qfai 1.0.5 → 1.0.6

package/README.md

@@ -132,6 +132,7 @@ Spec では `QFAI-CONTRACT-REF:` 行で参照する契約IDを宣言します（
 Scenario では `# QFAI-CONTRACT-REF:` のコメント行で契約参照を宣言します（`none` 可）。
 契約ファイルは `QFAI-CONTRACT-ID: <ID>` を **1ファイル1ID** で宣言します。
 契約IDは UI/API/DB/THEMA（THEMA は 3 桁）です。UI 契約は `themaRef` / `themeOverrides` / `assets` を追加できます。
+UI/API は YAML、DB は SQL（`.sql`）を正式フォーマットとして扱います。
 `assets.pack` は `ui/` 配下の相対パス、`assets.use` は `assets.yaml` の `items[].id` を参照します。
 `validate.json` / `report` の file path は root 相対で出力します（absolute は出力しません）。
 
@@ -240,6 +241,10 @@ qfai.config.yaml
       README.md
       constitution.md
       workflow.md
+      thinking.md
+      communication.md
+      quality.md
+      agent-selection.md
     steering/
       README.md
       product.md
@@ -271,8 +276,6 @@ qfai.config.yaml
       backend-engineer.md
       devops-ci-engineer.md
       code-reviewer.md
-tests/
-  qfai-traceability.sample.test.ts
 .github/
   workflows/
     qfai.yml

package/assets/init/.qfai/README.md

@@ -1,12 +1,13 @@
 # .qfai (QFAI Workspace)
 
-Generated by: `qfai init`  
-Version: v1.0.5 (proposed init structure)  
-Updated: 2026-01-12
+Generated by: `qfai init`
+Template version: 1.0.6
 
 This directory is the workspace for **QFAI (Quality‑First AI)**.  
 QFAI standardizes engineering work around a fixed workflow:
 
+Note: You may edit files under `.qfai/` to fit your project. QFAI prompts treat this workspace as project-local SSOT for guidance and artifacts.
+
 **SDD → ATDD → TDD → Implementation → Verification**
 
 ## Directory Overview

package/assets/init/.qfai/assistant/agents/README.md

@@ -32,3 +32,5 @@ Each role card includes YAML frontmatter:
 ## Response contract
 
 All subagents must use the "Findings / Recommendations / Proposed edits / Open Questions / Confidence" structure.
+
+Note: These are **role cards** (definitions). Actual delegation/execution depends on your AI product (Copilot, Claude Code, etc.). If subagents are not supported, the orchestrator prompt must emulate role-by-role reasoning in a single run.

package/assets/init/.qfai/assistant/agents/contract-designer.md

@@ -57,7 +57,7 @@ You are the **Contract Designer** in a QFAI-driven workflow.
 
 ## Expected Outputs
 
-- Contract file list + minimal YAML drafts.
+- Contract file list + drafts (**UI/API: YAML**, **DB: SQL**).
 - Rationale for each field.
 - Example payloads where helpful.
 

package/assets/init/.qfai/assistant/instructions/README.md

@@ -1,6 +1,10 @@
 # instructions/
 
-High-level governance for AI behavior.
+High-level governance for AI behavior in QFAI.
 
 - `constitution.md` : non-negotiable rules (quality, traceability, evidence)
 - `workflow.md` : the default QFAI workflow (SDD→ATDD→TDD→Implement→Verify)
+- `thinking.md` : ambiguity elimination + evidence-first reasoning
+- `communication.md` : reporting style + status format
+- `quality.md` : quality gates + testing expectations
+- `agent-selection.md` : how to delegate to roles (or emulate if unsupported)

package/assets/init/.qfai/assistant/instructions/agent-selection.md

@@ -0,0 +1,28 @@
+---
+id: agent-selection
+category: project
+update_frequency: occasional
+---
+
+# Agent Selection (Delegation playbook)
+
+## Goal
+
+Delegate work to specialized roles to reduce blind spots and improve quality.
+
+## Default delegation map
+
+- **Requirements Analyst**: clarify intent, scope, acceptance criteria, open questions
+- **Planner**: plan phases, risks, gating, rollback strategy
+- **Architect**: design, boundaries, compatibility considerations
+- **Contract Designer**: contracts (UI/API: YAML, DB: SQL), IDs, indexing implications
+- **QA Engineer**: risk-based checks, regression scope, quality gate review
+- **Test Engineer**: scenario.feature and test scaffolding strategy
+- **Front-end / Back-end Engineer**: implementation within repo conventions
+- **DevOps/CI Engineer**: verify-pack/CI impacts
+- **Code Reviewer**: style, maintainability, correctness
+
+## If subagents are not supported
+
+Emulate the delegation by doing role-by-role analysis in order:
+Requirements → Plan → Design → Contracts → Tests → Implementation → Review → QA.

package/assets/init/.qfai/assistant/instructions/communication.md

@@ -0,0 +1,27 @@
+---
+id: communication
+category: universal
+update_frequency: occasional
+---
+
+# Communication (Output and reporting contract)
+
+## Absolute rule — Output language
+
+All user-facing output must be in **the user's language**.  
+If multiple languages appear, choose the user's dominant language.
+
+## Reporting format (default)
+
+Use concise, structured bullet points:
+
+- **Summary**: what changed / decided
+- **Evidence**: key file paths / commands / logs referenced
+- **Impact**: user-visible changes, risks
+- **Verification**: what you ran and expected result
+- **Open Questions**: unresolved items (if any)
+
+## Error handling
+
+- Do not hide errors. Explain impact, scope, and recovery steps.
+- Avoid dumping excessive logs; show the minimum relevant excerpt.

package/assets/init/.qfai/assistant/instructions/quality.md

@@ -0,0 +1,27 @@
+---
+id: quality
+category: universal
+update_frequency: occasional
+---
+
+# Quality (Gates, tests, and safety)
+
+## Quality gates (baseline)
+
+When code changes are requested, the expected minimum gates are:
+
+- `pnpm format:check`
+- `pnpm lint`
+- `pnpm check-types`
+- `pnpm test`
+- `pnpm verify:pack` (when publishing/distribution matters)
+
+## Do not weaken safety nets
+
+- Do not suppress or disable checks via inline ignores unless explicitly approved.
+- If a rule must be changed, justify with evidence and update tests/docs accordingly.
+
+## Testing expectations
+
+- Prefer adding/adjusting tests over weakening validation.
+- Keep outputs deterministic (avoid time/randomness).

package/assets/init/.qfai/assistant/instructions/thinking.md

@@ -0,0 +1,29 @@
+---
+id: thinking
+category: universal
+update_frequency: rare
+---
+
+# Thinking (Evidence-first, ambiguity elimination)
+
+## Principles
+
+- Prefer **repo evidence** over assumptions (file paths, configs, tests, commands).
+- If something cannot be verified, write `TBD` and raise an Open Question (what evidence is missing).
+- Minimize ambiguity: define terms, scope, and measurable acceptance criteria.
+
+## Working method
+
+1. Restate the goal and constraints (brief).
+2. Enumerate unknowns and assumptions.
+3. Identify the evidence to check (files/commands).
+4. Decide with a rationale grounded in evidence.
+5. Record residual risk and the rollback path.
+
+## When to stop and ask
+
+Stop and ask the user if:
+
+- required inputs are missing (e.g., target behavior, API shape, UX intent),
+- multiple interpretations are plausible and impact is material,
+- a change could be breaking and intent is unclear.

package/assets/init/.qfai/assistant/prompts/qfai-implement.md

@@ -89,15 +89,6 @@ Simulate roles by running the same sequence yourself:
 
 - Write a short “role output” section per role, then consolidate into the final deliverable(s).
 
-## Behavior Rules (high leverage)
-
-- **Language**: Output MUST follow the user’s working language for this session.
-- **Question budget**: Ask at most **5** clarifying questions total. Prioritize blockers. If `--auto`, proceed with explicit assumptions.
-- **No hallucination**: Do not invent file paths, commands, or policies. Confirm via repo inspection.
-- **Evidence**: Do not claim completion without commands/results (format/lint/type/test/pack as applicable).
-- **Subagent contract**: When delegating, require the subagent response structure:
-  1. Findings 2) Recommendations 3) Proposed edits 4) Open Questions/Risks 5) Confidence
-
 ## Step 0 — Load Context (always)
 
 1. Read relevant **project steering** (if present):

package/assets/init/.qfai/assistant/prompts/qfai-pr.md

@@ -92,15 +92,6 @@ Simulate roles by running the same sequence yourself:
 
 - Write a short “role output” section per role, then consolidate into the final deliverable(s).
 
-## Behavior Rules (high leverage)
-
-- **Language**: Output MUST follow the user’s working language for this session.
-- **Question budget**: Ask at most **5** clarifying questions total. Prioritize blockers. If `--auto`, proceed with explicit assumptions.
-- **No hallucination**: Do not invent file paths, commands, or policies. Confirm via repo inspection.
-- **Evidence**: Do not claim completion without commands/results (format/lint/type/test/pack as applicable).
-- **Subagent contract**: When delegating, require the subagent response structure:
-  1. Findings 2) Recommendations 3) Proposed edits 4) Open Questions/Risks 5) Confidence
-
 ## Step 0 — Load Context (always)
 
 1. Read relevant **project steering** (if present):

package/assets/init/.qfai/assistant/prompts/qfai-require.md

@@ -88,15 +88,6 @@ Simulate roles by running the same sequence yourself:
 
 - Write a short “role output” section per role, then consolidate into the final deliverable(s).
 
-## Behavior Rules (high leverage)
-
-- **Language**: Output MUST follow the user’s working language for this session.
-- **Question budget**: Ask at most **5** clarifying questions total. Prioritize blockers. If `--auto`, proceed with explicit assumptions.
-- **No hallucination**: Do not invent file paths, commands, or policies. Confirm via repo inspection.
-- **Evidence**: Do not claim completion without commands/results (format/lint/type/test/pack as applicable).
-- **Subagent contract**: When delegating, require the subagent response structure:
-  1. Findings 2) Recommendations 3) Proposed edits 4) Open Questions/Risks 5) Confidence
-
 ## Step 0 — Load Context (always)
 
 1. Read relevant **project steering** (if present):

package/assets/init/.qfai/assistant/prompts/qfai-scenario-test.md

@@ -88,15 +88,6 @@ Simulate roles by running the same sequence yourself:
 
 - Write a short “role output” section per role, then consolidate into the final deliverable(s).
 
-## Behavior Rules (high leverage)
-
-- **Language**: Output MUST follow the user’s working language for this session.
-- **Question budget**: Ask at most **5** clarifying questions total. Prioritize blockers. If `--auto`, proceed with explicit assumptions.
-- **No hallucination**: Do not invent file paths, commands, or policies. Confirm via repo inspection.
-- **Evidence**: Do not claim completion without commands/results (format/lint/type/test/pack as applicable).
-- **Subagent contract**: When delegating, require the subagent response structure:
-  1. Findings 2) Recommendations 3) Proposed edits 4) Open Questions/Risks 5) Confidence
-
 ## Step 0 — Load Context (always)
 
 1. Read relevant **project steering** (if present):

package/assets/init/.qfai/assistant/prompts/qfai-spec.md

@@ -91,15 +91,6 @@ Simulate roles by running the same sequence yourself:
 
 - Write a short “role output” section per role, then consolidate into the final deliverable(s).
 
-## Behavior Rules (high leverage)
-
-- **Language**: Output MUST follow the user’s working language for this session.
-- **Question budget**: Ask at most **5** clarifying questions total. Prioritize blockers. If `--auto`, proceed with explicit assumptions.
-- **No hallucination**: Do not invent file paths, commands, or policies. Confirm via repo inspection.
-- **Evidence**: Do not claim completion without commands/results (format/lint/type/test/pack as applicable).
-- **Subagent contract**: When delegating, require the subagent response structure:
-  1. Findings 2) Recommendations 3) Proposed edits 4) Open Questions/Risks 5) Confidence
-
 ## Step 0 — Load Context (always)
 
 1. Read relevant **project steering** (if present):
@@ -262,7 +253,20 @@ Only create contracts when the spec requires a stable interface definition.
 
 If your repo defines contract schema or naming rules, follow them. Otherwise:
 
-- Use a clear filename and include “Purpose / Fields / Constraints / Examples” inside the YAML as comments.
+- **UI / API contracts:** write **YAML** and include “Purpose / Fields / Constraints / Examples” as YAML comments.
+- **DB contracts:** write **SQL (`.sql`)** and include a small header comment block, then representative DDL/schema notes.
+
+DB contract default conventions (unless your repo defines others):
+
+- Filename: `db-0001-<slug>.sql`
+- Header (must):
+  ```sql
+  -- QFAI-CONTRACT-ID: DB-0001
+  -- Purpose: <what this schema contract guarantees>
+  -- Constraints: <keys, nullability, ranges, invariants>
+  -- Examples: <optional example rows or queries>
+  ```
+- Body: DDL or schema notes that tests/scenarios can validate (minimal, spec-driven).
 
 ## Step 4 — QA + Review
 

package/assets/init/.qfai/assistant/prompts/qfai-unit-test.md

@@ -88,15 +88,6 @@ Simulate roles by running the same sequence yourself:
 
 - Write a short “role output” section per role, then consolidate into the final deliverable(s).
 
-## Behavior Rules (high leverage)
-
-- **Language**: Output MUST follow the user’s working language for this session.
-- **Question budget**: Ask at most **5** clarifying questions total. Prioritize blockers. If `--auto`, proceed with explicit assumptions.
-- **No hallucination**: Do not invent file paths, commands, or policies. Confirm via repo inspection.
-- **Evidence**: Do not claim completion without commands/results (format/lint/type/test/pack as applicable).
-- **Subagent contract**: When delegating, require the subagent response structure:
-  1. Findings 2) Recommendations 3) Proposed edits 4) Open Questions/Risks 5) Confidence
-
 ## Step 0 — Load Context (always)
 
 1. Read relevant **project steering** (if present):

package/assets/init/.qfai/assistant/prompts/qfai-verify.md

@@ -88,15 +88,6 @@ Simulate roles by running the same sequence yourself:
 
 - Write a short “role output” section per role, then consolidate into the final deliverable(s).
 
-## Behavior Rules (high leverage)
-
-- **Language**: Output MUST follow the user’s working language for this session.
-- **Question budget**: Ask at most **5** clarifying questions total. Prioritize blockers. If `--auto`, proceed with explicit assumptions.
-- **No hallucination**: Do not invent file paths, commands, or policies. Confirm via repo inspection.
-- **Evidence**: Do not claim completion without commands/results (format/lint/type/test/pack as applicable).
-- **Subagent contract**: When delegating, require the subagent response structure:
-  1. Findings 2) Recommendations 3) Proposed edits 4) Open Questions/Risks 5) Confidence
-
 ## Step 0 — Load Context (always)
 
 1. Read relevant **project steering** (if present):

package/assets/init/.qfai/assistant/prompts.local/README.md

@@ -4,3 +4,5 @@ Optional per-project overrides.
 
 If a file with the same name exists here (e.g., `qfai-spec.md`), a tool wrapper may prefer it to customize behavior for this repo.
 Keep overrides minimal; prefer updating steering/instructions instead.
+
+Note: After `qfai init`, this folder contains only this README. Place an override file here only when you intentionally diverge from the SSOT prompt bodies.

package/assets/init/.qfai/assistant/steering/README.md

@@ -31,3 +31,10 @@ Steering MUST be grounded in repo evidence. When possible, include:
 - directory anchors (e.g., `packages/qfai/src/cli`)
 
 If a fact cannot be verified, mark it as `TBD` and record what evidence is missing.
+
+### Default AI-managed flow (recommended)
+
+1. Run a QFAI command (e.g., `/qfai-require`, `/qfai-spec`, `/qfai-implement`).
+2. The agent loads `steering/*` and fills missing placeholders from **repo evidence**.
+3. If evidence is missing, write `TBD` and add an Open Question (what evidence is required).
+4. Optionally, review changes via PR if your team requires human approval.

package/assets/init/.qfai/contracts/README.md

@@ -6,6 +6,6 @@ Subfolders:
 
 - `ui/` : UI contracts (YAML)
 - `api/`: API contracts (YAML)
-- `db/` : DB contracts (YAML)
+- `db/` : DB contracts (SQL / .sql)
 
 Create contracts only when needed by the spec (keep minimal).

package/assets/init/.qfai/contracts/db/README.md

@@ -1,8 +1,25 @@
 # contracts/db/
 
-Place DB contracts here (YAML).
+Place **DB contracts** here as **SQL DDL files** (`.sql`).
 
-Guidelines:
+## File naming
 
-- Keep contracts minimal: define only what scenarios/tests need to validate.
-- Include examples where helpful.
+- `db-0001-<slug>.sql`
+  - Contract ID: `DB-0001` (4-digit fixed)
+
+## Required header
+
+Include the contract ID in the SQL header comments so QFAI can discover/index it:
+
+```sql
+-- QFAI-CONTRACT-ID: DB-0001
+-- DB-0001 (short description)
+```
+
+## Guidelines
+
+- Keep DB contracts minimal and spec-driven (only what scenarios/tests need).
+- This directory is for the **contract/schema snapshot**, not migrations history.
+- If your project uses an ORM schema (Prisma, etc.), decide one:
+  - (Recommended for v1.0.6) Keep a minimal `.sql` contract snapshot here, and link to the real schema in the spec.
+  - (Future) Add support for other schema formats via config (out of scope for v1.0.6).

package/assets/init/.qfai/specs/README.md

@@ -7,3 +7,5 @@ A **spec pack** lives under `specs/spec-XXXX/` and contains:
 - `scenario.feature` : Gherkin scenarios (ATDD skeleton)
 
 Create/update spec packs with `/qfai-spec`.
+
+Note: After `qfai init`, this folder contains only this README. Spec packs (`spec-XXXX/`) are created by running `/qfai-spec`.

package/dist/cli/index.cjs

@@ -1075,8 +1075,8 @@ var import_promises7 = require("fs/promises");
 var import_node_path9 = __toESM(require("path"), 1);
 var import_node_url2 = require("url");
 async function resolveToolVersion() {
-  if ("1.0.5".length > 0) {
-    return "1.0.5";
+  if ("1.0.6".length > 0) {
+    return "1.0.6";
   }
   try {
     const packagePath = resolvePackageJsonPath();

package/dist/cli/index.mjs

@@ -1056,8 +1056,8 @@ import { readFile as readFile4 } from "fs/promises";
 import path9 from "path";
 import { fileURLToPath as fileURLToPath2 } from "url";
 async function resolveToolVersion() {
-  if ("1.0.5".length > 0) {
-    return "1.0.5";
+  if ("1.0.6".length > 0) {
+    return "1.0.6";
   }
   try {
     const packagePath = resolvePackageJsonPath();

package/dist/index.cjs

@@ -1290,8 +1290,8 @@ var import_promises7 = require("fs/promises");
 var import_node_path8 = __toESM(require("path"), 1);
 var import_node_url = require("url");
 async function resolveToolVersion() {
-  if ("1.0.5".length > 0) {
-    return "1.0.5";
+  if ("1.0.6".length > 0) {
+    return "1.0.6";
   }
   try {
     const packagePath = resolvePackageJsonPath();

package/dist/index.mjs

@@ -1237,8 +1237,8 @@ import { readFile as readFile4 } from "fs/promises";
 import path8 from "path";
 import { fileURLToPath } from "url";
 async function resolveToolVersion() {
-  if ("1.0.5".length > 0) {
-    return "1.0.5";
+  if ("1.0.6".length > 0) {
+    return "1.0.6";
   }
   try {
     const packagePath = resolvePackageJsonPath();

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "qfai",
-  "version": "1.0.5",
+  "version": "1.0.6",
   "description": "Quality-first AI-driven development toolkit (SDD × ATDD × TDD).",
   "license": "MIT",
   "repository": {

package/assets/init/root/tests/qfai-traceability.sample.test.ts

@@ -1,2 +0,0 @@
-// Traceability sample: reference SC IDs via annotations (comments are enough).
-// QFAI:SC-0001