sinapse-ai 1.16.0 → 1.19.0

package/.sinapse-ai/development/tasks/undo-last.md

@@ -126,7 +126,7 @@ acceptance-criteria:
 
 - **Tool:** backup-manager
   - **Purpose:** Backup and restore operations
-  - **Source:** .sinapse-ai/utils/backup-manager.js
+  - **Source:** .sinapse-ai/infrastructure/scripts/backup-manager.js
 
 - **Tool:** version-control
   - **Purpose:** Git operations for rollback
@@ -138,10 +138,7 @@ acceptance-criteria:
 
 **Agent-specific code for this task:**
 
-- **Script:** rollback-changes.js
-  - **Purpose:** Rollback to previous state
-  - **Language:** JavaScript
-  - **Location:** .sinapse-ai/scripts/rollback-changes.js
+- N/A
 
 ---
 

package/.sinapse-ai/development/tasks/update-manifest.md

@@ -131,20 +131,13 @@ acceptance-criteria:
   - **Purpose:** File reading, modification, and backup
   - **Source:** Node.js fs module
 
-- **Tool:** ast-parser
-  - **Purpose:** Parse and modify code safely
-  - **Source:** .sinapse-ai/utils/ast-parser.js
-
 ---
 
 ## Scripts
 
 **Agent-specific code for this task:**
 
-- **Script:** modify-file.js
-  - **Purpose:** Safe file modification with backup
-  - **Language:** JavaScript
-  - **Location:** .sinapse-ai/scripts/modify-file.js
+- N/A
 
 ---
 

package/.sinapse-ai/development/tasks/ux-create-wireframe.md

@@ -137,7 +137,7 @@ acceptance-criteria:
 - **Script:** execute-task.js
   - **Purpose:** Generic task execution wrapper
   - **Language:** JavaScript
-  - **Location:** .sinapse-ai/scripts/execute-task.js
+  - **Location:** .sinapse-ai/core/execution/
 
 ---
 

package/.sinapse-ai/development/tasks/ux-ds-scan-artifact.md

@@ -136,11 +136,11 @@ acceptance-criteria:
 
 - **Tool:** task-runner
   - **Purpose:** Task execution and orchestration
-  - **Source:** .sinapse-ai/core/task-runner.js
+  - **Source:** .sinapse-ai/core/execution/
 
 - **Tool:** logger
   - **Purpose:** Execution logging and error tracking
-  - **Source:** .sinapse-ai/utils/logger.js
+  - **Source:** .sinapse-ai/core/execution/
 
 ---
 
@@ -151,7 +151,7 @@ acceptance-criteria:
 - **Script:** execute-task.js
   - **Purpose:** Generic task execution wrapper
   - **Language:** JavaScript
-  - **Location:** .sinapse-ai/scripts/execute-task.js
+  - **Location:** .sinapse-ai/core/execution/
 
 ---
 

package/.sinapse-ai/development/tasks/ux-user-research.md

@@ -136,11 +136,11 @@ acceptance-criteria:
 
 - **Tool:** task-runner
   - **Purpose:** Task execution and orchestration
-  - **Source:** .sinapse-ai/core/task-runner.js
+  - **Source:** .sinapse-ai/core/execution/
 
 - **Tool:** logger
   - **Purpose:** Execution logging and error tracking
-  - **Source:** .sinapse-ai/utils/logger.js
+  - **Source:** .sinapse-ai/core/execution/
 
 ---
 
@@ -151,7 +151,7 @@ acceptance-criteria:
 - **Script:** execute-task.js
   - **Purpose:** Generic task execution wrapper
   - **Language:** JavaScript
-  - **Location:** .sinapse-ai/scripts/execute-task.js
+  - **Location:** .sinapse-ai/core/execution/
 
 ---
 

package/.sinapse-ai/development/tasks/validate-next-story.md

@@ -127,10 +127,6 @@ acceptance-criteria:
 
 **External/shared resources used by this task:**
 
-- **Tool:** validation-engine
-  - **Purpose:** Rule-based validation and reporting
-  - **Source:** .sinapse-ai/utils/validation-engine.js
-
 - **Tool:** schema-validator
   - **Purpose:** JSON/YAML schema validation
   - **Source:** ajv or similar
@@ -141,10 +137,7 @@ acceptance-criteria:
 
 **Agent-specific code for this task:**
 
-- **Script:** run-validation.js
-  - **Purpose:** Execute validation rules and generate report
-  - **Language:** JavaScript
-  - **Location:** .sinapse-ai/scripts/run-validation.js
+- N/A
 
 ---
 

package/.sinapse-ai/development/templates/approval-table.md

@@ -0,0 +1,106 @@
+# Approval Table — Shared Terminal Presentation Convention
+
+> **Reusable presentation protocol.** Any task that asks the user to review or approve a
+> set of artifacts (requirements, epics, stories, tasks, risks, open questions) MUST render
+> them through this convention BEFORE asking for a decision. This is the single source of
+> truth for "scannable terminal table + verdict", so the experience is identical across
+> PRD, epic, story, and spec flows. Do not duplicate ad-hoc formats per task — reference
+> this file.
+
+## Why this exists
+
+Lists decided during planning (FRs, NFRs, epics, stories) used to be reported as counts or
+prose ("✅ 8 FR, 3 NFR"). That forces the user to open a file to actually review. This
+convention renders the real items as a scannable table the user can approve row by row,
+right in the terminal. Output language follows the project's language (PT-BR for the user
+when applicable); the schemas below are language-agnostic.
+
+## The protocol (3 parts, always in this order)
+
+1. **Verdict header** — one scannable line stating what is being reviewed and the current
+   status. Format:
+   ```
+   ── {ARTIFACT} REVIEW · {n} item(s) · status: {DRAFT|READY|APPROVED|NEEDS REVISION} ──
+   ```
+2. **The table** — render the items using the matching schema below. Keep it scannable:
+   one row per item, truncate long cells to ~60 chars with `…`, never wrap a cell across
+   lines. Use the priority/severity emoji legend when the schema has one.
+3. **The decision prompt** — after the table, ask for the decision in the form the calling
+   task requires (the 1-9 elicitation menu, or a GO / NO-GO, or `[GO] [PAUSE] [REVISE] [ABORT]`).
+   Never ask for approval without showing the table first.
+
+## Standard schemas
+
+Use the schema that matches the artifact. Add/trim columns only when the data genuinely
+requires it — keep the core columns stable so the experience is consistent.
+
+### Requirements (gather / spec)
+```
+| ID    | Description                          | Priority | Source            |
+| ----- | ------------------------------------ | -------- | ----------------- |
+| FR-1  | {functional requirement}             | 🔴 P0    | requirements.json |
+
+| ID    | Category    | Requirement              | Metric / Target   |
+| ----- | ----------- | ------------------------ | ----------------- |
+| NFR-1 | {category}  | {non-functional req}     | {measurable}      |
+
+| ID    | Type     | Constraint               | Impact            |
+| ----- | -------- | ------------------------ | ----------------- |
+| CON-1 | {type}   | {constraint}             | {impact}          |
+```
+
+### Epic list (PRD epic-list approval)
+```
+| #  | Epic                          | Goal (1 line)                         | Stories |
+| -- | ----------------------------- | ------------------------------------- | ------- |
+| 1  | Foundation & Core Infra       | {one-sentence goal}                   | ~{n}    |
+```
+
+### Epic detail — stories within an epic
+```
+| Story | As a … / I want …                 | ACs | Depends on |
+| ----- | --------------------------------- | --- | ---------- |
+| 1.1   | {user} wants {action}             | {n} | —          |
+```
+
+### Story summary (create / validate story)
+```
+| Field        | Value                                            |
+| ------------ | ------------------------------------------------ |
+| Story        | {epicNum}.{storyNum} — {title}                   |
+| Status       | {Draft \| Ready \| InProgress \| Done}           |
+| Acceptance   | {n} criteria                                     |
+| Tasks        | {n} (linked to AC: {list})                       |
+| Dependencies | {list or —}                                      |
+
+| AC# | Criterion                                  | Tasks      |
+| --- | ------------------------------------------ | ---------- |
+| 1   | {given/when/then}                          | T1, T3     |
+```
+
+### Risks & open questions (spec / plan)
+```
+| Risk            | Probability   | Impact        | Mitigation        |
+| --------------- | ------------- | ------------- | ----------------- |
+| {risk}          | 🟢/🟡/🔴      | 🟢/🟡/🔴      | {mitigation}      |
+
+| ID   | Question                  | Blocking | Assigned To |
+| ---- | ------------------------- | -------- | ----------- |
+| OQ-1 | {open question}           | Yes / No | @{agent}    |
+```
+
+## Legends
+
+- **Priority:** 🔴 P0 (must) · 🟠 P1 (should) · 🟡 P2 (could) · 🟢 P3 (won't-now)
+- **Severity / level:** 🔴 High · 🟡 Medium · 🟢 Low
+- **Status glyphs (optional inline):** ✓ done · ● in progress · ○ pending · ✗ blocked
+
+## Rules
+
+- Always render the table for a list ≥ 2 items. For a single item, a compact field table is
+  still preferred over prose.
+- The table is in addition to writing the artifact file — never replace the file, and never
+  replace the table with a count.
+- Keep within terminal width: prefer fewer, tighter columns over a wide table that wraps.
+- When iterating (NEEDS REVISION), re-render the full table with the changed rows marked
+  (prefix the changed cell with `» `), so the user sees what moved.

package/.sinapse-ai/development/templates/chrome-brain/scripts/chrome-ensure.cjs

@@ -53,7 +53,7 @@ function isAlive(port) {
       (res) => {
         res.resume();
         resolve(res.statusCode === 200);
-      }
+      },
     );
     req.on('error', () => resolve(false));
     req.on('timeout', () => {

package/.sinapse-ai/development/workflows/spec-pipeline.yaml

@@ -208,6 +208,7 @@ workflow:
 
       on_success:
         log: "✅ Requirements gathered: {requirements.functional.length} FR, {requirements.nonFunctional.length} NFR"
+        present_table: ".sinapse-ai/development/templates/approval-table.md"  # Render FR/NFR/CON as scannable approval tables, not just the count
         next: assess
 
       on_failure:

package/.sinapse-ai/infrastructure/scripts/changelog-generator.js

@@ -110,9 +110,13 @@ class ChangelogGenerator {
    */
   async getLastReleaseTag() {
     try {
-      const tag = execSync('git describe --tags --abbrev=0 2>/dev/null', {
+      // Suppress stderr via stdio, not `2>/dev/null` — the redirect breaks the
+      // command on Windows cmd, which would wrongly fall through to the first
+      // commit and generate the changelog from the start of history.
+      const tag = execSync('git describe --tags --abbrev=0', {
         cwd: this.rootPath,
         encoding: 'utf8',
+        stdio: ['ignore', 'pipe', 'ignore'],
       }).trim();
       return tag;
     } catch {

package/.sinapse-ai/infrastructure/scripts/pattern-extractor.js

@@ -797,7 +797,12 @@ export function Button({ variant = 'primary', size = 'md', disabled, children }:
     const patterns = [];
     const seenPatterns = new Set();
 
-    const hookFiles = files.filter((f) => f.includes('/hooks/') || f.includes('use'));
+    // Normalize separators: the walk yields '\\' on Windows, so '/hooks/' would
+    // never match there and hook detection would silently diverge by OS.
+    const hookFiles = files.filter((f) => {
+      const n = f.replace(/\\/g, '/');
+      return n.includes('/hooks/') || n.includes('use');
+    });
 
     for (const file of hookFiles) {
       const content = await this.readFile(file);
@@ -1152,7 +1157,11 @@ describe('MyComponent', () => {
     const patterns = [];
     const seenPatterns = new Set();
 
-    const utilFiles = files.filter((f) => f.includes('/utils/') || f.includes('/lib/') || f.includes('/helpers/'));
+    // Normalize separators ('\\' on Windows) so the '/dir/' checks match cross-OS.
+    const utilFiles = files.filter((f) => {
+      const n = f.replace(/\\/g, '/');
+      return n.includes('/utils/') || n.includes('/lib/') || n.includes('/helpers/');
+    });
 
     for (const file of utilFiles) {
       const content = await this.readFile(file);

package/.sinapse-ai/infrastructure/scripts/test-utilities.js

@@ -11,14 +11,41 @@
 
 const fs = require('fs');
 const path = require('path');
-const { exec } = require('child_process');
-const util = require('util');
-const execPromise = util.promisify(exec);
 
 const utilsDir = path.join(__dirname);
 let results = [];
 let utilities = [];
 
+/**
+ * Recursively count lines containing `needle` under `dir` (the cross-platform
+ * equivalent of `grep -r needle dir | wc -l`). Pure Node — no shell, no /bin/bash.
+ */
+function countMatchingLines(dir, needle) {
+  let count = 0;
+  let entries;
+  try {
+    entries = fs.readdirSync(dir, { withFileTypes: true });
+  } catch {
+    return 0;
+  }
+  for (const entry of entries) {
+    const full = path.join(dir, entry.name);
+    if (entry.isDirectory()) {
+      count += countMatchingLines(full, needle);
+    } else if (entry.isFile()) {
+      try {
+        const content = fs.readFileSync(full, 'utf8');
+        for (const line of content.split('\n')) {
+          if (line.includes(needle)) count++;
+        }
+      } catch {
+        // binary or unreadable file — skip
+      }
+    }
+  }
+  return count;
+}
+
 /**
  * Count integration references for a utility
  */
@@ -34,16 +61,10 @@ async function countIntegrationReferences(utilityName) {
 
   for (const dir of searchDirs) {
     if (!fs.existsSync(dir)) continue;
-
-    try {
-      const { stdout } = await execPromise(
-        `grep -r "${basename}" ${dir} 2>/dev/null | wc -l`,
-        { shell: '/bin/bash' },
-      );
-      totalCount += parseInt(stdout.trim()) || 0;
-    } catch {
-      // Directory doesn't exist or grep failed - not a problem
-    }
+    // Walk the tree in Node and count matching lines. Replaces a
+    // `grep -r ... | wc -l` shell pipeline hardcoded to `/bin/bash`, which does
+    // not exist on Windows (the call always failed there and returned 0).
+    totalCount += countMatchingLines(dir, basename);
   }
 
   return totalCount;

package/.sinapse-ai/infrastructure/scripts/usage-analytics.js

@@ -607,11 +607,15 @@ class UsageAnalytics {
   }
 
   determineContextType(filePath) {
-    if (filePath.includes('/tests/')) return 'test';
-    if (filePath.includes('/docs/')) return 'documentation';
-    if (filePath.includes('/utils/')) return 'utility';
-    if (filePath.includes('/tasks/')) return 'task';
-    if (filePath.includes('/agents/')) return 'agent';
+    // Normalize '\\' -> '/' first: filePath comes from path.relative, which yields
+    // backslashes on Windows, so the '/segment/' checks would all miss there and
+    // every file would be classified 'general'.
+    const p = filePath.replace(/\\/g, '/');
+    if (p.includes('/tests/')) return 'test';
+    if (p.includes('/docs/')) return 'documentation';
+    if (p.includes('/utils/')) return 'utility';
+    if (p.includes('/tasks/')) return 'task';
+    if (p.includes('/agents/')) return 'agent';
     return 'general';
   }
 

package/.sinapse-ai/infrastructure/scripts/validate-claude-integration.js

@@ -38,10 +38,18 @@ function validateClaudeIntegration(options = {}) {
     warnings.push(`Claude hooks dir not found yet: ${path.relative(projectRoot, hooksDir)}`);
   }
 
+  // The Claude mirror legitimately contains MORE files than the core source dir:
+  // ide-sync mirrors core agents (.sinapse-ai/development/agents) PLUS every squad
+  // -orqx PLUS redirect stubs. So an exact count match is the wrong invariant — it
+  // produced a spurious "30/12" warning. What matters here is that no core agent is
+  // missing; exact file-by-file parity (drift/orphaned) is covered by
+  // `npm run sync:ide:check`. We only flag an under-count.
   const sourceCount = countMarkdownFiles(sourceAgentsDir);
   const claudeCount = countMarkdownFiles(agentsDir);
-  if (sourceCount > 0 && claudeCount !== sourceCount) {
-    warnings.push(`Claude agent count differs from source (${claudeCount}/${sourceCount})`);
+  if (sourceCount > 0 && claudeCount < sourceCount) {
+    warnings.push(
+      `Claude mirror has fewer agents than core source (${claudeCount} < ${sourceCount}) — run \`npm run sync:ide\``,
+    );
   }
 
   return {