tribunal-kit 5.8.4 → 5.8.5

package/.agent/history/memory/MEMORY.md

@@ -1,6 +1,6 @@
 # 🧠 Tribunal Memory Index
 > Auto-generated by `tribunal-kit memory export`. Do not edit manually.
-> Entries: 151 | Semantic: 121 | Procedural: 30 | Episodic: 0 | Working: 0
+> Entries: 221 | Semantic: 177 | Procedural: 44 | Episodic: 0 | Working: 0
 
 ## SEMANTIC (Permanent Facts)
 | ID | Content | Tags | Source | Created |
@@ -126,6 +126,62 @@
 | 147 | [ ] Have I checked the environment for relevant existing implementations? | project-idiom, auto-learned | manual | 1784618632Z |
 | 148 | ❌ **Forbidden:** Declaring a task complete because the output "looks correct." | project-idiom, auto-learned | manual | 1784618632Z |
 | 149 | ✅ **Required:** You are explicitly forbidden from finalizing any task without providing **concrete evidence** (terminal output, passing tests, compile success, or equivalent proof) that your output works as intended. | project-idiom, auto-learned | manual | 1784618632Z |
+| 152 | ❌ Overriding project idioms without explicit justification -> ✅ Idioms represent team decisions; flag deviations with reasoning | project-idiom, auto-learned | manual | 1784705972Z |
+| 153 | ❌ Applying idioms from one project to a different project -> ✅ Idioms are project-specific; verify they apply to the current codebase | project-idiom, auto-learned | manual | 1784705972Z |
+| 154 | ❌ **Forbidden:** Declaring a task complete because the output "looks correct." | project-idiom, auto-learned | manual | 1784705972Z |
+| 155 | ✅ **Required:** You are explicitly forbidden from finalizing any task without providing **concrete evidence** (terminal output, passing tests, compile success, or equivalent proof) that your output works as intended. | project-idiom, auto-learned | manual | 1784705972Z |
+| 156 | [ ] Have I reviewed the user's specific constraints and requests? | project-idiom, auto-learned | manual | 1784705972Z |
+| 157 | [ ] Have I checked the environment for relevant existing implementations? | project-idiom, auto-learned | manual | 1784705972Z |
+| 158 | ❌ **Forbidden:** Declaring a task complete because the output "looks correct." | project-idiom, auto-learned | manual | 1784705972Z |
+| 159 | ✅ **Required:** You are explicitly forbidden from finalizing any task without providing **concrete evidence** (terminal output, passing tests, compile success, or equivalent proof) that your output works as intended. | project-idiom, auto-learned | manual | 1784705972Z |
+| 162 | ❌ Overriding project idioms without explicit justification -> ✅ Idioms represent team decisions; flag deviations with reasoning | project-idiom, auto-learned | manual | 1784706426Z |
+| 163 | ❌ Applying idioms from one project to a different project -> ✅ Idioms are project-specific; verify they apply to the current codebase | project-idiom, auto-learned | manual | 1784706426Z |
+| 164 | ❌ **Forbidden:** Declaring a task complete because the output "looks correct." | project-idiom, auto-learned | manual | 1784706426Z |
+| 165 | ✅ **Required:** You are explicitly forbidden from finalizing any task without providing **concrete evidence** (terminal output, passing tests, compile success, or equivalent proof) that your output works as intended. | project-idiom, auto-learned | manual | 1784706426Z |
+| 166 | [ ] Have I reviewed the user's specific constraints and requests? | project-idiom, auto-learned | manual | 1784706426Z |
+| 167 | [ ] Have I checked the environment for relevant existing implementations? | project-idiom, auto-learned | manual | 1784706426Z |
+| 168 | ❌ **Forbidden:** Declaring a task complete because the output "looks correct." | project-idiom, auto-learned | manual | 1784706426Z |
+| 169 | ✅ **Required:** You are explicitly forbidden from finalizing any task without providing **concrete evidence** (terminal output, passing tests, compile success, or equivalent proof) that your output works as intended. | project-idiom, auto-learned | manual | 1784706426Z |
+| 172 | ❌ Overriding project idioms without explicit justification -> ✅ Idioms represent team decisions; flag deviations with reasoning | project-idiom, auto-learned | manual | 1784706433Z |
+| 173 | ❌ Applying idioms from one project to a different project -> ✅ Idioms are project-specific; verify they apply to the current codebase | project-idiom, auto-learned | manual | 1784706433Z |
+| 174 | ❌ **Forbidden:** Declaring a task complete because the output "looks correct." | project-idiom, auto-learned | manual | 1784706433Z |
+| 175 | ✅ **Required:** You are explicitly forbidden from finalizing any task without providing **concrete evidence** (terminal output, passing tests, compile success, or equivalent proof) that your output works as intended. | project-idiom, auto-learned | manual | 1784706433Z |
+| 176 | [ ] Have I reviewed the user's specific constraints and requests? | project-idiom, auto-learned | manual | 1784706433Z |
+| 177 | [ ] Have I checked the environment for relevant existing implementations? | project-idiom, auto-learned | manual | 1784706433Z |
+| 178 | ❌ **Forbidden:** Declaring a task complete because the output "looks correct." | project-idiom, auto-learned | manual | 1784706433Z |
+| 179 | ✅ **Required:** You are explicitly forbidden from finalizing any task without providing **concrete evidence** (terminal output, passing tests, compile success, or equivalent proof) that your output works as intended. | project-idiom, auto-learned | manual | 1784706433Z |
+| 182 | ❌ Overriding project idioms without explicit justification -> ✅ Idioms represent team decisions; flag deviations with reasoning | project-idiom, auto-learned | manual | 1784706809Z |
+| 183 | ❌ Applying idioms from one project to a different project -> ✅ Idioms are project-specific; verify they apply to the current codebase | project-idiom, auto-learned | manual | 1784706809Z |
+| 184 | ❌ **Forbidden:** Declaring a task complete because the output "looks correct." | project-idiom, auto-learned | manual | 1784706809Z |
+| 185 | ✅ **Required:** You are explicitly forbidden from finalizing any task without providing **concrete evidence** (terminal output, passing tests, compile success, or equivalent proof) that your output works as intended. | project-idiom, auto-learned | manual | 1784706809Z |
+| 186 | [ ] Have I reviewed the user's specific constraints and requests? | project-idiom, auto-learned | manual | 1784706809Z |
+| 187 | [ ] Have I checked the environment for relevant existing implementations? | project-idiom, auto-learned | manual | 1784706809Z |
+| 188 | ❌ **Forbidden:** Declaring a task complete because the output "looks correct." | project-idiom, auto-learned | manual | 1784706809Z |
+| 189 | ✅ **Required:** You are explicitly forbidden from finalizing any task without providing **concrete evidence** (terminal output, passing tests, compile success, or equivalent proof) that your output works as intended. | project-idiom, auto-learned | manual | 1784706809Z |
+| 192 | ❌ Overriding project idioms without explicit justification -> ✅ Idioms represent team decisions; flag deviations with reasoning | project-idiom, auto-learned | manual | 1784706814Z |
+| 193 | ❌ Applying idioms from one project to a different project -> ✅ Idioms are project-specific; verify they apply to the current codebase | project-idiom, auto-learned | manual | 1784706814Z |
+| 194 | ❌ **Forbidden:** Declaring a task complete because the output "looks correct." | project-idiom, auto-learned | manual | 1784706814Z |
+| 195 | ✅ **Required:** You are explicitly forbidden from finalizing any task without providing **concrete evidence** (terminal output, passing tests, compile success, or equivalent proof) that your output works as intended. | project-idiom, auto-learned | manual | 1784706814Z |
+| 196 | [ ] Have I reviewed the user's specific constraints and requests? | project-idiom, auto-learned | manual | 1784706814Z |
+| 197 | [ ] Have I checked the environment for relevant existing implementations? | project-idiom, auto-learned | manual | 1784706814Z |
+| 198 | ❌ **Forbidden:** Declaring a task complete because the output "looks correct." | project-idiom, auto-learned | manual | 1784706814Z |
+| 199 | ✅ **Required:** You are explicitly forbidden from finalizing any task without providing **concrete evidence** (terminal output, passing tests, compile success, or equivalent proof) that your output works as intended. | project-idiom, auto-learned | manual | 1784706814Z |
+| 202 | ❌ Overriding project idioms without explicit justification -> ✅ Idioms represent team decisions; flag deviations with reasoning | project-idiom, auto-learned | manual | 1785138690Z |
+| 203 | ❌ Applying idioms from one project to a different project -> ✅ Idioms are project-specific; verify they apply to the current codebase | project-idiom, auto-learned | manual | 1785138690Z |
+| 204 | ❌ **Forbidden:** Declaring a task complete because the output "looks correct." | project-idiom, auto-learned | manual | 1785138690Z |
+| 205 | ✅ **Required:** You are explicitly forbidden from finalizing any task without providing **concrete evidence** (terminal output, passing tests, compile success, or equivalent proof) that your output works as intended. | project-idiom, auto-learned | manual | 1785138690Z |
+| 206 | [ ] Have I reviewed the user's specific constraints and requests? | project-idiom, auto-learned | manual | 1785138690Z |
+| 207 | [ ] Have I checked the environment for relevant existing implementations? | project-idiom, auto-learned | manual | 1785138690Z |
+| 208 | ❌ **Forbidden:** Declaring a task complete because the output "looks correct." | project-idiom, auto-learned | manual | 1785138690Z |
+| 209 | ✅ **Required:** You are explicitly forbidden from finalizing any task without providing **concrete evidence** (terminal output, passing tests, compile success, or equivalent proof) that your output works as intended. | project-idiom, auto-learned | manual | 1785138690Z |
+| 212 | ❌ Overriding project idioms without explicit justification -> ✅ Idioms represent team decisions; flag deviations with reasoning | project-idiom, auto-learned | manual | 1785138695Z |
+| 213 | ❌ Applying idioms from one project to a different project -> ✅ Idioms are project-specific; verify they apply to the current codebase | project-idiom, auto-learned | manual | 1785138695Z |
+| 214 | ❌ **Forbidden:** Declaring a task complete because the output "looks correct." | project-idiom, auto-learned | manual | 1785138695Z |
+| 215 | ✅ **Required:** You are explicitly forbidden from finalizing any task without providing **concrete evidence** (terminal output, passing tests, compile success, or equivalent proof) that your output works as intended. | project-idiom, auto-learned | manual | 1785138695Z |
+| 216 | [ ] Have I reviewed the user's specific constraints and requests? | project-idiom, auto-learned | manual | 1785138695Z |
+| 217 | [ ] Have I checked the environment for relevant existing implementations? | project-idiom, auto-learned | manual | 1785138695Z |
+| 218 | ❌ **Forbidden:** Declaring a task complete because the output "looks correct." | project-idiom, auto-learned | manual | 1785138695Z |
+| 219 | ✅ **Required:** You are explicitly forbidden from finalizing any task without providing **concrete evidence** (terminal output, passing tests, compile success, or equivalent proof) that your output works as intended. | project-idiom, auto-learned | manual | 1785138696Z |
 
 ## PROCEDURAL (How-To Recipes)
 | ID | Content | Tags | Source | Created |
@@ -160,4 +216,18 @@
 | 141 | Run `npm run validate-payload && jest --coverage` to test the project | build-script, test, auto-learned | manual | 1784396209Z |
 | 150 | Run `echo 'No build step required for this project'` to build the project | build-script, build, auto-learned | manual | 1784618632Z |
 | 151 | Run `npm run validate-payload && jest --coverage` to test the project | build-script, test, auto-learned | manual | 1784618632Z |
+| 160 | Run `echo 'No build step required for this project'` to build the project | build-script, build, auto-learned | manual | 1784705972Z |
+| 161 | Run `npm run validate-payload && jest --coverage` to test the project | build-script, test, auto-learned | manual | 1784705972Z |
+| 170 | Run `echo 'No build step required for this project'` to build the project | build-script, build, auto-learned | manual | 1784706426Z |
+| 171 | Run `npm run validate-payload && jest --coverage` to test the project | build-script, test, auto-learned | manual | 1784706426Z |
+| 180 | Run `echo 'No build step required for this project'` to build the project | build-script, build, auto-learned | manual | 1784706433Z |
+| 181 | Run `npm run validate-payload && jest --coverage` to test the project | build-script, test, auto-learned | manual | 1784706433Z |
+| 190 | Run `echo 'No build step required for this project'` to build the project | build-script, build, auto-learned | manual | 1784706809Z |
+| 191 | Run `npm run validate-payload && jest --coverage` to test the project | build-script, test, auto-learned | manual | 1784706809Z |
+| 200 | Run `echo 'No build step required for this project'` to build the project | build-script, build, auto-learned | manual | 1784706814Z |
+| 201 | Run `npm run validate-payload && jest --coverage` to test the project | build-script, test, auto-learned | manual | 1784706814Z |
+| 210 | Run `npm run build:rust` to build the project | build-script, build, auto-learned | manual | 1785138690Z |
+| 211 | Run `npm run validate-payload && jest --coverage` to test the project | build-script, test, auto-learned | manual | 1785138690Z |
+| 220 | Run `npm run build:rust` to build the project | build-script, build, auto-learned | manual | 1785138696Z |
+| 221 | Run `npm run validate-payload && jest --coverage` to test the project | build-script, test, auto-learned | manual | 1785138696Z |
 

package/.agent/rules/GEMINI.md

@@ -222,7 +222,7 @@ The Human Gate is never skipped. No code is written to a file without explicit u
 | Database/SQL          | logic + security + sql + schema + complexity-reviewer                                               |
 | Mobile/Cross-platform | logic + security + mobile-reviewer + type-safety + complexity-reviewer                              |
 | Any domain            | + performance (if optimization)                                                                     |
-| Before merge          | /tribunal-full (all 20)                                                                             |
+| Before merge          | /tribunal-full (all 19 reviewers)                                                                   |
 
 ---
 

package/.agent/scripts/_utils.js

@@ -226,9 +226,34 @@ function parseArgs(argv, schema = {}) {
 
 // ── Command Runner ──────────────────────────────────────────────────────────
 
+const WINDOWS_CMD_SHIMS = new Set(["npm", "npx", "pnpm", "yarn", "bun", "bunx"]);
+
+/**
+ * Select the executable that should be passed to spawnSync.
+ * Only package-manager shims need a .cmd suffix on Windows; native executables
+ * such as node, cargo, and git must retain their original names.
+ *
+ * @param {string} cmd - Command to run
+ * @param {string} [platform] - Platform override for deterministic testing
+ * @returns {string}
+ */
+function normalizeCommand(cmd, platform = process.platform) {
+  const lowerCaseCommand = cmd.toLowerCase();
+  const hasPath = cmd.includes("/") || cmd.includes("\\\\");
+  if (
+    platform === "win32" &&
+    !hasPath &&
+    !lowerCaseCommand.endsWith(".cmd") &&
+    WINDOWS_CMD_SHIMS.has(lowerCaseCommand)
+  ) {
+    return `${cmd}.cmd`;
+  }
+  return cmd;
+}
+
 /**
  * Run a command synchronously and return a structured result.
- * Handles Windows .cmd shims automatically.
+ * Handles known Windows package-manager shims without rewriting native tools.
  *
  * @param {string} cmd - Command to run
  * @param {string[]} args - Arguments
@@ -237,12 +262,7 @@ function parseArgs(argv, schema = {}) {
  */
 function runCommand(cmd, args = [], opts = {}) {
   const { spawnSync } = require("child_process");
-  const executable =
-    process.platform === "win32" &&
-    !cmd.endsWith(".cmd") &&
-    !cmd.includes(path.sep)
-      ? `${cmd}.cmd`
-      : cmd;
+  const executable = normalizeCommand(cmd);
 
   const result = spawnSync(executable, args, {
     encoding: "utf8",
@@ -275,5 +295,6 @@ module.exports = {
   // CLI
   parseArgs,
   // Commands
+  normalizeCommand,
   runCommand,
 };

package/.agent/scripts/context_broker.js

@@ -355,25 +355,6 @@ const BASELINE_SKILLS = [
 
 // ── Skill catalogue (loaded from disk) ───────────────────────────────────────
 
-/**
- * Find the .agent directory by walking up from cwd.
- * @returns {string} path to .agent/
- */
-function findAgentDir() {
-  let current = path.resolve(process.cwd());
-  const root = path.parse(current).root;
-  while (current !== root) {
-    const candidate = path.join(current, ".agent");
-    if (fs.existsSync(candidate) && fs.statSync(candidate).isDirectory())
-      return candidate;
-    current = path.dirname(current);
-  }
-  console.error(
-    `${RED}✖ .agent/ not found. Run: npx tribunal-kit init${RESET}`,
-  );
-  process.exit(1);
-}
-
 /**
  * Parse the YAML frontmatter from a SKILL.md file.
  * Returns { name, description, ... } or null on parse failure.

package/.agent/scripts/guardrail_engine.js

@@ -104,7 +104,7 @@ function ruleReviewerCount(content, manifest, _ctx) {
   const actualCount = manifest.agents ? manifest.agents.reviewer_count : null;
   if (actualCount === null) return violations;
 
-  const countRegex = /(\d+)[-\s]+(?:parallel\s+)?(?:reviewers?|agents?)/gi;
+  const countRegex = /(\d+)\s*(?:-\s*)?(?:(?:parallel|domain(?:-specific)?|tribunal|code)\s+)*reviewers?\b/gi;
   let match;
   while ((match = countRegex.exec(content)) !== null) {
     const claimed = parseInt(match[1], 10);

package/.agent/scripts/integrity_manifest.js

@@ -82,9 +82,10 @@ function crawlAgents(agentDir) {
     const content = safeReadFile(path.join(agentsDir, file));
     const fm = parseFrontmatter(content);
     const isReviewer =
-      /reviewer|auditor|tester/i.test(name) ||
+      /reviewer|auditor|tester|throughput-optimizer/i.test(name) ||
       (fm.role && /reviewer/i.test(fm.role));
 
+
     if (isReviewer) {
       reviewers.push(name);
     } else {
@@ -239,7 +240,10 @@ function extractNumericClaims(filePath, content, agentDir, actualCounts) {
   const lines = content.split(/\r?\n/);
 
   const claimPatterns = [
-    { regex: /(\d+)[-\s]+(?:parallel\s+)?(?:reviewers?|agents?)/gi, entity: "reviewers" },
+    {
+      regex: /(\d+)\s*(?:-\s*)?(?:(?:parallel|domain(?:-specific)?|tribunal|code)\s+)*reviewers?\b/gi,
+      entity: "reviewers",
+    },
     { regex: /(\d+)\s*(?:specialist\s+)?agents?\b/gi, entity: "agents" },
     { regex: /(\d+)\s*(?:modular\s+)?skills?\b/gi, entity: "skills" },
     { regex: /(\d+)\s*(?:slash\s+)?workflows?\b/gi, entity: "workflows" },
@@ -253,8 +257,10 @@ function extractNumericClaims(filePath, content, agentDir, actualCounts) {
       let match;
       while ((match = regex.exec(line)) !== null) {
         const claimed = parseInt(match[1], 10);
-        // Skip tiny numbers (likely not count claims) and very large numbers
-        if (claimed < 3 || claimed > 500) continue;
+        // This manifest describes the full Tribunal payload, so small local
+        // counts (for example, "3 skills" in one skill group) are not global
+        // asset claims. Skip them along with implausibly large values.
+        if (claimed < 5 || claimed > 500) continue;
 
         const actual = actualCounts[entity];
         if (actual !== undefined) {

package/.agent/scripts/marathon_harness.js

@@ -24,6 +24,7 @@
 const fs = require("fs");
 const path = require("path");
 const { execSync } = require("child_process");
+const { findAgentDir } = require("./_utils");
 
 const {
   GREEN,
@@ -42,7 +43,11 @@ const {
 } = require("./_colors");
 
 // ── Paths ────────────────────────────────────────────────────────────────────
-const MARATHON_DIR = path.resolve(".agent", "history", "marathon");
+function getMarathonDir(startDir = process.cwd()) {
+  return path.join(findAgentDir(startDir), "history", "marathon");
+}
+
+const MARATHON_DIR = getMarathonDir();
 const FEATURE_LIST_FILE = path.join(MARATHON_DIR, "feature_list.json");
 const PROGRESS_FILE = path.join(MARATHON_DIR, "progress.json");
 const ARCHIVE_DIR = path.join(MARATHON_DIR, "archive");
@@ -1049,6 +1054,10 @@ function main() {
   }
 }
 
+module.exports = {
+  getMarathonDir,
+};
+
 if (require.main === module) {
   main();
 }

package/.agent/scripts/skill_evolution.js

@@ -46,11 +46,18 @@ const {
 const { findAgentDir } = require("./_utils");
 
 
-const AGENT_DIR = findAgentDir();
-const SKILL_DIR = path.join(AGENT_DIR, "skills", "project-idioms");
-const SKILL_FILE = path.join(SKILL_DIR, "SKILL.md");
-const HISTORY_DIR = path.join(AGENT_DIR, "history", "skill-evolution");
-const LOG_FILE = path.join(HISTORY_DIR, "digest-log.json");
+function getPaths(startDir = process.cwd()) {
+  const agentDir = findAgentDir(startDir);
+  const skillDir = path.join(agentDir, "skills", "project-idioms");
+  const historyDir = path.join(agentDir, "history", "skill-evolution");
+  return {
+    agentDir,
+    skillDir,
+    skillFile: path.join(skillDir, "SKILL.md"),
+    historyDir,
+    logFile: path.join(historyDir, "digest-log.json"),
+  };
+}
 
 // ── Architectural Weight Patterns ────────────────────────────────────────────
 const HIGH_WEIGHT_PATTERNS = [
@@ -205,8 +212,9 @@ function countTokensEstimate(text) {
 
 // ── Idiom management ──────────────────────────────────────────────────────────
 function loadExistingIdioms() {
-  if (!fs.existsSync(SKILL_FILE)) return [];
-  const content = fs.readFileSync(SKILL_FILE, "utf8");
+  const { skillFile } = getPaths();
+  if (!fs.existsSync(skillFile)) return [];
+  const content = fs.readFileSync(skillFile, "utf8");
   const idioms = [];
   const pattern =
     /\|\s*(\d+)\s*\|\s*`([^`]+)`\s*\|\s*([^|]+)\|\s*([^|]+)\|\s*([^|]+)\|/g;
@@ -346,10 +354,11 @@ function parseLlmYamlResponse(response) {
 
 // ── Log helpers ────────────────────────────────────────────────────────────────
 function loadLog() {
-  fs.mkdirSync(HISTORY_DIR, { recursive: true });
-  if (fs.existsSync(LOG_FILE)) {
+  const { historyDir, logFile } = getPaths();
+  fs.mkdirSync(historyDir, { recursive: true });
+  if (fs.existsSync(logFile)) {
     try {
-      return JSON.parse(fs.readFileSync(LOG_FILE, "utf8"));
+      return JSON.parse(fs.readFileSync(logFile, "utf8"));
     } catch {
       /* fallthrough */
     }
@@ -358,8 +367,9 @@ function loadLog() {
 }
 
 function saveLog(log) {
-  fs.mkdirSync(HISTORY_DIR, { recursive: true });
-  fs.writeFileSync(LOG_FILE, JSON.stringify(log, null, 2), "utf8");
+  const { historyDir, logFile } = getPaths();
+  fs.mkdirSync(historyDir, { recursive: true });
+  fs.writeFileSync(logFile, JSON.stringify(log, null, 2), "utf8");
 }
 
 // ── Auto-LLM API Integration ─────────────────────────────────────────────────
@@ -835,8 +845,9 @@ async function cmdDigest(args) {
   if (added > 0) {
     log.total_idioms = merged.length;
     const skillMd = renderSkillMd(merged, (log.cycles || []).length + 1);
-    fs.mkdirSync(SKILL_DIR, { recursive: true });
-    fs.writeFileSync(SKILL_FILE, skillMd, "utf8");
+    const { skillDir, skillFile } = getPaths();
+    fs.mkdirSync(skillDir, { recursive: true });
+    fs.writeFileSync(skillFile, skillMd, "utf8");
     console.log(`  ${GREEN}✔ ${added} new idiom(s) added to SKILL.md${RESET}`);
   } else {
     console.log(`  ${DIM}  No new unique idioms added to SKILL.md.${RESET}`);
@@ -932,22 +943,24 @@ async function cmdDigest(args) {
 }
 
 function cmdShow() {
-  if (!fs.existsSync(SKILL_FILE)) {
+  const { skillFile } = getPaths();
+  if (!fs.existsSync(skillFile)) {
     console.log(
       `${YELLOW}No project-idioms skill found. Run 'digest' first.${RESET}`,
     );
     return;
   }
-  console.log(fs.readFileSync(SKILL_FILE, "utf8"));
+  console.log(fs.readFileSync(skillFile, "utf8"));
 }
 
 function cmdReset() {
-  if (fs.existsSync(SKILL_FILE)) {
-    fs.unlinkSync(SKILL_FILE);
+  const { skillFile, logFile } = getPaths();
+  if (fs.existsSync(skillFile)) {
+    fs.unlinkSync(skillFile);
     console.log(`${GREEN}✔ project-idioms/SKILL.md deleted.${RESET}`);
   }
-  if (fs.existsSync(LOG_FILE)) {
-    fs.unlinkSync(LOG_FILE);
+  if (fs.existsSync(logFile)) {
+    fs.unlinkSync(logFile);
     console.log(`${GREEN}✔ Digest log cleared.${RESET}`);
   }
   console.log(`${DIM}Run 'digest' to start a fresh evolution cycle.${RESET}`);
@@ -958,7 +971,8 @@ function cmdStatus() {
   const cycles = log.cycles || [];
   const totalSaved = log.total_tokens_saved || 0;
   const totalIdioms = log.total_idioms || 0;
-  const idiomsExist = fs.existsSync(SKILL_FILE);
+  const { skillFile } = getPaths();
+  const idiomsExist = fs.existsSync(skillFile);
 
   console.log(
     `\n${BOLD}${CYAN}━━━ Skill Evolution Status ━━━━━━━━━━━━━━━━━━━━━━━━${RESET}`,
@@ -1033,6 +1047,7 @@ module.exports = {
   architecturalWeight,
   parseLlmYamlResponse,
   loadExistingIdioms,
+  getPaths,
 };
 
 if (require.main === module) {

package/.agent/scripts/swarm_dispatcher.js

@@ -52,7 +52,7 @@ class SwarmDashboard {
   }
 
   start() {
-    console.clear();
+    this.render();
     this.timer = setInterval(() => this.render(), 80);
   }
 
@@ -89,12 +89,63 @@ const MAX_GOAL_LENGTH = 200;
 const MAX_CONTEXT_LENGTH = 800;
 const MAX_WORKERS_PER_SWARM = 5;
 
-const { findAgentDir } = require("./_utils");
+function getBinaryPath() {
+  if (process.env.TRIBUNAL_FORCE_JS === "1") return null;
+  const isWindows = process.platform === "win32";
+  const ext = isWindows ? ".exe" : "";
+  const candidates = [
+    process.env.TRIBUNAL_CORE_PATH,
+    path.resolve(__dirname, "..", "..", "tribunal-kit", "target", "release", `tribunal-core${ext}`),
+    path.resolve(__dirname, "..", "..", "target", "release", `tribunal-core${ext}`),
+    path.resolve(process.cwd(), "target", "release", `tribunal-core${ext}`),
+    path.resolve(process.cwd(), "tribunal-kit", "target", "release", `tribunal-core${ext}`),
+  ];
+  for (const c of candidates) {
+    if (c && fs.existsSync(c)) return c;
+  }
+  return null;
+}
 
+function nativeValidate(file, schemaType) {
+  const bin = getBinaryPath();
+  if (!bin || !file || !fs.existsSync(file)) return null;
+  try {
+    const { spawnSync } = require("child_process");
+    const res = spawnSync(bin, ["validate", "--file", file, "--schema", schemaType], {
+      encoding: "utf-8",
+    });
+    if (res.status === 0) {
+      return true;
+    }
+  } catch {
+    // fallback to JS engine
+  }
+  return null;
+}
+
+function findAgentDir(startPath) {
+  let current = path.resolve(startPath);
+  const root = path.parse(current).root;
+  while (current !== root) {
+    const agentDir = path.join(current, ".agent");
+    if (fs.existsSync(agentDir) && fs.statSync(agentDir).isDirectory()) {
+      return agentDir;
+    }
+    current = path.dirname(current);
+  }
+  return null;
+}
 
 // ─── Legacy mode: validate orchestrator micro-worker payloads ──────────────────
 
-function validatePayload(payloadData, workspaceRoot, agentsDir) {
+function validatePayload(payloadData, workspaceRoot, agentsDir, payloadFile = null) {
+  if (payloadFile) {
+    const nativeResult = nativeValidate(payloadFile, "micro-worker");
+    if (nativeResult === true) {
+      return true;
+    }
+  }
+
   if (!payloadData.dispatch_micro_workers) {
     console.error(
       "ERROR: Payload missing required 'dispatch_micro_workers' array.",
@@ -301,7 +352,14 @@ function validateWorkerResult(res, index) {
   return errors;
 }
 
-function validateSwarmPayload(payloadData, agentsDir) {
+function validateSwarmPayload(payloadData, agentsDir, payloadFile = null) {
+  if (payloadFile) {
+    const nativeResult = nativeValidate(payloadFile, "swarm");
+    if (nativeResult === true) {
+      return true;
+    }
+  }
+
   let items;
   if (typeof payloadData === "object" && payloadData !== null) {
     if (Array.isArray(payloadData)) {
@@ -417,7 +475,7 @@ function main() {
   }
 
   if (mode === "swarm") {
-    if (!validateSwarmPayload(payloadData, agentsDir)) {
+    if (!validateSwarmPayload(payloadData, agentsDir, file)) {
       console.error("ERROR: Swarm payload validation failed.");
       process.exit(1);
     }
@@ -494,7 +552,7 @@ function main() {
       }
     }
   } else {
-    if (!validatePayload(payloadData, workspaceRoot, agentsDir)) {
+    if (!validatePayload(payloadData, workspaceRoot, agentsDir, file)) {
       console.error("ERROR: Payload validation failed.");
       process.exit(1);
     }
@@ -536,6 +594,7 @@ function main() {
 }
 
 module.exports = {
+  SwarmDashboard,
   validateWorkerRequest,
   validateWorkerResult,
   validateSwarmPayload,

package/.agent/skills/better-colors/SKILL.md

@@ -40,15 +40,16 @@ $$\text{Format: } \text{oklch}(L \quad C \quad H [\quad / \quad A])$$
   - `250`: Blue / Indigo
   - `300`: Purple / Violet
 
-### CSS Custom Properties Setup
+### CSS Custom Properties & Relative Color Syntax
 
 ```css
 :root {
   /* Brand Primary: Vibrant Indigo */
   --color-primary: oklch(0.55 0.22 260);
-  --color-primary-hover: oklch(0.50 0.24 260);
-  --color-primary-active: oklch(0.45 0.25 260);
-  --color-primary-subtle: oklch(0.94 0.04 260);
+  /* Modern CSS Relative Color Syntax (Modifying Lightness/Chroma dynamically) */
+  --color-primary-hover: oklch(from var(--color-primary) calc(l - 0.08) c h);
+  --color-primary-active: oklch(from var(--color-primary) calc(l - 0.14) c h);
+  --color-primary-subtle: oklch(from var(--color-primary) 0.95 0.04 h);
 
   /* Neutrals: Subtly Tinted (Chroma ~ 0.015) */
   --bg-surface: oklch(0.99 0.005 260);
@@ -58,6 +59,13 @@ $$\text{Format: } \text{oklch}(L \quad C \quad H [\quad / \quad A])$$
   --border-subtle: oklch(0.88 0.015 260);
 }
 
+/* Native CSS light-dark() declaration mode */
+:root {
+  color-scheme: light dark;
+  --surface-adaptive: light-dark(oklch(0.99 0.005 260), oklch(0.14 0.015 260));
+  --text-adaptive: light-dark(oklch(0.20 0.02 260), oklch(0.96 0.01 260));
+}
+
 @media (prefers-color-scheme: dark) {
   :root {
     --bg-surface: oklch(0.14 0.015 260);

package/.agent/skills/fixing-accessibility/SKILL.md

@@ -19,15 +19,17 @@ Guidelines for detecting, prioritizing, and fixing accessibility defects across
 
 ---
 
-## 1. Priority Priority Matrix
+## 1. Priority Priority Matrix (WCAG 2.2 AA)
 
 | Priority | Category | Critical Requirements |
 | --- | --- | --- |
 | **P1** | **Accessible Names** | Every icon button MUST have `aria-label` or `aria-labelledby`. Decorative icons MUST have `aria-hidden="true"`. |
 | **P2** | **Keyboard Access** | All interactive controls MUST be reachable and operable via `Tab` / `Shift+Tab`, `Space`, `Enter`. NEVER use `<div>` or `<span>` as clickable elements without `role="button"`, `tabIndex={0}`, and key handlers. |
-| **P3** | **Focus & Dialogs** | Modals/Dialogs MUST trap focus within the dialog container while open, return focus to trigger on close, and dismiss on `Escape`. |
-| **P4** | **Color Contrast** | Text MUST satisfy WCAG AA 4.5:1 ratio (3.0:1 for large text). Focus rings MUST have at least 3.0:1 contrast against adjacent background. |
-| **P5** | **Form Validation** | Inputs MUST have explicit `<label>` or `aria-labelledby`. Errors MUST use `aria-invalid="true"` and `aria-describedby` pointing to error text. |
+| **P3** | **Focus & Dialogs (SC 2.4.11)** | Modals/Dialogs MUST trap focus within the dialog container while open, return focus on close, dismiss on `Escape`, and ensure focused items are **never fully obscured by sticky headers/footers**. |
+| **P4** | **Focus Appearance (SC 2.4.13)** | Focus indicators MUST be clearly visible (`:focus-visible`), achieve at least **3:1 contrast** against adjacent background, and have a min **2px thickness**. |
+| **P5** | **Target Size (SC 2.5.8)** | Interactive controls MUST measure at least **24x24 CSS pixels** (min **44x44px** recommended for mobile touch targets) or have sufficient non-intersecting spacing. |
+| **P6** | **Color & Contrast** | Text MUST satisfy WCAG AA 4.5:1 ratio (3.0:1 for large text). Color alone MUST NOT be the only indicator of state or error. |
+| **P7** | **Form Validation** | Inputs MUST have explicit `<label>` or `aria-labelledby`. Errors MUST use `aria-invalid="true"` and `aria-describedby` pointing to error text. |
 
 ---