olympus-ai 4.4.12 → 4.4.13

package/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "olympus-ai",
-  "version": "4.4.12",
+  "version": "4.4.13",
   "description": "Olympus: Multi-agent orchestration for Claude Code. Summon the gods of code.",
   "author": {
     "name": "mikev10",

package/dist/installer/index.d.ts

@@ -24,7 +24,7 @@ export declare const HOOKS_DIR: string;
 export declare const SETTINGS_FILE: string;
 export declare const VERSION_FILE: string;
 /** Current version - MUST match package.json */
-export declare const VERSION = "4.4.12";
+export declare const VERSION = "4.4.13";
 /** Installation result */
 export interface InstallResult {
     success: boolean;

package/dist/installer/index.js

@@ -40,7 +40,7 @@ export const HOOKS_DIR = join(CLAUDE_CONFIG_DIR, 'hooks');
 export const SETTINGS_FILE = join(CLAUDE_CONFIG_DIR, 'settings.json');
 export const VERSION_FILE = join(CLAUDE_CONFIG_DIR, '.olympus-version.json');
 /** Current version - MUST match package.json */
-export const VERSION = '4.4.12';
+export const VERSION = '4.4.13';
 /**
  * Read a content file from the resources/ directory.
  *

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "olympus-ai",
-  "version": "4.4.12",
+  "version": "4.4.13",
   "description": "Olympus: Multi-agent orchestration for Claude Code. Summon the gods of code.",
   "type": "module",
   "main": "dist/index.js",

package/resources/rules/common/markdown-formatting.md

@@ -77,9 +77,9 @@ All 54 active markdownlint rules organized by category. Rules disabled by projec
 | MD010 | `no-hard-tabs`         | No tab characters — use spaces only                        |
 | MD014 | `commands-show-output` | Don't prefix commands with `$` unless showing output below |
 | MD031 | `blanks-around-fences` | One blank line BEFORE and AFTER every fenced code block    |
-| MD040 | `fenced-code-language` | Fenced code blocks MUST specify a language (e.g., ` ```typescript `) |
-| MD046 | `code-block-style` | Use consistent code block style — use **fenced** (backtick) style, not indented |
-| MD048 | `code-fence-style` | Use consistent fence delimiters — use **backticks** (`` ` ``), not tildes (`~`) |
+| MD040 | `fenced-code-language` | Fenced code blocks MUST specify a language (e.g., ` ```typescript `)            |
+| MD046 | `code-block-style`     | Use consistent code block style — use **fenced** (backtick) style, not indented |
+| MD048 | `code-fence-style`     | Use consistent fence delimiters — use **backticks** (`` ` ``), not tildes (`~`) |
 
 ### Whitespace and Formatting
 

package/resources/rules/inception/requirements-analysis.md

@@ -94,7 +94,7 @@ Analyze whatever the user has provided:
 **MANDATORY**: Evaluate ALL of these areas and ask questions for ANY that are unclear:
 - **Functional Requirements**: Core features, user interactions, system behaviors
 - **Non-Functional Requirements**: Performance, security, scalability, usability
-- **User Scenarios**: Use cases, user journeys, edge cases, error scenarios
+- **Scenario Coverage**: Use cases, user journeys, edge cases, error scenarios
 - **Business Context**: Goals, constraints, success criteria, stakeholder needs
 - **Technical Context**: Integration points, data requirements, system boundaries
 - **Quality Attributes**: Reliability, maintainability, testability, accessibility
@@ -188,8 +188,10 @@ When done, let me know and I will incorporate your approved findings into the re
      - Scope estimate
      - Complexity estimate
    - Include both functional and non-functional requirements
+   - Include scenario coverage when relevant (use cases, edge cases, error scenarios)
    - Incorporate user's answers to clarifying questions
    - Provide brief summary of key requirements
+   - **Do NOT include a User Stories section** — user stories with personas and acceptance criteria are generated in the dedicated User Stories stage
 
 ### Step 8: MANDATORY: Update State Tracking
 

package/resources/skills/plan/SKILL.md

@@ -895,10 +895,19 @@ Write result to `aidlc-docs/{workflowId}/inception/prfaq.md`. If generation fail
   Task(
     subagent_type="momus",
     description="Workflow planning review",
-    prompt="Critically review this inception plan for: (1) gaps in requirements, (2) unrealistic acceptance criteria, (3) missing edge cases, (4) architectural risks, (5) incorrect depth/risk assessment. Intent: {intent.md}. Requirements: {requirements.md}. Execution plan: {execution-plan.md}."
+    prompt="Critically review this inception plan for: (1) gaps in requirements, (2) unrealistic acceptance criteria, (3) missing edge cases, (4) architectural risks, (5) incorrect depth/risk assessment. Intent: {intent.md}. Requirements: {requirements.md}. Execution plan: {execution-plan.md}. For each finding, include a clear recommendation stating whether the issue should be fixed now, deferred, or is informational only."
   )
   ```
-  Save output to `aidlc-docs/{workflowId}/inception/intent-review.md` with metadata (reviewer: momus, trigger: automatic, trust level, verdict). Present Momus feedback to the user. Address critical issues.
+
+  **Write findings to file**: After Momus returns, create `aidlc-docs/{workflowId}/inception/momus-review.md` with metadata (reviewer: momus, trigger: automatic, trust level, verdict) and the full review. Format each finding with:
+  - The finding description and severity (critical, important, minor)
+  - **Recommendation**: Whether to fix now (with suggested change), defer to a later stage, or note as informational — be explicit about the recommended action
+  - **Decision**: `[ ]` — empty checkbox for the user to mark with their decision
+  - **User Comments**: blank line for the user to add notes
+
+  Include a response guide at the top of the file explaining how to review (check findings to incorporate, add comments, or skip).
+
+  Do not silently incorporate or fix findings. The user must review the file and respond before any changes are made. This is a human approval gate.
 
 - **Trust 2+** (and not Risk Tier 3): Tell the user: "Optional: Run `/review` for Momus feedback on the inception plan."
 
@@ -923,12 +932,13 @@ Write result to `aidlc-docs/{workflowId}/inception/prfaq.md`. If generation fail
 - `aidlc-docs/{workflowId}/inception/plans/execution-plan.md`
 - `aidlc-docs/{workflowId}/inception/plans/workflow-routing.md`
 - `aidlc-docs/{workflowId}/inception/prfaq.md` (if generated)
-- `aidlc-docs/{workflowId}/inception/intent-review.md` (if Momus ran)
+- `aidlc-docs/{workflowId}/inception/momus-review.md` (if Momus ran)
 
 ### What needs your review
 - [ ] Depth assessment ({SHALLOW|MEDIUM|DEEP}, score {N}/30) is appropriate
 - [ ] Risk tier ({1|2|3}) correctly reflects implementation risk
 - [ ] Execution plan covers all required phases and stages
+- [ ] Momus review findings reviewed (if applicable) — check findings to incorporate, add comments in the file
 
 ---
 
@@ -1280,7 +1290,7 @@ Update `aidlc-docs/{workflowId}/checkpoint.json`:
 4. **CHECKPOINTS ARE MANDATORY**: Save checkpoint state after every stage transition (update inception_stages, current_inception_stage, state file, audit). This enables resume on interruption.
 5. **TRUST ADJUSTS CEREMONY**: Higher trust = fewer questions + lighter gates. Lower trust = more thorough validation.
 6. **REVIEW REQUIRED AFTER EVERY STAGE**: Use the exact REVIEW REQUIRED / WHAT'S NEXT format after each stage completes.
-7. **RESEARCH IS SILENT**: Agent research dispatches (explore, librarian) happen without announcing them to the user. Only surface findings in the artifacts. **Exception**: Metis blind spot analysis writes findings to a dedicated file for user review — do not silently incorporate.
+7. **RESEARCH IS SILENT**: Agent research dispatches (explore, librarian) happen without announcing them to the user. Only surface findings in the artifacts. **Exception**: Metis blind spot analysis and Momus review findings write to dedicated files for user review — do not silently incorporate or fix.
 8. **STATE TRACKING IS TRIPLE**: Every stage update must write to checkpoint.json + aidlc-state.md + audit.md.
 9. **RESUME IS IDEMPOTENT**: Each stage checks its `inception_stages` entry before executing. `completed` or `skipped` → skip to next. `in_progress` with `questions_file` set → resume Q&A without regenerating.