diff --git a/CHANGELOG.md b/CHANGELOG.md
index c75037925..46dc6af55 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -2,6 +2,14 @@
 
 All notable changes to this project will be documented in this file.
 
+## [2.2.2] - 2026-07-06
+
+Move `unit-test-instructions` production from `build-and-test` to `code-generation` so the per-unit subagent generating code is guided by the same test instructions file that `build-and-test` later executes. Eliminates the duplication where test guidance was produced after code generation had already completed.
+
+* `unit-test-instructions` is now produced per-unit by `code-generation` (new Step 4) and consumed by `build-and-test` — previously `build-and-test` both produced and consumed it, creating it too late to guide the code-generation subagent.
+* `build-and-test.produces[]` no longer includes `unit-test-instructions`; `build-and-test.consumes[]` now lists it as a required input from code-generation.
+* The code-generation subagent delegation prompt now includes `unit-test-instructions.md` alongside `code-generation-plan.md`.
+
 ## [2.2.1] - 2026-07-04
 
 Stops help requests from accidentally creating intents. `/aidlc intent help` was parsed as "switch to an intent named help"; the failed switch died with an error inviting the agent to "describe what to build to start a new one", which a conductor mid-task read as an instruction and birthed an unwanted intent. Bare `/aidlc help` and `/aidlc -h` were worse: they fell into the freeform-description funnel and offered to birth an intent literally named "help" (or silently advanced the active stage). All help-shaped inputs now route to the help text. **Upgrade:** re-copy your `dist/<harness>/` shell into the project.
diff --git a/README.md b/README.md
index 8330e9744..2473f733c 100644
--- a/README.md
+++ b/README.md
@@ -8,7 +8,7 @@ A native implementation of the **AI-DLC methodology** (AI-Driven Development Lif
 
 The methodology lives once, in a harness-neutral `core/`; each harness adds a thin surface that decides how it shows up on that harness. So you edit the methodology in one place, and every harness distribution is generated from it — no harness gets special treatment. (See [Repository layout](#repository-layout) for how the pieces fit together.)
 
-![version](https://img.shields.io/badge/version-2.2.1-blue)
+![version](https://img.shields.io/badge/version-2.2.2-blue)
 ![license](https://img.shields.io/badge/license-MIT--0-green)
 ![Kiro IDE](https://img.shields.io/badge/harness-Kiro%20IDE-orange)
 ![Kiro CLI](https://img.shields.io/badge/harness-Kiro%20CLI-orange)
diff --git a/core/aidlc-common/stages/construction/build-and-test.md b/core/aidlc-common/stages/construction/build-and-test.md
index 5b480de25..5c8e5b473 100644
--- a/core/aidlc-common/stages/construction/build-and-test.md
+++ b/core/aidlc-common/stages/construction/build-and-test.md
@@ -9,7 +9,6 @@ support_agents:
 mode: inline
 produces:
   - build-instructions
-  - unit-test-instructions
   - integration-test-instructions
   - performance-test-instructions
   - security-test-instructions
@@ -18,6 +17,8 @@ produces:
 consumes:
   - artifact: code-generation-plan
     required: true
+  - artifact: unit-test-instructions
+    required: true
   - artifact: code-summary
     required: true
 requires_stage:
@@ -36,7 +37,7 @@ scopes:
   - security-patch
   - workshop
 inputs: ALL code generation outputs across all units
-outputs: build-instructions.md, unit-test-instructions.md, integration-test-instructions.md, performance-test-instructions.md, security-test-instructions.md, build-and-test-summary.md, test-results.md (under this stage's record dir, engine-resolved)
+outputs: build-instructions.md, integration-test-instructions.md, performance-test-instructions.md, security-test-instructions.md, build-and-test-summary.md, test-results.md (under this stage's record dir, engine-resolved)
 ---
 
 # Build and Test
@@ -51,7 +52,7 @@ Load aidlc-quality-agent (lead) persona from `agents/aidlc-quality-agent.md` and
 
 ### Step 2: Analyze Testing Requirements
 
-Read code generation outputs across all units from `<record>/construction/*/code-generation/code-summary.md`. Review NFR requirements across units (if they exist) to identify performance and security testing needs. Catalog all test types required.
+Read code generation outputs across all units from `<record>/construction/*/code-generation/code-summary.md` and per-unit test instructions from `<record>/construction/*/code-generation/unit-test-instructions.md`. Review NFR requirements across units (if they exist) to identify performance and security testing needs. Catalog all test types required.
 
 ### Step 3: Generate Build Instructions
 
@@ -64,17 +65,14 @@ Create `<record>/construction/build-and-test/build-instructions.md`:
 
 ### Step 4-8: Generate Test Instructions (Strategy-Aware)
 
-Consult the active test strategy from `aidlc-state.md` → `**Test Strategy**` (see stage-protocol.md §8 "Test Strategy"). Generate test instruction files based on the strategy level:
+Consult the active test strategy from `aidlc-state.md` → `**Test Strategy**` (see stage-protocol.md §8 "Test Strategy"). Generate test instruction files based on the strategy level.
 
-**Minimal strategy** — generate ONLY:
-- `unit-test-instructions.md`: Requirement-driven unit tests (1 test per requirement, happy-path floor per component). ~5-15 tests total. Skip all other test types.
+**Minimal strategy** — no additional test instruction files are generated (unit tests are already covered per-unit by code-generation).
 
 **Standard strategy** — generate:
-- `unit-test-instructions.md`: 5-8 tests per component, key behavior coverage
 - `integration-test-instructions.md`: Key boundary tests, cross-unit interaction
 
 **Comprehensive strategy** — generate all applicable:
-- `unit-test-instructions.md`: 10-15 tests per component, thorough coverage
 - `integration-test-instructions.md`: Cross-unit interaction, external dependency handling
 - `performance-test-instructions.md` (IF NFR performance requirements exist): Load testing, benchmarks, regression detection
 - `security-test-instructions.md` (IF NFR security requirements exist): SAST/DAST, auth testing, injection testing
@@ -104,7 +102,7 @@ Create `<record>/construction/build-and-test/build-and-test-summary.md`:
 Attempt to execute the build and test commands documented in the instruction files:
 
 1. **Build**: Run the build commands from `build-instructions.md` via Bash. Capture output.
-2. **Unit tests**: Run the unit test command from `unit-test-instructions.md` via Bash. Capture pass/fail counts.
+2. **Unit tests**: Run the unit test commands from the per-unit `unit-test-instructions.md` files (under `<record>/construction/*/code-generation/`) via Bash. Capture pass/fail counts.
 3. **Integration tests** (if applicable): Run integration test commands. Capture results.
 4. **Report results**: Create or update `<record>/construction/build-and-test/test-results.md` with:
    - Build status (success/failure + output)
@@ -154,7 +152,8 @@ The imported sensors check those outputs:
 - **`required-sections`** verifies each instruction file contains the
   registry default (≥2 H2 headings).
 - **`upstream-coverage`** verifies the prose references the upstream
-  artefacts this stage consumes (`code-generation-plan`, `code-summary`).
+  artefacts this stage consumes (`code-generation-plan`, `unit-test-instructions`,
+  `code-summary`).
 - **`type-check`** runs against any TypeScript/TSX code touched as part
   of test generation (matches `**/*.{ts,tsx}`).
 
diff --git a/core/aidlc-common/stages/construction/code-generation.md b/core/aidlc-common/stages/construction/code-generation.md
index ec253d796..77b449324 100644
--- a/core/aidlc-common/stages/construction/code-generation.md
+++ b/core/aidlc-common/stages/construction/code-generation.md
@@ -12,6 +12,7 @@ for_each: unit-of-work
 workspace_requires: true
 produces:
   - code-generation-plan
+  - unit-test-instructions
   - code-summary
 consumes:
   - artifact: business-logic-model
@@ -49,7 +50,7 @@ scopes:
   - security-patch
   - workshop
 inputs: ALL prior design artifacts for this unit
-outputs: application code + code-generation-plan.md, code-summary.md (under this stage's per-unit record dir, engine-resolved)
+outputs: application code + code-generation-plan.md, unit-test-instructions.md, code-summary.md (under this stage's per-unit record dir, engine-resolved)
 ---
 
 # Code Generation
@@ -133,7 +134,24 @@ Present a structured question to get plan approval before proceeding to generati
 - "Approve Plan" — proceed to code generation
 - "Request Changes" — revise the plan
 
-### Step 4: PART 2 — Generation
+### Step 4: Generate Unit Test Instructions
+
+Create `<record>/construction/{unit-name}/code-generation/unit-test-instructions.md` based on the active test strategy (stage-protocol.md §8 "Test Strategy"):
+
+- **Minimal strategy**: Requirement-driven unit tests (1 test per requirement, happy-path floor per component). ~5-15 tests total.
+- **Standard strategy**: 5-8 tests per component, key behavior coverage
+- **Comprehensive strategy**: 10-15 tests per component, thorough coverage
+
+Include:
+- Test framework setup and configuration
+- How to run unit tests (commands, flags, filters)
+- Expected coverage targets
+- Mocking/stubbing guidance
+- Test data management
+
+This file guides the subagent during code generation and is later consumed by build-and-test for execution.
+
+### Step 5: PART 2 — Generation
 
 Before delegating, display to the user:
 "Generating code for [N] plan steps. This may take several minutes depending on project complexity. I'll show a summary when complete."
@@ -146,12 +164,13 @@ Include in the delegation prompt:
 - Design artifacts for the CURRENT UNIT ONLY (not all units)
 - A 1-2 line summary of each inception-phase artifact with its file path (requirements summary, stories summary, app design summary) — the subagent can Read specific files if it needs full content
 - The approved code-generation-plan.md (full content)
+- The unit-test-instructions.md (full content)
 - Project workspace details (languages, frameworks, conventions from aidlc-state.md)
 - Instructions to execute each plan step sequentially and mark checkboxes as completed
 
 The subagent generates all code, test files, and configuration artifacts in the workspace.
 
-### Step 5: Generate Code Summary
+### Step 6: Generate Code Summary
 
 After subagent completes, create `<record>/construction/{unit-name}/code-generation/code-summary.md` documenting:
 - Files created/modified
@@ -159,11 +178,11 @@ After subagent completes, create `<record>/construction/{unit-name}/code-generat
 - Test coverage summary
 - Any deviations from the plan
 
-### Step 6: Update State
+### Step 7: Update State
 
 Update `<record>/aidlc-state.md`: mark Code Generation for {unit-name} as `[x]` completed and update "Current Status".
 
-### Step 7: Completion
+### Step 8: Completion
 
 Present completion message and approval gate:
 
@@ -186,7 +205,8 @@ Approval gate: strictly 2-option (Approve / Request Changes).
 This stage produces TypeScript/JavaScript code in the active Bolt
 worktree. Generated code lives at the workspace root (NEVER under
 the record dir); the planning + summary artefacts (`code-generation-plan.md`,
-`code-summary.md`) live under `<record>/construction/{unit-name}/code-generation/`.
+`unit-test-instructions.md`, `code-summary.md`) live under
+`<record>/construction/{unit-name}/code-generation/`.
 
 The imported sensors check the code outputs:
 
diff --git a/core/tools/aidlc-version.ts b/core/tools/aidlc-version.ts
index 834a24531..9ea354589 100644
--- a/core/tools/aidlc-version.ts
+++ b/core/tools/aidlc-version.ts
@@ -1,4 +1,4 @@
 // Hand-edited single source of truth for the AIDLC framework version.
 // Bumped in the same commit that adds the matching ## [N.N.N] heading
 // to CHANGELOG.md. Pinned by tests/unit/t68-version-changelog-sync.test.ts.
-export const AIDLC_VERSION = "2.2.1";
+export const AIDLC_VERSION = "2.2.2";
diff --git a/dist/claude/.claude/aidlc-common/stages/construction/build-and-test.md b/dist/claude/.claude/aidlc-common/stages/construction/build-and-test.md
index 033e3ad77..6947f62f2 100644
--- a/dist/claude/.claude/aidlc-common/stages/construction/build-and-test.md
+++ b/dist/claude/.claude/aidlc-common/stages/construction/build-and-test.md
@@ -9,7 +9,6 @@ support_agents:
 mode: inline
 produces:
   - build-instructions
-  - unit-test-instructions
   - integration-test-instructions
   - performance-test-instructions
   - security-test-instructions
@@ -18,6 +17,8 @@ produces:
 consumes:
   - artifact: code-generation-plan
     required: true
+  - artifact: unit-test-instructions
+    required: true
   - artifact: code-summary
     required: true
 requires_stage:
@@ -36,7 +37,7 @@ scopes:
   - security-patch
   - workshop
 inputs: ALL code generation outputs across all units
-outputs: build-instructions.md, unit-test-instructions.md, integration-test-instructions.md, performance-test-instructions.md, security-test-instructions.md, build-and-test-summary.md, test-results.md (under this stage's record dir, engine-resolved)
+outputs: build-instructions.md, integration-test-instructions.md, performance-test-instructions.md, security-test-instructions.md, build-and-test-summary.md, test-results.md (under this stage's record dir, engine-resolved)
 ---
 
 # Build and Test
@@ -51,7 +52,7 @@ Load aidlc-quality-agent (lead) persona from `agents/aidlc-quality-agent.md` and
 
 ### Step 2: Analyze Testing Requirements
 
-Read code generation outputs across all units from `<record>/construction/*/code-generation/code-summary.md`. Review NFR requirements across units (if they exist) to identify performance and security testing needs. Catalog all test types required.
+Read code generation outputs across all units from `<record>/construction/*/code-generation/code-summary.md` and per-unit test instructions from `<record>/construction/*/code-generation/unit-test-instructions.md`. Review NFR requirements across units (if they exist) to identify performance and security testing needs. Catalog all test types required.
 
 ### Step 3: Generate Build Instructions
 
@@ -64,17 +65,14 @@ Create `<record>/construction/build-and-test/build-instructions.md`:
 
 ### Step 4-8: Generate Test Instructions (Strategy-Aware)
 
-Consult the active test strategy from `aidlc-state.md` → `**Test Strategy**` (see stage-protocol.md §8 "Test Strategy"). Generate test instruction files based on the strategy level:
+Consult the active test strategy from `aidlc-state.md` → `**Test Strategy**` (see stage-protocol.md §8 "Test Strategy"). Generate test instruction files based on the strategy level.
 
-**Minimal strategy** — generate ONLY:
-- `unit-test-instructions.md`: Requirement-driven unit tests (1 test per requirement, happy-path floor per component). ~5-15 tests total. Skip all other test types.
+**Minimal strategy** — no additional test instruction files are generated (unit tests are already covered per-unit by code-generation).
 
 **Standard strategy** — generate:
-- `unit-test-instructions.md`: 5-8 tests per component, key behavior coverage
 - `integration-test-instructions.md`: Key boundary tests, cross-unit interaction
 
 **Comprehensive strategy** — generate all applicable:
-- `unit-test-instructions.md`: 10-15 tests per component, thorough coverage
 - `integration-test-instructions.md`: Cross-unit interaction, external dependency handling
 - `performance-test-instructions.md` (IF NFR performance requirements exist): Load testing, benchmarks, regression detection
 - `security-test-instructions.md` (IF NFR security requirements exist): SAST/DAST, auth testing, injection testing
@@ -104,7 +102,7 @@ Create `<record>/construction/build-and-test/build-and-test-summary.md`:
 Attempt to execute the build and test commands documented in the instruction files:
 
 1. **Build**: Run the build commands from `build-instructions.md` via Bash. Capture output.
-2. **Unit tests**: Run the unit test command from `unit-test-instructions.md` via Bash. Capture pass/fail counts.
+2. **Unit tests**: Run the unit test commands from the per-unit `unit-test-instructions.md` files (under `<record>/construction/*/code-generation/`) via Bash. Capture pass/fail counts.
 3. **Integration tests** (if applicable): Run integration test commands. Capture results.
 4. **Report results**: Create or update `<record>/construction/build-and-test/test-results.md` with:
    - Build status (success/failure + output)
@@ -154,7 +152,8 @@ The imported sensors check those outputs:
 - **`required-sections`** verifies each instruction file contains the
   registry default (≥2 H2 headings).
 - **`upstream-coverage`** verifies the prose references the upstream
-  artefacts this stage consumes (`code-generation-plan`, `code-summary`).
+  artefacts this stage consumes (`code-generation-plan`, `unit-test-instructions`,
+  `code-summary`).
 - **`type-check`** runs against any TypeScript/TSX code touched as part
   of test generation (matches `**/*.{ts,tsx}`).
 
diff --git a/dist/claude/.claude/aidlc-common/stages/construction/code-generation.md b/dist/claude/.claude/aidlc-common/stages/construction/code-generation.md
index 23256859f..aec9f4646 100644
--- a/dist/claude/.claude/aidlc-common/stages/construction/code-generation.md
+++ b/dist/claude/.claude/aidlc-common/stages/construction/code-generation.md
@@ -12,6 +12,7 @@ for_each: unit-of-work
 workspace_requires: true
 produces:
   - code-generation-plan
+  - unit-test-instructions
   - code-summary
 consumes:
   - artifact: business-logic-model
@@ -49,7 +50,7 @@ scopes:
   - security-patch
   - workshop
 inputs: ALL prior design artifacts for this unit
-outputs: application code + code-generation-plan.md, code-summary.md (under this stage's per-unit record dir, engine-resolved)
+outputs: application code + code-generation-plan.md, unit-test-instructions.md, code-summary.md (under this stage's per-unit record dir, engine-resolved)
 ---
 
 # Code Generation
@@ -133,7 +134,24 @@ Present a structured question to get plan approval before proceeding to generati
 - "Approve Plan" — proceed to code generation
 - "Request Changes" — revise the plan
 
-### Step 4: PART 2 — Generation
+### Step 4: Generate Unit Test Instructions
+
+Create `<record>/construction/{unit-name}/code-generation/unit-test-instructions.md` based on the active test strategy (stage-protocol.md §8 "Test Strategy"):
+
+- **Minimal strategy**: Requirement-driven unit tests (1 test per requirement, happy-path floor per component). ~5-15 tests total.
+- **Standard strategy**: 5-8 tests per component, key behavior coverage
+- **Comprehensive strategy**: 10-15 tests per component, thorough coverage
+
+Include:
+- Test framework setup and configuration
+- How to run unit tests (commands, flags, filters)
+- Expected coverage targets
+- Mocking/stubbing guidance
+- Test data management
+
+This file guides the subagent during code generation and is later consumed by build-and-test for execution.
+
+### Step 5: PART 2 — Generation
 
 Before delegating, display to the user:
 "Generating code for [N] plan steps. This may take several minutes depending on project complexity. I'll show a summary when complete."
@@ -146,12 +164,13 @@ Include in the delegation prompt:
 - Design artifacts for the CURRENT UNIT ONLY (not all units)
 - A 1-2 line summary of each inception-phase artifact with its file path (requirements summary, stories summary, app design summary) — the subagent can Read specific files if it needs full content
 - The approved code-generation-plan.md (full content)
+- The unit-test-instructions.md (full content)
 - Project workspace details (languages, frameworks, conventions from aidlc-state.md)
 - Instructions to execute each plan step sequentially and mark checkboxes as completed
 
 The subagent generates all code, test files, and configuration artifacts in the workspace.
 
-### Step 5: Generate Code Summary
+### Step 6: Generate Code Summary
 
 After subagent completes, create `<record>/construction/{unit-name}/code-generation/code-summary.md` documenting:
 - Files created/modified
@@ -159,11 +178,11 @@ After subagent completes, create `<record>/construction/{unit-name}/code-generat
 - Test coverage summary
 - Any deviations from the plan
 
-### Step 6: Update State
+### Step 7: Update State
 
 Update `<record>/aidlc-state.md`: mark Code Generation for {unit-name} as `[x]` completed and update "Current Status".
 
-### Step 7: Completion
+### Step 8: Completion
 
 Present completion message and approval gate:
 
@@ -186,7 +205,8 @@ Approval gate: strictly 2-option (Approve / Request Changes).
 This stage produces TypeScript/JavaScript code in the active Bolt
 worktree. Generated code lives at the workspace root (NEVER under
 the record dir); the planning + summary artefacts (`code-generation-plan.md`,
-`code-summary.md`) live under `<record>/construction/{unit-name}/code-generation/`.
+`unit-test-instructions.md`, `code-summary.md`) live under
+`<record>/construction/{unit-name}/code-generation/`.
 
 The imported sensors check the code outputs:
 
diff --git a/dist/claude/.claude/tools/aidlc-version.ts b/dist/claude/.claude/tools/aidlc-version.ts
index 834a24531..9ea354589 100644
--- a/dist/claude/.claude/tools/aidlc-version.ts
+++ b/dist/claude/.claude/tools/aidlc-version.ts
@@ -1,4 +1,4 @@
 // Hand-edited single source of truth for the AIDLC framework version.
 // Bumped in the same commit that adds the matching ## [N.N.N] heading
 // to CHANGELOG.md. Pinned by tests/unit/t68-version-changelog-sync.test.ts.
-export const AIDLC_VERSION = "2.2.1";
+export const AIDLC_VERSION = "2.2.2";
diff --git a/dist/claude/.claude/tools/data/stage-graph.json b/dist/claude/.claude/tools/data/stage-graph.json
index 915e6b0cd..3c98392e0 100644
--- a/dist/claude/.claude/tools/data/stage-graph.json
+++ b/dist/claude/.claude/tools/data/stage-graph.json
@@ -1853,6 +1853,7 @@
     "workspace_requires": true,
     "produces": [
       "code-generation-plan",
+      "unit-test-instructions",
       "code-summary"
     ],
     "consumes": [
@@ -1913,7 +1914,7 @@
     "reviewer": "aidlc-architecture-reviewer-agent",
     "reviewer_max_iterations": 2,
     "inputs": "ALL prior design artifacts for this unit",
-    "outputs": "application code + code-generation-plan.md, code-summary.md (under this stage's per-unit record dir, engine-resolved)",
+    "outputs": "application code + code-generation-plan.md, unit-test-instructions.md, code-summary.md (under this stage's per-unit record dir, engine-resolved)",
     "rules_in_context": [
       {
         "path": "aidlc/spaces/default/memory/org.md",
@@ -1959,7 +1960,6 @@
     "mode": "inline",
     "produces": [
       "build-instructions",
-      "unit-test-instructions",
       "integration-test-instructions",
       "performance-test-instructions",
       "security-test-instructions",
@@ -1971,6 +1971,10 @@
         "artifact": "code-generation-plan",
         "required": true
       },
+      {
+        "artifact": "unit-test-instructions",
+        "required": true
+      },
       {
         "artifact": "code-summary",
         "required": true
@@ -1995,7 +1999,7 @@
       "workshop"
     ],
     "inputs": "ALL code generation outputs across all units",
-    "outputs": "build-instructions.md, unit-test-instructions.md, integration-test-instructions.md, performance-test-instructions.md, security-test-instructions.md, build-and-test-summary.md, test-results.md (under this stage's record dir, engine-resolved)",
+    "outputs": "build-instructions.md, integration-test-instructions.md, performance-test-instructions.md, security-test-instructions.md, build-and-test-summary.md, test-results.md (under this stage's record dir, engine-resolved)",
     "rules_in_context": [
       {
         "path": "aidlc/spaces/default/memory/org.md",
diff --git a/dist/codex/.codex/aidlc-common/stages/construction/build-and-test.md b/dist/codex/.codex/aidlc-common/stages/construction/build-and-test.md
index 3f8d1c66a..fccb8724d 100644
--- a/dist/codex/.codex/aidlc-common/stages/construction/build-and-test.md
+++ b/dist/codex/.codex/aidlc-common/stages/construction/build-and-test.md
@@ -9,7 +9,6 @@ support_agents:
 mode: inline
 produces:
   - build-instructions
-  - unit-test-instructions
   - integration-test-instructions
   - performance-test-instructions
   - security-test-instructions
@@ -18,6 +17,8 @@ produces:
 consumes:
   - artifact: code-generation-plan
     required: true
+  - artifact: unit-test-instructions
+    required: true
   - artifact: code-summary
     required: true
 requires_stage:
@@ -36,7 +37,7 @@ scopes:
   - security-patch
   - workshop
 inputs: ALL code generation outputs across all units
-outputs: build-instructions.md, unit-test-instructions.md, integration-test-instructions.md, performance-test-instructions.md, security-test-instructions.md, build-and-test-summary.md, test-results.md (under this stage's record dir, engine-resolved)
+outputs: build-instructions.md, integration-test-instructions.md, performance-test-instructions.md, security-test-instructions.md, build-and-test-summary.md, test-results.md (under this stage's record dir, engine-resolved)
 ---
 
 # Build and Test
@@ -51,7 +52,7 @@ Load aidlc-quality-agent (lead) persona from `agents/aidlc-quality-agent.md` and
 
 ### Step 2: Analyze Testing Requirements
 
-Read code generation outputs across all units from `<record>/construction/*/code-generation/code-summary.md`. Review NFR requirements across units (if they exist) to identify performance and security testing needs. Catalog all test types required.
+Read code generation outputs across all units from `<record>/construction/*/code-generation/code-summary.md` and per-unit test instructions from `<record>/construction/*/code-generation/unit-test-instructions.md`. Review NFR requirements across units (if they exist) to identify performance and security testing needs. Catalog all test types required.
 
 ### Step 3: Generate Build Instructions
 
@@ -64,17 +65,14 @@ Create `<record>/construction/build-and-test/build-instructions.md`:
 
 ### Step 4-8: Generate Test Instructions (Strategy-Aware)
 
-Consult the active test strategy from `aidlc-state.md` → `**Test Strategy**` (see stage-protocol.md §8 "Test Strategy"). Generate test instruction files based on the strategy level:
+Consult the active test strategy from `aidlc-state.md` → `**Test Strategy**` (see stage-protocol.md §8 "Test Strategy"). Generate test instruction files based on the strategy level.
 
-**Minimal strategy** — generate ONLY:
-- `unit-test-instructions.md`: Requirement-driven unit tests (1 test per requirement, happy-path floor per component). ~5-15 tests total. Skip all other test types.
+**Minimal strategy** — no additional test instruction files are generated (unit tests are already covered per-unit by code-generation).
 
 **Standard strategy** — generate:
-- `unit-test-instructions.md`: 5-8 tests per component, key behavior coverage
 - `integration-test-instructions.md`: Key boundary tests, cross-unit interaction
 
 **Comprehensive strategy** — generate all applicable:
-- `unit-test-instructions.md`: 10-15 tests per component, thorough coverage
 - `integration-test-instructions.md`: Cross-unit interaction, external dependency handling
 - `performance-test-instructions.md` (IF NFR performance requirements exist): Load testing, benchmarks, regression detection
 - `security-test-instructions.md` (IF NFR security requirements exist): SAST/DAST, auth testing, injection testing
@@ -104,7 +102,7 @@ Create `<record>/construction/build-and-test/build-and-test-summary.md`:
 Attempt to execute the build and test commands documented in the instruction files:
 
 1. **Build**: Run the build commands from `build-instructions.md` via Bash. Capture output.
-2. **Unit tests**: Run the unit test command from `unit-test-instructions.md` via Bash. Capture pass/fail counts.
+2. **Unit tests**: Run the unit test commands from the per-unit `unit-test-instructions.md` files (under `<record>/construction/*/code-generation/`) via Bash. Capture pass/fail counts.
 3. **Integration tests** (if applicable): Run integration test commands. Capture results.
 4. **Report results**: Create or update `<record>/construction/build-and-test/test-results.md` with:
    - Build status (success/failure + output)
@@ -154,7 +152,8 @@ The imported sensors check those outputs:
 - **`required-sections`** verifies each instruction file contains the
   registry default (≥2 H2 headings).
 - **`upstream-coverage`** verifies the prose references the upstream
-  artefacts this stage consumes (`code-generation-plan`, `code-summary`).
+  artefacts this stage consumes (`code-generation-plan`, `unit-test-instructions`,
+  `code-summary`).
 - **`type-check`** runs against any TypeScript/TSX code touched as part
   of test generation (matches `**/*.{ts,tsx}`).
 
diff --git a/dist/codex/.codex/aidlc-common/stages/construction/code-generation.md b/dist/codex/.codex/aidlc-common/stages/construction/code-generation.md
index 80338c225..a3760a8f3 100644
--- a/dist/codex/.codex/aidlc-common/stages/construction/code-generation.md
+++ b/dist/codex/.codex/aidlc-common/stages/construction/code-generation.md
@@ -12,6 +12,7 @@ for_each: unit-of-work
 workspace_requires: true
 produces:
   - code-generation-plan
+  - unit-test-instructions
   - code-summary
 consumes:
   - artifact: business-logic-model
@@ -49,7 +50,7 @@ scopes:
   - security-patch
   - workshop
 inputs: ALL prior design artifacts for this unit
-outputs: application code + code-generation-plan.md, code-summary.md (under this stage's per-unit record dir, engine-resolved)
+outputs: application code + code-generation-plan.md, unit-test-instructions.md, code-summary.md (under this stage's per-unit record dir, engine-resolved)
 ---
 
 # Code Generation
@@ -133,7 +134,24 @@ Present a structured question to get plan approval before proceeding to generati
 - "Approve Plan" — proceed to code generation
 - "Request Changes" — revise the plan
 
-### Step 4: PART 2 — Generation
+### Step 4: Generate Unit Test Instructions
+
+Create `<record>/construction/{unit-name}/code-generation/unit-test-instructions.md` based on the active test strategy (stage-protocol.md §8 "Test Strategy"):
+
+- **Minimal strategy**: Requirement-driven unit tests (1 test per requirement, happy-path floor per component). ~5-15 tests total.
+- **Standard strategy**: 5-8 tests per component, key behavior coverage
+- **Comprehensive strategy**: 10-15 tests per component, thorough coverage
+
+Include:
+- Test framework setup and configuration
+- How to run unit tests (commands, flags, filters)
+- Expected coverage targets
+- Mocking/stubbing guidance
+- Test data management
+
+This file guides the subagent during code generation and is later consumed by build-and-test for execution.
+
+### Step 5: PART 2 — Generation
 
 Before delegating, display to the user:
 "Generating code for [N] plan steps. This may take several minutes depending on project complexity. I'll show a summary when complete."
@@ -146,12 +164,13 @@ Include in the delegation prompt:
 - Design artifacts for the CURRENT UNIT ONLY (not all units)
 - A 1-2 line summary of each inception-phase artifact with its file path (requirements summary, stories summary, app design summary) — the subagent can Read specific files if it needs full content
 - The approved code-generation-plan.md (full content)
+- The unit-test-instructions.md (full content)
 - Project workspace details (languages, frameworks, conventions from aidlc-state.md)
 - Instructions to execute each plan step sequentially and mark checkboxes as completed
 
 The subagent generates all code, test files, and configuration artifacts in the workspace.
 
-### Step 5: Generate Code Summary
+### Step 6: Generate Code Summary
 
 After subagent completes, create `<record>/construction/{unit-name}/code-generation/code-summary.md` documenting:
 - Files created/modified
@@ -159,11 +178,11 @@ After subagent completes, create `<record>/construction/{unit-name}/code-generat
 - Test coverage summary
 - Any deviations from the plan
 
-### Step 6: Update State
+### Step 7: Update State
 
 Update `<record>/aidlc-state.md`: mark Code Generation for {unit-name} as `[x]` completed and update "Current Status".
 
-### Step 7: Completion
+### Step 8: Completion
 
 Present completion message and approval gate:
 
@@ -186,7 +205,8 @@ Approval gate: strictly 2-option (Approve / Request Changes).
 This stage produces TypeScript/JavaScript code in the active Bolt
 worktree. Generated code lives at the workspace root (NEVER under
 the record dir); the planning + summary artefacts (`code-generation-plan.md`,
-`code-summary.md`) live under `<record>/construction/{unit-name}/code-generation/`.
+`unit-test-instructions.md`, `code-summary.md`) live under
+`<record>/construction/{unit-name}/code-generation/`.
 
 The imported sensors check the code outputs:
 
diff --git a/dist/codex/.codex/tools/aidlc-version.ts b/dist/codex/.codex/tools/aidlc-version.ts
index 834a24531..9ea354589 100644
--- a/dist/codex/.codex/tools/aidlc-version.ts
+++ b/dist/codex/.codex/tools/aidlc-version.ts
@@ -1,4 +1,4 @@
 // Hand-edited single source of truth for the AIDLC framework version.
 // Bumped in the same commit that adds the matching ## [N.N.N] heading
 // to CHANGELOG.md. Pinned by tests/unit/t68-version-changelog-sync.test.ts.
-export const AIDLC_VERSION = "2.2.1";
+export const AIDLC_VERSION = "2.2.2";
diff --git a/dist/codex/.codex/tools/data/stage-graph.json b/dist/codex/.codex/tools/data/stage-graph.json
index 1040cf96a..45bc36ea6 100644
--- a/dist/codex/.codex/tools/data/stage-graph.json
+++ b/dist/codex/.codex/tools/data/stage-graph.json
@@ -1853,6 +1853,7 @@
     "workspace_requires": true,
     "produces": [
       "code-generation-plan",
+      "unit-test-instructions",
       "code-summary"
     ],
     "consumes": [
@@ -1913,7 +1914,7 @@
     "reviewer": "aidlc-architecture-reviewer-agent",
     "reviewer_max_iterations": 2,
     "inputs": "ALL prior design artifacts for this unit",
-    "outputs": "application code + code-generation-plan.md, code-summary.md (under this stage's per-unit record dir, engine-resolved)",
+    "outputs": "application code + code-generation-plan.md, unit-test-instructions.md, code-summary.md (under this stage's per-unit record dir, engine-resolved)",
     "rules_in_context": [
       {
         "path": "aidlc/spaces/default/memory/org.md",
@@ -1959,7 +1960,6 @@
     "mode": "inline",
     "produces": [
       "build-instructions",
-      "unit-test-instructions",
       "integration-test-instructions",
       "performance-test-instructions",
       "security-test-instructions",
@@ -1971,6 +1971,10 @@
         "artifact": "code-generation-plan",
         "required": true
       },
+      {
+        "artifact": "unit-test-instructions",
+        "required": true
+      },
       {
         "artifact": "code-summary",
         "required": true
@@ -1995,7 +1999,7 @@
       "workshop"
     ],
     "inputs": "ALL code generation outputs across all units",
-    "outputs": "build-instructions.md, unit-test-instructions.md, integration-test-instructions.md, performance-test-instructions.md, security-test-instructions.md, build-and-test-summary.md, test-results.md (under this stage's record dir, engine-resolved)",
+    "outputs": "build-instructions.md, integration-test-instructions.md, performance-test-instructions.md, security-test-instructions.md, build-and-test-summary.md, test-results.md (under this stage's record dir, engine-resolved)",
     "rules_in_context": [
       {
         "path": "aidlc/spaces/default/memory/org.md",
diff --git a/dist/kiro-ide/.kiro/aidlc-common/stages/construction/build-and-test.md b/dist/kiro-ide/.kiro/aidlc-common/stages/construction/build-and-test.md
index 48592056d..1e2079307 100644
--- a/dist/kiro-ide/.kiro/aidlc-common/stages/construction/build-and-test.md
+++ b/dist/kiro-ide/.kiro/aidlc-common/stages/construction/build-and-test.md
@@ -9,7 +9,6 @@ support_agents:
 mode: inline
 produces:
   - build-instructions
-  - unit-test-instructions
   - integration-test-instructions
   - performance-test-instructions
   - security-test-instructions
@@ -18,6 +17,8 @@ produces:
 consumes:
   - artifact: code-generation-plan
     required: true
+  - artifact: unit-test-instructions
+    required: true
   - artifact: code-summary
     required: true
 requires_stage:
@@ -36,7 +37,7 @@ scopes:
   - security-patch
   - workshop
 inputs: ALL code generation outputs across all units
-outputs: build-instructions.md, unit-test-instructions.md, integration-test-instructions.md, performance-test-instructions.md, security-test-instructions.md, build-and-test-summary.md, test-results.md (under this stage's record dir, engine-resolved)
+outputs: build-instructions.md, integration-test-instructions.md, performance-test-instructions.md, security-test-instructions.md, build-and-test-summary.md, test-results.md (under this stage's record dir, engine-resolved)
 ---
 
 # Build and Test
@@ -51,7 +52,7 @@ Load aidlc-quality-agent (lead) persona from `agents/aidlc-quality-agent.md` and
 
 ### Step 2: Analyze Testing Requirements
 
-Read code generation outputs across all units from `<record>/construction/*/code-generation/code-summary.md`. Review NFR requirements across units (if they exist) to identify performance and security testing needs. Catalog all test types required.
+Read code generation outputs across all units from `<record>/construction/*/code-generation/code-summary.md` and per-unit test instructions from `<record>/construction/*/code-generation/unit-test-instructions.md`. Review NFR requirements across units (if they exist) to identify performance and security testing needs. Catalog all test types required.
 
 ### Step 3: Generate Build Instructions
 
@@ -64,17 +65,14 @@ Create `<record>/construction/build-and-test/build-instructions.md`:
 
 ### Step 4-8: Generate Test Instructions (Strategy-Aware)
 
-Consult the active test strategy from `aidlc-state.md` → `**Test Strategy**` (see stage-protocol.md §8 "Test Strategy"). Generate test instruction files based on the strategy level:
+Consult the active test strategy from `aidlc-state.md` → `**Test Strategy**` (see stage-protocol.md §8 "Test Strategy"). Generate test instruction files based on the strategy level.
 
-**Minimal strategy** — generate ONLY:
-- `unit-test-instructions.md`: Requirement-driven unit tests (1 test per requirement, happy-path floor per component). ~5-15 tests total. Skip all other test types.
+**Minimal strategy** — no additional test instruction files are generated (unit tests are already covered per-unit by code-generation).
 
 **Standard strategy** — generate:
-- `unit-test-instructions.md`: 5-8 tests per component, key behavior coverage
 - `integration-test-instructions.md`: Key boundary tests, cross-unit interaction
 
 **Comprehensive strategy** — generate all applicable:
-- `unit-test-instructions.md`: 10-15 tests per component, thorough coverage
 - `integration-test-instructions.md`: Cross-unit interaction, external dependency handling
 - `performance-test-instructions.md` (IF NFR performance requirements exist): Load testing, benchmarks, regression detection
 - `security-test-instructions.md` (IF NFR security requirements exist): SAST/DAST, auth testing, injection testing
@@ -104,7 +102,7 @@ Create `<record>/construction/build-and-test/build-and-test-summary.md`:
 Attempt to execute the build and test commands documented in the instruction files:
 
 1. **Build**: Run the build commands from `build-instructions.md` via Bash. Capture output.
-2. **Unit tests**: Run the unit test command from `unit-test-instructions.md` via Bash. Capture pass/fail counts.
+2. **Unit tests**: Run the unit test commands from the per-unit `unit-test-instructions.md` files (under `<record>/construction/*/code-generation/`) via Bash. Capture pass/fail counts.
 3. **Integration tests** (if applicable): Run integration test commands. Capture results.
 4. **Report results**: Create or update `<record>/construction/build-and-test/test-results.md` with:
    - Build status (success/failure + output)
@@ -154,7 +152,8 @@ The imported sensors check those outputs:
 - **`required-sections`** verifies each instruction file contains the
   registry default (≥2 H2 headings).
 - **`upstream-coverage`** verifies the prose references the upstream
-  artefacts this stage consumes (`code-generation-plan`, `code-summary`).
+  artefacts this stage consumes (`code-generation-plan`, `unit-test-instructions`,
+  `code-summary`).
 - **`type-check`** runs against any TypeScript/TSX code touched as part
   of test generation (matches `**/*.{ts,tsx}`).
 
diff --git a/dist/kiro-ide/.kiro/aidlc-common/stages/construction/code-generation.md b/dist/kiro-ide/.kiro/aidlc-common/stages/construction/code-generation.md
index 24424e530..70822cc67 100644
--- a/dist/kiro-ide/.kiro/aidlc-common/stages/construction/code-generation.md
+++ b/dist/kiro-ide/.kiro/aidlc-common/stages/construction/code-generation.md
@@ -12,6 +12,7 @@ for_each: unit-of-work
 workspace_requires: true
 produces:
   - code-generation-plan
+  - unit-test-instructions
   - code-summary
 consumes:
   - artifact: business-logic-model
@@ -49,7 +50,7 @@ scopes:
   - security-patch
   - workshop
 inputs: ALL prior design artifacts for this unit
-outputs: application code + code-generation-plan.md, code-summary.md (under this stage's per-unit record dir, engine-resolved)
+outputs: application code + code-generation-plan.md, unit-test-instructions.md, code-summary.md (under this stage's per-unit record dir, engine-resolved)
 ---
 
 # Code Generation
@@ -133,7 +134,24 @@ Present a structured question to get plan approval before proceeding to generati
 - "Approve Plan" — proceed to code generation
 - "Request Changes" — revise the plan
 
-### Step 4: PART 2 — Generation
+### Step 4: Generate Unit Test Instructions
+
+Create `<record>/construction/{unit-name}/code-generation/unit-test-instructions.md` based on the active test strategy (stage-protocol.md §8 "Test Strategy"):
+
+- **Minimal strategy**: Requirement-driven unit tests (1 test per requirement, happy-path floor per component). ~5-15 tests total.
+- **Standard strategy**: 5-8 tests per component, key behavior coverage
+- **Comprehensive strategy**: 10-15 tests per component, thorough coverage
+
+Include:
+- Test framework setup and configuration
+- How to run unit tests (commands, flags, filters)
+- Expected coverage targets
+- Mocking/stubbing guidance
+- Test data management
+
+This file guides the subagent during code generation and is later consumed by build-and-test for execution.
+
+### Step 5: PART 2 — Generation
 
 Before delegating, display to the user:
 "Generating code for [N] plan steps. This may take several minutes depending on project complexity. I'll show a summary when complete."
@@ -146,12 +164,13 @@ Include in the delegation prompt:
 - Design artifacts for the CURRENT UNIT ONLY (not all units)
 - A 1-2 line summary of each inception-phase artifact with its file path (requirements summary, stories summary, app design summary) — the subagent can Read specific files if it needs full content
 - The approved code-generation-plan.md (full content)
+- The unit-test-instructions.md (full content)
 - Project workspace details (languages, frameworks, conventions from aidlc-state.md)
 - Instructions to execute each plan step sequentially and mark checkboxes as completed
 
 The subagent generates all code, test files, and configuration artifacts in the workspace.
 
-### Step 5: Generate Code Summary
+### Step 6: Generate Code Summary
 
 After subagent completes, create `<record>/construction/{unit-name}/code-generation/code-summary.md` documenting:
 - Files created/modified
@@ -159,11 +178,11 @@ After subagent completes, create `<record>/construction/{unit-name}/code-generat
 - Test coverage summary
 - Any deviations from the plan
 
-### Step 6: Update State
+### Step 7: Update State
 
 Update `<record>/aidlc-state.md`: mark Code Generation for {unit-name} as `[x]` completed and update "Current Status".
 
-### Step 7: Completion
+### Step 8: Completion
 
 Present completion message and approval gate:
 
@@ -186,7 +205,8 @@ Approval gate: strictly 2-option (Approve / Request Changes).
 This stage produces TypeScript/JavaScript code in the active Bolt
 worktree. Generated code lives at the workspace root (NEVER under
 the record dir); the planning + summary artefacts (`code-generation-plan.md`,
-`code-summary.md`) live under `<record>/construction/{unit-name}/code-generation/`.
+`unit-test-instructions.md`, `code-summary.md`) live under
+`<record>/construction/{unit-name}/code-generation/`.
 
 The imported sensors check the code outputs:
 
diff --git a/dist/kiro-ide/.kiro/tools/aidlc-version.ts b/dist/kiro-ide/.kiro/tools/aidlc-version.ts
index 834a24531..9ea354589 100644
--- a/dist/kiro-ide/.kiro/tools/aidlc-version.ts
+++ b/dist/kiro-ide/.kiro/tools/aidlc-version.ts
@@ -1,4 +1,4 @@
 // Hand-edited single source of truth for the AIDLC framework version.
 // Bumped in the same commit that adds the matching ## [N.N.N] heading
 // to CHANGELOG.md. Pinned by tests/unit/t68-version-changelog-sync.test.ts.
-export const AIDLC_VERSION = "2.2.1";
+export const AIDLC_VERSION = "2.2.2";
diff --git a/dist/kiro-ide/.kiro/tools/data/stage-graph.json b/dist/kiro-ide/.kiro/tools/data/stage-graph.json
index 3cf324185..2262a21fb 100644
--- a/dist/kiro-ide/.kiro/tools/data/stage-graph.json
+++ b/dist/kiro-ide/.kiro/tools/data/stage-graph.json
@@ -1853,6 +1853,7 @@
     "workspace_requires": true,
     "produces": [
       "code-generation-plan",
+      "unit-test-instructions",
       "code-summary"
     ],
     "consumes": [
@@ -1913,7 +1914,7 @@
     "reviewer": "aidlc-architecture-reviewer-agent",
     "reviewer_max_iterations": 2,
     "inputs": "ALL prior design artifacts for this unit",
-    "outputs": "application code + code-generation-plan.md, code-summary.md (under this stage's per-unit record dir, engine-resolved)",
+    "outputs": "application code + code-generation-plan.md, unit-test-instructions.md, code-summary.md (under this stage's per-unit record dir, engine-resolved)",
     "rules_in_context": [
       {
         "path": "aidlc/spaces/default/memory/org.md",
@@ -1959,7 +1960,6 @@
     "mode": "inline",
     "produces": [
       "build-instructions",
-      "unit-test-instructions",
       "integration-test-instructions",
       "performance-test-instructions",
       "security-test-instructions",
@@ -1971,6 +1971,10 @@
         "artifact": "code-generation-plan",
         "required": true
       },
+      {
+        "artifact": "unit-test-instructions",
+        "required": true
+      },
       {
         "artifact": "code-summary",
         "required": true
@@ -1995,7 +1999,7 @@
       "workshop"
     ],
     "inputs": "ALL code generation outputs across all units",
-    "outputs": "build-instructions.md, unit-test-instructions.md, integration-test-instructions.md, performance-test-instructions.md, security-test-instructions.md, build-and-test-summary.md, test-results.md (under this stage's record dir, engine-resolved)",
+    "outputs": "build-instructions.md, integration-test-instructions.md, performance-test-instructions.md, security-test-instructions.md, build-and-test-summary.md, test-results.md (under this stage's record dir, engine-resolved)",
     "rules_in_context": [
       {
         "path": "aidlc/spaces/default/memory/org.md",
diff --git a/dist/kiro/.kiro/aidlc-common/stages/construction/build-and-test.md b/dist/kiro/.kiro/aidlc-common/stages/construction/build-and-test.md
index 48592056d..1e2079307 100644
--- a/dist/kiro/.kiro/aidlc-common/stages/construction/build-and-test.md
+++ b/dist/kiro/.kiro/aidlc-common/stages/construction/build-and-test.md
@@ -9,7 +9,6 @@ support_agents:
 mode: inline
 produces:
   - build-instructions
-  - unit-test-instructions
   - integration-test-instructions
   - performance-test-instructions
   - security-test-instructions
@@ -18,6 +17,8 @@ produces:
 consumes:
   - artifact: code-generation-plan
     required: true
+  - artifact: unit-test-instructions
+    required: true
   - artifact: code-summary
     required: true
 requires_stage:
@@ -36,7 +37,7 @@ scopes:
   - security-patch
   - workshop
 inputs: ALL code generation outputs across all units
-outputs: build-instructions.md, unit-test-instructions.md, integration-test-instructions.md, performance-test-instructions.md, security-test-instructions.md, build-and-test-summary.md, test-results.md (under this stage's record dir, engine-resolved)
+outputs: build-instructions.md, integration-test-instructions.md, performance-test-instructions.md, security-test-instructions.md, build-and-test-summary.md, test-results.md (under this stage's record dir, engine-resolved)
 ---
 
 # Build and Test
@@ -51,7 +52,7 @@ Load aidlc-quality-agent (lead) persona from `agents/aidlc-quality-agent.md` and
 
 ### Step 2: Analyze Testing Requirements
 
-Read code generation outputs across all units from `<record>/construction/*/code-generation/code-summary.md`. Review NFR requirements across units (if they exist) to identify performance and security testing needs. Catalog all test types required.
+Read code generation outputs across all units from `<record>/construction/*/code-generation/code-summary.md` and per-unit test instructions from `<record>/construction/*/code-generation/unit-test-instructions.md`. Review NFR requirements across units (if they exist) to identify performance and security testing needs. Catalog all test types required.
 
 ### Step 3: Generate Build Instructions
 
@@ -64,17 +65,14 @@ Create `<record>/construction/build-and-test/build-instructions.md`:
 
 ### Step 4-8: Generate Test Instructions (Strategy-Aware)
 
-Consult the active test strategy from `aidlc-state.md` → `**Test Strategy**` (see stage-protocol.md §8 "Test Strategy"). Generate test instruction files based on the strategy level:
+Consult the active test strategy from `aidlc-state.md` → `**Test Strategy**` (see stage-protocol.md §8 "Test Strategy"). Generate test instruction files based on the strategy level.
 
-**Minimal strategy** — generate ONLY:
-- `unit-test-instructions.md`: Requirement-driven unit tests (1 test per requirement, happy-path floor per component). ~5-15 tests total. Skip all other test types.
+**Minimal strategy** — no additional test instruction files are generated (unit tests are already covered per-unit by code-generation).
 
 **Standard strategy** — generate:
-- `unit-test-instructions.md`: 5-8 tests per component, key behavior coverage
 - `integration-test-instructions.md`: Key boundary tests, cross-unit interaction
 
 **Comprehensive strategy** — generate all applicable:
-- `unit-test-instructions.md`: 10-15 tests per component, thorough coverage
 - `integration-test-instructions.md`: Cross-unit interaction, external dependency handling
 - `performance-test-instructions.md` (IF NFR performance requirements exist): Load testing, benchmarks, regression detection
 - `security-test-instructions.md` (IF NFR security requirements exist): SAST/DAST, auth testing, injection testing
@@ -104,7 +102,7 @@ Create `<record>/construction/build-and-test/build-and-test-summary.md`:
 Attempt to execute the build and test commands documented in the instruction files:
 
 1. **Build**: Run the build commands from `build-instructions.md` via Bash. Capture output.
-2. **Unit tests**: Run the unit test command from `unit-test-instructions.md` via Bash. Capture pass/fail counts.
+2. **Unit tests**: Run the unit test commands from the per-unit `unit-test-instructions.md` files (under `<record>/construction/*/code-generation/`) via Bash. Capture pass/fail counts.
 3. **Integration tests** (if applicable): Run integration test commands. Capture results.
 4. **Report results**: Create or update `<record>/construction/build-and-test/test-results.md` with:
    - Build status (success/failure + output)
@@ -154,7 +152,8 @@ The imported sensors check those outputs:
 - **`required-sections`** verifies each instruction file contains the
   registry default (≥2 H2 headings).
 - **`upstream-coverage`** verifies the prose references the upstream
-  artefacts this stage consumes (`code-generation-plan`, `code-summary`).
+  artefacts this stage consumes (`code-generation-plan`, `unit-test-instructions`,
+  `code-summary`).
 - **`type-check`** runs against any TypeScript/TSX code touched as part
   of test generation (matches `**/*.{ts,tsx}`).
 
diff --git a/dist/kiro/.kiro/aidlc-common/stages/construction/code-generation.md b/dist/kiro/.kiro/aidlc-common/stages/construction/code-generation.md
index 24424e530..70822cc67 100644
--- a/dist/kiro/.kiro/aidlc-common/stages/construction/code-generation.md
+++ b/dist/kiro/.kiro/aidlc-common/stages/construction/code-generation.md
@@ -12,6 +12,7 @@ for_each: unit-of-work
 workspace_requires: true
 produces:
   - code-generation-plan
+  - unit-test-instructions
   - code-summary
 consumes:
   - artifact: business-logic-model
@@ -49,7 +50,7 @@ scopes:
   - security-patch
   - workshop
 inputs: ALL prior design artifacts for this unit
-outputs: application code + code-generation-plan.md, code-summary.md (under this stage's per-unit record dir, engine-resolved)
+outputs: application code + code-generation-plan.md, unit-test-instructions.md, code-summary.md (under this stage's per-unit record dir, engine-resolved)
 ---
 
 # Code Generation
@@ -133,7 +134,24 @@ Present a structured question to get plan approval before proceeding to generati
 - "Approve Plan" — proceed to code generation
 - "Request Changes" — revise the plan
 
-### Step 4: PART 2 — Generation
+### Step 4: Generate Unit Test Instructions
+
+Create `<record>/construction/{unit-name}/code-generation/unit-test-instructions.md` based on the active test strategy (stage-protocol.md §8 "Test Strategy"):
+
+- **Minimal strategy**: Requirement-driven unit tests (1 test per requirement, happy-path floor per component). ~5-15 tests total.
+- **Standard strategy**: 5-8 tests per component, key behavior coverage
+- **Comprehensive strategy**: 10-15 tests per component, thorough coverage
+
+Include:
+- Test framework setup and configuration
+- How to run unit tests (commands, flags, filters)
+- Expected coverage targets
+- Mocking/stubbing guidance
+- Test data management
+
+This file guides the subagent during code generation and is later consumed by build-and-test for execution.
+
+### Step 5: PART 2 — Generation
 
 Before delegating, display to the user:
 "Generating code for [N] plan steps. This may take several minutes depending on project complexity. I'll show a summary when complete."
@@ -146,12 +164,13 @@ Include in the delegation prompt:
 - Design artifacts for the CURRENT UNIT ONLY (not all units)
 - A 1-2 line summary of each inception-phase artifact with its file path (requirements summary, stories summary, app design summary) — the subagent can Read specific files if it needs full content
 - The approved code-generation-plan.md (full content)
+- The unit-test-instructions.md (full content)
 - Project workspace details (languages, frameworks, conventions from aidlc-state.md)
 - Instructions to execute each plan step sequentially and mark checkboxes as completed
 
 The subagent generates all code, test files, and configuration artifacts in the workspace.
 
-### Step 5: Generate Code Summary
+### Step 6: Generate Code Summary
 
 After subagent completes, create `<record>/construction/{unit-name}/code-generation/code-summary.md` documenting:
 - Files created/modified
@@ -159,11 +178,11 @@ After subagent completes, create `<record>/construction/{unit-name}/code-generat
 - Test coverage summary
 - Any deviations from the plan
 
-### Step 6: Update State
+### Step 7: Update State
 
 Update `<record>/aidlc-state.md`: mark Code Generation for {unit-name} as `[x]` completed and update "Current Status".
 
-### Step 7: Completion
+### Step 8: Completion
 
 Present completion message and approval gate:
 
@@ -186,7 +205,8 @@ Approval gate: strictly 2-option (Approve / Request Changes).
 This stage produces TypeScript/JavaScript code in the active Bolt
 worktree. Generated code lives at the workspace root (NEVER under
 the record dir); the planning + summary artefacts (`code-generation-plan.md`,
-`code-summary.md`) live under `<record>/construction/{unit-name}/code-generation/`.
+`unit-test-instructions.md`, `code-summary.md`) live under
+`<record>/construction/{unit-name}/code-generation/`.
 
 The imported sensors check the code outputs:
 
diff --git a/dist/kiro/.kiro/tools/aidlc-version.ts b/dist/kiro/.kiro/tools/aidlc-version.ts
index 834a24531..9ea354589 100644
--- a/dist/kiro/.kiro/tools/aidlc-version.ts
+++ b/dist/kiro/.kiro/tools/aidlc-version.ts
@@ -1,4 +1,4 @@
 // Hand-edited single source of truth for the AIDLC framework version.
 // Bumped in the same commit that adds the matching ## [N.N.N] heading
 // to CHANGELOG.md. Pinned by tests/unit/t68-version-changelog-sync.test.ts.
-export const AIDLC_VERSION = "2.2.1";
+export const AIDLC_VERSION = "2.2.2";
diff --git a/dist/kiro/.kiro/tools/data/stage-graph.json b/dist/kiro/.kiro/tools/data/stage-graph.json
index 3cf324185..2262a21fb 100644
--- a/dist/kiro/.kiro/tools/data/stage-graph.json
+++ b/dist/kiro/.kiro/tools/data/stage-graph.json
@@ -1853,6 +1853,7 @@
     "workspace_requires": true,
     "produces": [
       "code-generation-plan",
+      "unit-test-instructions",
       "code-summary"
     ],
     "consumes": [
@@ -1913,7 +1914,7 @@
     "reviewer": "aidlc-architecture-reviewer-agent",
     "reviewer_max_iterations": 2,
     "inputs": "ALL prior design artifacts for this unit",
-    "outputs": "application code + code-generation-plan.md, code-summary.md (under this stage's per-unit record dir, engine-resolved)",
+    "outputs": "application code + code-generation-plan.md, unit-test-instructions.md, code-summary.md (under this stage's per-unit record dir, engine-resolved)",
     "rules_in_context": [
       {
         "path": "aidlc/spaces/default/memory/org.md",
@@ -1959,7 +1960,6 @@
     "mode": "inline",
     "produces": [
       "build-instructions",
-      "unit-test-instructions",
       "integration-test-instructions",
       "performance-test-instructions",
       "security-test-instructions",
@@ -1971,6 +1971,10 @@
         "artifact": "code-generation-plan",
         "required": true
       },
+      {
+        "artifact": "unit-test-instructions",
+        "required": true
+      },
       {
         "artifact": "code-summary",
         "required": true
@@ -1995,7 +1999,7 @@
       "workshop"
     ],
     "inputs": "ALL code generation outputs across all units",
-    "outputs": "build-instructions.md, unit-test-instructions.md, integration-test-instructions.md, performance-test-instructions.md, security-test-instructions.md, build-and-test-summary.md, test-results.md (under this stage's record dir, engine-resolved)",
+    "outputs": "build-instructions.md, integration-test-instructions.md, performance-test-instructions.md, security-test-instructions.md, build-and-test-summary.md, test-results.md (under this stage's record dir, engine-resolved)",
     "rules_in_context": [
       {
         "path": "aidlc/spaces/default/memory/org.md",
diff --git a/docs/reference/04-stages/construction.md b/docs/reference/04-stages/construction.md
index 2532d7549..b5e2c158d 100644
--- a/docs/reference/04-stages/construction.md
+++ b/docs/reference/04-stages/construction.md
@@ -569,7 +569,7 @@ share infrastructure resources.
 | support_agents    | (none -- focused implementation)                                                                  |
 | mode              | subagent (Task tool subagent_type: aidlc-developer-agent)                                               |
 | Inputs            | ALL prior design artifacts for this unit                                                          |
-| Outputs           | application code (workspace root) + `<record>/construction/{unit-name}/code-generation/` -- code-generation-plan.md, code-summary.md |
+| Outputs           | application code (workspace root) + `<record>/construction/{unit-name}/code-generation/` -- code-generation-plan.md, unit-test-instructions.md, code-summary.md |
 
 ### Purpose
 
@@ -654,9 +654,22 @@ This stage has a **two-part structure**: planning followed by generation.
    - "Approve Plan" -- proceed to code generation
    - "Request Changes" -- revise the plan
 
-#### PART 2 -- Generation (Steps 4-7)
+4. **Generate Unit Test Instructions** -- Create
+   `<record>/construction/{unit-name}/code-generation/unit-test-instructions.md`
+   based on the active test strategy (stage-protocol.md §8 "Test Strategy"):
+   - **Minimal**: Requirement-driven unit tests (1 test per requirement,
+     happy-path floor per component). ~5-15 tests total.
+   - **Standard**: 5-8 tests per component, key behavior coverage
+   - **Comprehensive**: 10-15 tests per component, thorough coverage
+
+   Include: test framework setup and configuration, how to run unit tests
+   (commands, flags, filters), expected coverage targets, mocking/stubbing
+   guidance, test data management. This file guides the subagent during code
+   generation and is later consumed by build-and-test for execution.
+
+#### PART 2 -- Generation (Steps 5-8)
 
-4. **Generate Code** -- Before delegating, display to the user:
+5. **Generate Code** -- Before delegating, display to the user:
    "Generating code for [N] plan steps. This may take several minutes
    depending on project complexity. I'll show a summary when complete."
 
@@ -672,6 +685,7 @@ This stage has a **two-part structure**: planning followed by generation.
      (requirements summary, stories summary, app design summary) -- the
      subagent can Read specific files if it needs full content
    - The approved code-generation-plan.md (full content)
+   - The unit-test-instructions.md (full content)
    - Project workspace details (languages, frameworks, conventions from
      aidlc-state.md)
    - Instructions to execute each plan step sequentially and mark checkboxes
@@ -682,7 +696,7 @@ This stage has a **two-part structure**: planning followed by generation.
    full content. The subagent generates all code, test files, and
    configuration artifacts in the workspace.
 
-5. **Generate Code Summary** -- After subagent completes, create
+6. **Generate Code Summary** -- After subagent completes, create
    `<record>/construction/{unit-name}/code-generation/code-summary.md`
    documenting:
    - Files created/modified
@@ -690,16 +704,17 @@ This stage has a **two-part structure**: planning followed by generation.
    - Test coverage summary
    - Any deviations from the plan
 
-6. **Update State** -- Update `<record>/aidlc-state.md`: mark Code
+7. **Update State** -- Update `<record>/aidlc-state.md`: mark Code
    Generation for {unit-name} as `[x]` completed and update "Current Status".
 
-7. **Completion** -- Present completion message and approval gate.
+8. **Completion** -- Present completion message and approval gate.
 
 ### Outputs
 
 | Artifact                  | Description                                                         |
 |---------------------------|---------------------------------------------------------------------|
 | code-generation-plan.md   | Detailed plan with checkboxes, story traceability, step sequencing  |
+| unit-test-instructions.md | Test framework setup, run commands, coverage targets, mocking       |
 | code-summary.md           | Files created/modified, decisions, test coverage, plan deviations   |
 | (application code)        | All source code, tests, and config written to workspace root        |
 
@@ -745,7 +760,7 @@ Strictly 2-option: Approve / Request Changes.
 | support_agents    | aidlc-devsecops-agent                                                                                   |
 | mode              | inline                                                                                            |
 | Inputs            | ALL code generation outputs across all units                                                      |
-| Outputs           | `<record>/construction/build-and-test/` -- build-instructions.md, unit-test-instructions.md, integration-test-instructions.md, performance-test-instructions.md, security-test-instructions.md, build-and-test-summary.md, test-results.md, plus conditional test instruction files |
+| Outputs           | `<record>/construction/build-and-test/` -- build-instructions.md, integration-test-instructions.md, performance-test-instructions.md, security-test-instructions.md, build-and-test-summary.md, test-results.md, plus conditional test instruction files |
 
 ### Purpose
 
@@ -758,6 +773,8 @@ testing expertise.
 
 - Code generation outputs across all units from
   `<record>/construction/*/code-generation/code-summary.md`
+- Per-unit test instructions from
+  `<record>/construction/*/code-generation/unit-test-instructions.md`
 - NFR requirements across units (if they exist) for performance and security
   testing needs
 
@@ -778,15 +795,7 @@ testing expertise.
    - Build verification steps
    - Troubleshooting common build issues
 
-4. **Generate Unit Test Instructions** -- Create
-   `<record>/construction/build-and-test/unit-test-instructions.md`:
-   - Test framework setup and configuration
-   - How to run unit tests (commands, flags, filters)
-   - Expected test coverage targets
-   - Mocking/stubbing guidance
-   - Test data management
-
-5. **Generate Integration Test Instructions** -- Create
+4. **Generate Integration Test Instructions** -- Create
    `<record>/construction/build-and-test/integration-test-instructions.md`:
    - Test environment prerequisites (databases, services, queues)
    - How to run integration tests
@@ -794,7 +803,7 @@ testing expertise.
    - External dependency handling (stubs, test doubles, sandboxes)
    - Test data setup and teardown
 
-6. **Generate Performance Test Instructions** (CONDITIONAL) -- IF NFR
+5. **Generate Performance Test Instructions** (CONDITIONAL) -- IF NFR
    performance requirements exist for any unit, create
    `performance-test-instructions.md`:
    - Load testing tools and configuration
@@ -803,7 +812,7 @@ testing expertise.
    - Stress and soak test procedures
    - Performance regression detection
 
-7. **Generate Security Test Instructions** (CONDITIONAL) -- IF NFR security
+6. **Generate Security Test Instructions** (CONDITIONAL) -- IF NFR security
    requirements exist for any unit, create
    `security-test-instructions.md`:
    - Security scanning tools (SAST, DAST, dependency audit)
@@ -812,7 +821,7 @@ testing expertise.
    - Compliance verification steps
    - Vulnerability assessment procedures
 
-8. **Generate Additional Test Types** (CONDITIONAL) -- As applicable based on
+7. **Generate Additional Test Types** (CONDITIONAL) -- As applicable based on
    project architecture, create specifically named files:
    - **contract-test-instructions.md**: For microservice APIs --
      consumer-driven contracts, schema validation, API compatibility
@@ -823,7 +832,7 @@ testing expertise.
 
    All files go in `<record>/construction/build-and-test/`.
 
-9. **Generate Build and Test Summary** -- Create
+8. **Generate Build and Test Summary** -- Create
    `<record>/construction/build-and-test/build-and-test-summary.md`:
    - Overall build status and prerequisites
    - Test type inventory (which test types were generated)
@@ -831,13 +840,13 @@ testing expertise.
    - Readiness assessment (build-ready, test-ready, deployment-ready)
    - Known limitations or outstanding items
 
-10. **Execute Build and Tests** -- Attempt to execute the build and test
+9. **Execute Build and Tests** -- Attempt to execute the build and test
     commands documented in the instruction files **via Bash**:
 
     a. **Build**: Run the build commands from build-instructions.md via Bash.
        Capture output.
-    b. **Unit tests**: Run the unit test command from
-       unit-test-instructions.md via Bash. Capture pass/fail counts.
+    b. **Unit tests**: Run the unit test commands from the per-unit
+       unit-test-instructions.md files via Bash. Capture pass/fail counts.
     c. **Integration tests** (if applicable): Run integration test commands.
        Capture results.
     d. **Report results**: Create or update
@@ -859,18 +868,17 @@ testing expertise.
     **On success:** Update the Build and Test Summary with actual results (not
     just instructions).
 
-11. **Update State** -- Update `<record>/aidlc-state.md`: mark Build and
+10. **Update State** -- Update `<record>/aidlc-state.md`: mark Build and
     Test as `[x]` completed and update "Current Status". Mark CONSTRUCTION
     phase as complete.
 
-12. **Completion** -- Present completion message and approval gate.
+11. **Completion** -- Present completion message and approval gate.
 
 ### Outputs
 
 | Artifact                          | Description                                                     | Condition          |
 |-----------------------------------|-----------------------------------------------------------------|--------------------|
 | build-instructions.md             | Dependency install, env setup, build commands, troubleshooting  | Always             |
-| unit-test-instructions.md         | Test framework setup, run commands, coverage targets, mocking   | Always             |
 | integration-test-instructions.md  | Prerequisites, cross-unit testing, external deps, data setup    | Always             |
 | performance-test-instructions.md  | Load testing, NFR scenarios, baselines, stress/soak tests       | If NFR perf exists |
 | security-test-instructions.md     | SAST/DAST, auth testing, injection testing, compliance          | If NFR sec exists  |
diff --git a/docs/reference/16-artifact-vocabulary.md b/docs/reference/16-artifact-vocabulary.md
index 64239dbc9..228bc0b91 100644
--- a/docs/reference/16-artifact-vocabulary.md
+++ b/docs/reference/16-artifact-vocabulary.md
@@ -128,9 +128,10 @@ The canonical names are split so the two never collide on the wire:
 
 - `build-test-results` — emitted by `build-and-test`. Pairs with
   sibling names in that stage: `build-instructions`,
-  `unit-test-instructions`, `integration-test-instructions`,
-  `performance-test-instructions`, `security-test-instructions`,
-  `build-and-test-summary`.
+      `integration-test-instructions`, `performance-test-instructions`,
+  `security-test-instructions`, `build-and-test-summary`.
+- `unit-test-instructions` is produced per-unit by `code-generation`
+  and consumed by `build-and-test`.
 - `load-test-results` — emitted by `performance-validation`. Pairs with
   `load-test-plan` already produced by the same stage.
 
diff --git a/tests/fixtures/designer-export/export.json b/tests/fixtures/designer-export/export.json
index 7244872c5..cc178a366 100644
--- a/tests/fixtures/designer-export/export.json
+++ b/tests/fixtures/designer-export/export.json
@@ -1854,6 +1854,7 @@
       "workspace_requires": true,
       "produces": [
         "code-generation-plan",
+        "unit-test-instructions",
         "code-summary"
       ],
       "consumes": [
@@ -1914,7 +1915,7 @@
       "reviewer": "aidlc-architecture-reviewer-agent",
       "reviewer_max_iterations": 2,
       "inputs": "ALL prior design artifacts for this unit",
-      "outputs": "application code + code-generation-plan.md, code-summary.md (under this stage's per-unit record dir, engine-resolved)",
+      "outputs": "application code + code-generation-plan.md, unit-test-instructions.md, code-summary.md (under this stage's per-unit record dir, engine-resolved)",
       "rules_in_context": [
         {
           "path": "aidlc/spaces/default/memory/org.md",
@@ -1960,7 +1961,6 @@
       "mode": "inline",
       "produces": [
         "build-instructions",
-        "unit-test-instructions",
         "integration-test-instructions",
         "performance-test-instructions",
         "security-test-instructions",
@@ -1972,6 +1972,10 @@
           "artifact": "code-generation-plan",
           "required": true
         },
+        {
+          "artifact": "unit-test-instructions",
+          "required": true
+        },
         {
           "artifact": "code-summary",
           "required": true
@@ -1996,7 +2000,7 @@
         "workshop"
       ],
       "inputs": "ALL code generation outputs across all units",
-      "outputs": "build-instructions.md, unit-test-instructions.md, integration-test-instructions.md, performance-test-instructions.md, security-test-instructions.md, build-and-test-summary.md, test-results.md (under this stage's record dir, engine-resolved)",
+      "outputs": "build-instructions.md, integration-test-instructions.md, performance-test-instructions.md, security-test-instructions.md, build-and-test-summary.md, test-results.md (under this stage's record dir, engine-resolved)",
       "rules_in_context": [
         {
           "path": "aidlc/spaces/default/memory/org.md",
diff --git a/tests/unit/gen-coverage-registry.test.ts b/tests/unit/gen-coverage-registry.test.ts
index 0e5302dba..fd8bb7f69 100644
--- a/tests/unit/gen-coverage-registry.test.ts
+++ b/tests/unit/gen-coverage-registry.test.ts
@@ -840,6 +840,7 @@ describe("mechanismsOf is body-derived (milestone 3)", () => {
     "unit/t18.test.ts",
     "unit/t19.test.ts",
     "unit/t20.test.ts",
+    "unit/t215-code-gen-unit-test-instructions-coverage.test.ts",
     "unit/t27.test.ts",
     "unit/t29.test.ts",
     "unit/t30-hook-session-end.test.ts",
diff --git a/tests/unit/t186-foreach-per-unit-iteration.test.ts b/tests/unit/t186-foreach-per-unit-iteration.test.ts
index 7f36eb873..185e5a38b 100644
--- a/tests/unit/t186-foreach-per-unit-iteration.test.ts
+++ b/tests/unit/t186-foreach-per-unit-iteration.test.ts
@@ -486,7 +486,7 @@ describe("t186 engine-driven per-unit for_each iteration (issue #368)", () => {
   // deadlock the multi-batch swarm, since the swarm re-emits batch 1 and never
   // advances). The guard is scoped to exclude the autonomous swarm, so the report
   // takes the normal forward path, not a per-unit-coverage error.
-  const CG_PRODUCES = ["code-generation-plan", "code-summary"];
+  const CG_PRODUCES = ["code-generation-plan", "unit-test-instructions", "code-summary"];
   test("13: autonomous multi-batch code-generation swarm is not blocked by the coverage guard", () => {
     const proj = seedProject("code-generation", "on");
     // code-generation must be in-flight (not pending) for an approve to be valid;
diff --git a/tests/unit/t215-code-gen-unit-test-instructions-coverage.test.ts b/tests/unit/t215-code-gen-unit-test-instructions-coverage.test.ts
new file mode 100644
index 000000000..4360dcf01
--- /dev/null
+++ b/tests/unit/t215-code-gen-unit-test-instructions-coverage.test.ts
@@ -0,0 +1,167 @@
+// covers: subcommand:aidlc-orchestrate:next
+//
+// Regression test for PR #403: unit-test-instructions moved from
+// build-and-test.produces[] to code-generation.produces[]. The per-unit
+// iteration loop must treat a unit as UNCOVERED when unit-test-instructions.md
+// is missing, even if code-generation-plan.md and code-summary.md exist.
+// mechanism = cli (spawn aidlc-orchestrate.ts next).
+
+import { afterEach, describe, expect, test } from "bun:test";
+import { spawnSync } from "node:child_process";
+import { mkdirSync, writeFileSync } from "node:fs";
+import { join } from "node:path";
+import {
+  AIDLC_SRC,
+  cleanupTestProject,
+  createTestProject,
+  DEFAULT_RECORD_DIR,
+  DEFAULT_SPACE,
+  resetAidlcEnv,
+  seededRecordDir,
+  seededStateFile,
+} from "../harness/fixtures.ts";
+
+resetAidlcEnv();
+
+const BUN = process.execPath;
+const ORCH = join(AIDLC_SRC, "tools", "aidlc-orchestrate.ts");
+const RP = `aidlc/spaces/${DEFAULT_SPACE}/intents/${DEFAULT_RECORD_DIR}`;
+
+const CG_PRODUCES_FULL = ["code-generation-plan", "unit-test-instructions", "code-summary"];
+const CG_PRODUCES_MISSING_UTI = ["code-generation-plan", "code-summary"];
+
+interface Directive {
+  kind?: string;
+  stage?: string;
+  unit?: string;
+  gate?: unknown;
+  produces?: string[];
+  [k: string]: unknown;
+}
+
+const tempDirs: string[] = [];
+afterEach(() => {
+  while (tempDirs.length) cleanupTestProject(tempDirs.pop());
+});
+
+function constructionStateAtCodeGen(): string {
+  return `# AI-DLC State Tracking
+
+## Project Information
+- **Project**: unit-test-instructions coverage test
+- **Project Type**: Greenfield
+- **Scope**: feature
+- **State Version**: 7
+- **Skeleton Stance**: on
+
+## Scope Configuration
+- **Stages to Execute**: all
+- **Stages to Skip**: none
+- **Depth**: Standard
+- **Test Strategy**: Standard
+
+## Stage Progress
+
+### CONSTRUCTION PHASE
+- [x] functional-design — EXECUTE
+- [x] nfr-requirements — EXECUTE
+- [x] nfr-design — EXECUTE
+- [x] infrastructure-design — EXECUTE
+- [-] code-generation — EXECUTE
+- [ ] build-and-test — EXECUTE
+
+### INCEPTION PHASE
+- [-] application-design — EXECUTE
+
+## Current Status
+- **Lifecycle Phase**: CONSTRUCTION
+- **Current Stage**: code-generation
+- **Status**: Running
+`;
+}
+
+function seedProject(): string {
+  const proj = createTestProject();
+  tempDirs.push(proj);
+  writeFileSync(seededStateFile(proj), constructionStateAtCodeGen());
+  return proj;
+}
+
+function seedBoltDag(proj: string, units: string[]): void {
+  writeFileSync(
+    join(seededRecordDir(proj), "runtime-graph.json"),
+    JSON.stringify(
+      {
+        bolt_dag: {
+          units: units.map((name) => ({ name, depends_on: [] })),
+          batches: [units],
+        },
+      },
+      null,
+      2,
+    ),
+  );
+}
+
+function coverUnit(proj: string, unit: string, producesNames: string[]): void {
+  const dir = join(seededRecordDir(proj), "construction", unit, "code-generation");
+  mkdirSync(dir, { recursive: true });
+  for (const name of producesNames) {
+    writeFileSync(join(dir, `${name}.md`), `# ${name} for ${unit}\n`);
+  }
+}
+
+function runNext(proj: string): Directive {
+  const r = spawnSync(BUN, [ORCH, "next", "--project-dir", proj], {
+    encoding: "utf-8",
+    env: (() => {
+      const e = { ...process.env };
+      delete e.AWS_AIDLC_DEFAULT_SCOPE;
+      return e;
+    })(),
+  });
+  try {
+    return JSON.parse((r.stdout ?? "").trim()) as Directive;
+  } catch {
+    throw new Error(
+      `runNext did not emit parseable JSON. status=${r.status}\n${r.stdout}\n${r.stderr}`,
+    );
+  }
+}
+
+describe("t215 code-generation per-unit coverage requires unit-test-instructions (PR #403)", () => {
+  test("1: unit missing unit-test-instructions.md is NOT covered — engine re-emits that unit", () => {
+    const proj = seedProject();
+    seedBoltDag(proj, ["alpha", "beta"]);
+    // Cover alpha with only plan + summary (missing unit-test-instructions)
+    coverUnit(proj, "alpha", CG_PRODUCES_MISSING_UTI);
+    const d = runNext(proj);
+    expect(d.kind).toBe("run-stage");
+    expect(d.stage).toBe("code-generation");
+    // Engine should re-emit alpha since it's not fully covered
+    expect(d.unit).toBe("alpha");
+  }, 30000);
+
+  test("2: unit with all three produces (including unit-test-instructions) IS covered — engine advances", () => {
+    const proj = seedProject();
+    seedBoltDag(proj, ["alpha", "beta"]);
+    // Cover alpha with the full produces list
+    coverUnit(proj, "alpha", CG_PRODUCES_FULL);
+    const d = runNext(proj);
+    expect(d.kind).toBe("run-stage");
+    expect(d.stage).toBe("code-generation");
+    // Engine should advance to beta since alpha is fully covered
+    expect(d.unit).toBe("beta");
+  }, 30000);
+
+  test("3: produces path includes unit-test-instructions.md for the emitted unit", () => {
+    const proj = seedProject();
+    seedBoltDag(proj, ["alpha", "beta"]);
+    const d = runNext(proj);
+    expect(d.kind).toBe("run-stage");
+    expect(d.unit).toBe("alpha");
+    expect(d.produces).toContain(
+      `${RP}/construction/alpha/code-generation/unit-test-instructions.md`,
+    );
+  }, 30000);
+});