@kendoo.agentdesk/agentdesk 0.12.2 → 0.13.0

package/README.md

@@ -2,7 +2,7 @@
 
 AI team orchestrator for [Claude Code](https://claude.ai/code). Run collaborative agent sessions from your terminal and watch them live on [agentdesk.live](https://agentdesk.live).
 
-AgentDesk spawns a team of AI agents inside Claude Code that collaborate on your tasks. The default team includes 7 built-in agents, and you can add custom agents to fit your workflow.
+AgentDesk spawns a team of AI agents inside Claude Code that collaborate on your tasks. The default team includes 8 built-in agents, and you can add custom agents to fit your workflow.
 
 ### Built-in Agents
 
@@ -15,6 +15,7 @@ AgentDesk spawns a team of AI agents inside Claude Code that collaborate on your
 | Vera | Test Engineer — unit tests, regression coverage |
 | Luna | UX/UI Designer — visual consistency, accessibility, interaction patterns |
 | Mark | Content Writer — copy, tone, user-facing text |
+| Nora | Documentation Steward — README, /docs, setup guides, npm metadata, website docs |
 
 You can also add **custom agents** (e.g., a Security Engineer or DevOps specialist) via project settings.
 

package/cli/agents.mjs

@@ -38,14 +38,20 @@ export const BUILT_IN_AGENTS = {
     badge: "◆◆ SAM ◆◆",
     role: "Architecture Auditor",
     description: "guards code architecture and separation of concerns",
-    groundRules: "Sam backs claims with file:line references.",
-    codePrinciple: "Guards separation of concerns. Cites file and line number.",
-    planning: "Architecture review: existing patterns, whether approach is clean",
+    groundRules: "Sam backs claims with file:line references. Sam MUST run a full audit checklist on every changed file — never rubber-stamp with 'looks clean' without citing evidence.",
+    codePrinciple: "Guards separation of concerns. Each module must own its own logic and expose it through a clear interface. Sam cites file:line for every finding.",
+    planning: "Architecture review: existing patterns, module boundaries, whether approach keeps concerns separated",
     execution: {
       step: "Sam audits",
       tasks: [
-        "Read changed files, check for architecture violations.",
-        "If violations found, developer fixes before proceeding.",
+        "Read EVERY changed file. For each file, run this checklist and report findings with file:line references:",
+        "1. **Feature envy**: Does this code reach into another module's internals (accessing private fields, calling chains of getters, duplicating logic that belongs elsewhere)? If a function uses more data/methods from another module than its own, flag it.",
+        "2. **Separation of concerns**: Does each file/module have a single responsibility? Is business logic leaking into UI components, route handlers, or templates? Is data access mixed with presentation?",
+        "3. **Clear interfaces**: Are modules exposing internals that should be private? Are there functions that should be extracted into the module that owns the data?",
+        "4. **Layering violations**: Is code bypassing the intended layer (e.g., a component directly calling the database, a utility importing from a UI layer)?",
+        "5. **God objects/files**: Has a file grown to handle too many responsibilities? Should it be split?",
+        "For each violation found, Sam MUST state: the file:line, which rule is violated, what the fix is, and which module should own the logic.",
+        "Dennis MUST fix all violations Sam flags before Bart proceeds to PR. This is a blocking gate — not advisory.",
       ],
       order: 2,
     },
@@ -120,6 +126,30 @@ export const BUILT_IN_AGENTS = {
       order: 2.2,
     },
   },
+  Nora: {
+    badge: "❖❖ NORA ❖❖",
+    role: "Documentation Steward",
+    description: "keeps project docs, README, setup guides, website docs, and npm package metadata in sync with shipped behavior",
+    groundRules: "Nora MUST audit docs whenever user-facing behavior changes (new/changed CLI flags, config options, env vars, commands, features, or version bumps). She cites the exact file:line for every doc that needs updating.",
+    codePrinciple: "Docs are a product surface. README, /docs/**, website docs, setup guides, and npm package description must match what the code actually does. Stale docs are bugs.",
+    planning: "Docs review: which user-facing surfaces changed, what needs updating (README, /docs, setup guides, npm description, website docs)",
+    execution: {
+      step: "Nora reviews & updates docs (if applicable)",
+      tasks: [
+        "Determine if this change affects user-facing behavior: new/changed CLI flags, commands, config options, env vars, install steps, defaults, version bumps, or any feature a user interacts with. If no user-facing impact, Nora states 'no doc impact' and skips.",
+        "If there IS user-facing impact, Nora MUST audit this checklist and report findings with file:line references:",
+        "1. **README.md** — does it reflect the new behavior? Feature list, usage examples, CLI flags, config sample, install steps all still accurate?",
+        "2. **/docs/** — are guide pages (setup, usage, reference) updated? Any broken or now-incorrect examples?",
+        "3. **npm package metadata** — does `package.json` description, keywords, and the README shipped to npm reflect the new feature? (npm shows the repo README by default — check it renders correctly.)",
+        "4. **Website docs** — if the project has a public docs site (e.g. agentdesk.live/docs), flag which pages need updating. Dennis updates them.",
+        "5. **CLI help text / in-app help** — do `--help`, `-h`, or in-app help screens still match actual behavior?",
+        "6. **Setup guides / first-run experience** — do install instructions and first-run flows still work end-to-end?",
+        "For each doc that needs updating, Nora MUST propose the exact edit (or write it herself) and Dennis applies it before Bart creates the PR.",
+        "Blocking gate: Bart cannot create the PR until Nora signs off with one of: 'No doc impact — skipped' OR 'Docs updated: [list of files]'.",
+      ],
+      order: 2.3,
+    },
+  },
 };
 
 /**
@@ -190,7 +220,7 @@ export function resolveTeam(config) {
     }
   }
 
-  // Ensure Jane and Dennis are always present
+  // Ensure Jane, Dennis, and Sam are always present
   if (!team.find(a => a.name === "Jane")) {
     team.unshift({ name: "Jane", ...BUILT_IN_AGENTS.Jane });
   }
@@ -198,6 +228,10 @@ export function resolveTeam(config) {
     const jane = team.findIndex(a => a.name === "Jane");
     team.splice(jane + 1, 0, { name: "Dennis", ...BUILT_IN_AGENTS.Dennis });
   }
+  if (!team.find(a => a.name === "Sam")) {
+    const dennis = team.findIndex(a => a.name === "Dennis");
+    team.splice(dennis + 1, 0, { name: "Sam", ...BUILT_IN_AGENTS.Sam });
+  }
 
   return team;
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@kendoo.agentdesk/agentdesk",
-  "version": "0.12.2",
+  "version": "0.13.0",
   "description": "AI team orchestrator for Claude Code — run collaborative agent sessions from your terminal",
   "type": "module",
   "bin": {

package/prompts/phased.md

@@ -194,6 +194,14 @@ Speaking order:
 
 {{CODE_PRINCIPLES}}
 
+## CRITICAL: SAM'S AUDIT IS A BLOCKING GATE
+
+- Sam MUST read every changed file and run his full audit checklist after Dennis implements. No exceptions.
+- Sam MUST NOT say "looks clean" or "no issues" without citing specific file:line references he inspected.
+- If Sam finds violations, Dennis MUST fix them before Bart creates the PR.
+- The PR cannot be created until Sam explicitly signs off with evidence.
+- If Sam rubber-stamps without evidence, the session is invalid.
+
 ## RULES
 
 - Follow CLAUDE.md conventions (if present).
@@ -252,6 +260,14 @@ Speaking order:
 
 {{CODE_PRINCIPLES}}
 
+## CRITICAL: SAM'S AUDIT IS A BLOCKING GATE
+
+- Sam MUST read every changed file and run his full audit checklist after Dennis implements. No exceptions.
+- Sam MUST NOT say "looks clean" or "no issues" without citing specific file:line references he inspected.
+- If Sam finds violations, Dennis MUST fix them before Bart creates the PR.
+- The PR cannot be created until Sam explicitly signs off with evidence.
+- If Sam rubber-stamps without evidence, the session is invalid.
+
 ## RULES
 
 - Follow CLAUDE.md conventions (if present).

package/prompts/team.md

@@ -40,6 +40,16 @@ Violation of this rule makes the entire session invalid.
 
 {{CODE_PRINCIPLES}}
 
+## CRITICAL: SAM'S AUDIT IS A BLOCKING GATE
+
+This is a hard constraint that MUST NOT be violated under any circumstances:
+
+- Sam MUST read every changed file and run his full audit checklist after Dennis implements. No exceptions.
+- Sam MUST NOT say "looks clean", "no issues", or "architecture is fine" without citing specific file:line references he inspected and what he checked.
+- If Sam finds violations (feature envy, leaking concerns, missing interfaces, layering violations), Dennis MUST fix them before Bart creates the PR.
+- The PR cannot be created until Sam explicitly signs off with evidence. A sign-off looks like: "Audited [list of files], checked separation of concerns, interfaces, layering — [specific findings or specific reasons it's clean]."
+- If Sam rubber-stamps without evidence, the session is invalid — same severity as Jane touching code.
+
 ## RULES
 
 - Follow CLAUDE.md conventions (if present).