as compliance-mapping (LDG-2289). The mappings to SOC 2 change management and EU AI Act Articles 12, 19 and 50 are Planned. This is evidence for an auditor, not a certification.
  • Plan row: LDG-2289.
openspec/changes/wf-audit-evidence/tasks.md

1. Connect and authorize

  • 1.1 Confirm the org has a connection for each of: PromptCompute, RepoOps, AI Source of Truth, PromptReports. Stop and tell the user which is missing.
  • 1.2 Get a PromptReports ID agent token for each product the run calls (RFC 8693 exchange). Where a product does not accept ID tokens yet, use its connect key.
  • 1.3 Register policy for data.export.bulk with require_approval as the default.

2. Build the steps

  • 2.1 PromptCompute: Read the signed approval ledger.
  • 2.2 RepoOps: Read tasks, events and outcomes.
  • 2.3 PromptReports: Map events to controls. Blocked: this part is Planned; until it ships, record the step's evidence as missing and continue.
  • 2.4 PromptCompute: A person approves the export.
  • 2.5 AI Source of Truth: Record the export's hash. Blocked: this part is Proposed; until it ships, record the step's evidence as missing and continue.

3. Verify and record

  • 3.1 Run a dry run with the user's inputs: Framework = SOC 2 change management; Period = 2026 Q4.
  • 3.2 Run it once for real; approve each gate as the user.
  • 3.3 Measure the outcomes: Controls with evidence attached; Hours of manual evidence collection per audit.
  • 3.4 Confirm the run left a signed record with a task id and one workflow request counted.
openspec/changes/wf-audit-evidence/specs/workflow-audit-evidence/spec.md

ADDED Requirements

Requirement: Each control cites events

Each mapped control SHALL list the signed events that support it, or say none.

Scenario: Each control cites events

  • WHEN a control has no supporting events
  • THEN the export lists it as a gap, not as passed

Requirement: Evidence stays honest

Each step SHALL report what it saw as measured, inferred or missing, and SHALL never report missing as measured.

Scenario: A product part is not built

  • WHEN a step depends on a part that is Planned or Proposed
  • THEN the step records missing, names the part, and the approval card shows the gap

Requirement: One run is one workflow request

The runner SHALL count one workflow request when the first step starts, and SHALL NOT count retries, resumes or dry runs.

Scenario: A gated step resumes after two days

  • WHEN a person approves a step and the run resumes
  • THEN the org's usage does not change