# openspec/changes/wf-security-questionnaire/proposal.md

## Why

Questionnaires take days, the answers are copied from last year, and nobody can prove them. You answer in hours, and every answer holds up when the buyer checks it.

## What Changes

- Add the `security-questionnaire` workflow to the PromptReports workflow library, run by the workflow runner as an agent acting for a person.
- PromptCompute: Read approvals evidence (today: Planned).
- RepoOps: Read incident and recovery evidence (today: Planned).
- CommitOps: Read promises kept (today: Planned).
- AI Source of Truth: Check each drafted answer against the record (today: Built).
- PromptCompute: A person approves sending it (today: Built).
- Gate `file.share.external` 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, AI Source of Truth, RepoOps, CommitOps.
- Outside the family: The questionnaire file (spreadsheet or portal export).
- State: Proposed. Named in the research as later work. Builds on the trust packet.
- Plan row: none yet (Proposed).

# openspec/changes/wf-security-questionnaire/tasks.md

## 1. Connect and authorize

- [ ] 1.1 Confirm the org has a connection for each of: PromptCompute, AI Source of Truth, RepoOps, CommitOps. 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 `file.share.external` with `require_approval` as the default.

## 2. Build the steps

- [ ] 2.1 PromptCompute: Read approvals evidence. Blocked: this part is Planned; until it ships, record the step's evidence as `missing` and continue.
- [ ] 2.2 RepoOps: Read incident and recovery evidence. Blocked: this part is Planned; until it ships, record the step's evidence as `missing` and continue.
- [ ] 2.3 CommitOps: Read promises kept. Blocked: this part is Planned; until it ships, record the step's evidence as `missing` and continue.
- [ ] 2.4 AI Source of Truth: Check each drafted answer against the record.
- [ ] 2.5 PromptCompute: A person approves sending it.

## 3. Verify and record

- [ ] 3.1 Run a dry run with the user's inputs: Questionnaire = contoso-sig-lite.xlsx; Buyer = Contoso Health.
- [ ] 3.2 Run it once for real; approve each gate as the user.
- [ ] 3.3 Measure the outcomes: Hours from questionnaire received to answered; Answers with an evidence link; Questions flagged as cannot answer.
- [ ] 3.4 Confirm the run left a signed record with a task id and one workflow request counted.

# openspec/changes/wf-security-questionnaire/specs/workflow-security-questionnaire/spec.md

## ADDED Requirements

### Requirement: No guessed answers

An answer without evidence SHALL be marked cannot answer and left for a person.

#### Scenario: No guessed answers

- **WHEN** the record has nothing on data residency
- **THEN** the question is flagged, not drafted

### 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