@zhijiewang/openharness 0.11.1 → 0.12.0

package/README.md

@@ -17,13 +17,34 @@
 
 AI coding agent in your terminal. Works with any LLM -- free local models or cloud APIs.
 
-[![npm version](https://img.shields.io/npm/v/@zhijiewang/openharness)](https://www.npmjs.com/package/@zhijiewang/openharness) [![npm downloads](https://img.shields.io/npm/dm/@zhijiewang/openharness)](https://www.npmjs.com/package/@zhijiewang/openharness) [![license](https://img.shields.io/npm/l/@zhijiewang/openharness)](LICENSE) ![Node.js 18+](https://img.shields.io/badge/node-18%2B-green) ![TypeScript](https://img.shields.io/badge/typescript-strict-blue) [![GitHub stars](https://img.shields.io/github/stars/zhijiewong/openharness)](https://github.com/zhijiewong/openharness) [![GitHub issues](https://img.shields.io/github/issues-raw/zhijiewong/openharness)](https://github.com/zhijiewong/openharness/issues) [![PRs Welcome](https://img.shields.io/badge/PRs-welcome-brightgreen)](https://github.com/zhijiewong/openharness/pulls)
+<p align="center">
+  <img src="assets/openharness_v0.11.1_4.gif" alt="OpenHarness demo" width="800" />
+</p>
 
----
+[![npm version](https://img.shields.io/npm/v/@zhijiewang/openharness)](https://www.npmjs.com/package/@zhijiewang/openharness) [![npm downloads](https://img.shields.io/npm/dm/@zhijiewang/openharness)](https://www.npmjs.com/package/@zhijiewang/openharness) [![license](https://img.shields.io/npm/l/@zhijiewang/openharness)](LICENSE) ![tests](https://img.shields.io/badge/tests-624-brightgreen) ![tools](https://img.shields.io/badge/tools-35-blue) ![Node.js 18+](https://img.shields.io/badge/node-18%2B-green) ![TypeScript](https://img.shields.io/badge/typescript-strict-blue) [![GitHub stars](https://img.shields.io/github/stars/zhijiewong/openharness)](https://github.com/zhijiewong/openharness) [![GitHub issues](https://img.shields.io/github/issues-raw/zhijiewong/openharness)](https://github.com/zhijiewong/openharness/issues) [![PRs Welcome](https://img.shields.io/badge/PRs-welcome-brightgreen)](https://github.com/zhijiewong/openharness/pulls)
 
-<video src="https://github.com/user-attachments/assets/ed19a2cc-14d3-4db3-aa5b-3dc07c444498" controls width="100%"></video>
+---
 
-*OpenHarness reading files, running commands, and editing code — powered by a local Ollama model.*
+## Table of Contents
+
+- [Quick Start](#quick-start)
+- [Why OpenHarness?](#why-openharness)
+- [Terminal UI](#terminal-ui)
+- [Tools (35)](#tools-35)
+- [Slash Commands (33)](#slash-commands-33)
+- [Permission Modes](#permission-modes)
+- [Hooks](#hooks)
+- [Checkpoints & Rewind](#checkpoints--rewind)
+- [Agent Roles](#agent-roles)
+- [Headless Mode & CI/CD](#headless-mode)
+- [Cybergotchi](#cybergotchi)
+- [MCP Servers](#mcp-servers)
+- [Providers](#providers)
+- [FAQ](#faq)
+- [Install](#install)
+- [Development](#development)
+- [Contributing](#contributing)
+- [Community](#community)
 
 ---
 
@@ -42,7 +63,17 @@ oh                                    # auto-detect local model
 oh --model ollama/qwen2.5:7b         # specific model
 oh --model gpt-4o                     # cloud model (needs OPENAI_API_KEY)
 oh --trust                            # auto-approve all tool calls
-oh run "fix the tests" --json         # headless mode for CI/CD
+oh --auto                             # auto-approve, block dangerous bash
+oh -p "fix the tests" --trust         # headless mode (single prompt, exit)
+oh run "review code" --json           # CI/CD with JSON output
+```
+
+**In-session commands:**
+```
+/rewind                               # undo last AI file change (checkpoint restore)
+/roles                                # list agent specializations
+/vim                                  # toggle vim mode
+Ctrl+O                                # flush transcript to scrollback for review
 ```
 
 ## Why OpenHarness?
@@ -53,7 +84,7 @@ Most AI coding agents are locked to one provider or cost $20+/month. OpenHarness
 |---|---|---|---|---|
 | Any LLM | Yes (Ollama, OpenAI, Anthropic, OpenRouter, any OpenAI-compatible) | Anthropic only | Yes | Yes |
 | Free local models | Ollama native | No | Yes | Yes |
-| Tools | 25 with permission gates | 43+ | File-focused | 20+ |
+| Tools | 35 with permission gates | 43+ | File-focused | 20+ |
 | Permission modes | 7 (ask, trust, deny, acceptEdits, plan, auto, bypass) | 7 | Basic | Basic |
 | Git integration | Auto-commit + /undo + /rewind checkpoints | Yes | Deep git | Basic |
 | Slash commands | 30+ built-in | 80+ | Some | Some |
@@ -127,33 +158,59 @@ statusLineFormat: '{model} │ {tokens} │ {cost} │ {ctx}'
 
 Available variables: `{model}`, `{tokens}` (input↑ output↓), `{cost}` ($X.XXXX), `{ctx}` (context usage bar). Empty sections are automatically collapsed.
 
-## Tools (25)
+## Tools (35)
 
 | Tool | Risk | Description |
 |------|------|-------------|
-| Bash | high | Execute shell commands with live streaming output |
-| Read | low | Read files with line ranges |
+| **Core** | | |
+| Bash | high | Execute shell commands with live streaming output (AST safety analysis) |
+| Read | low | Read files with line ranges, PDF support |
 | ImageRead | low | Read images/PDFs for multimodal analysis |
 | Write | medium | Create or overwrite files |
 | Edit | medium | Search-and-replace edits |
+| MultiEdit | medium | Atomic multi-file edits (all succeed or none) |
 | Glob | low | Find files by pattern |
-| Grep | low | Regex content search |
+| Grep | low | Regex content search with context lines |
 | LS | low | List directory contents with sizes |
+| **Web** | | |
 | WebFetch | medium | Fetch URL content (SSRF-protected) |
 | WebSearch | medium | Search the web |
+| RemoteTrigger | high | HTTP requests to webhooks/APIs |
+| **Tasks** | | |
 | TaskCreate | low | Create structured tasks |
 | TaskUpdate | low | Update task status |
 | TaskList | low | List all tasks |
+| TaskGet | low | Get task details |
+| TaskStop | low | Stop a running task |
+| TaskOutput | low | Get task output |
+| **Agents** | | |
+| Agent | medium | Spawn a sub-agent (with role specialization) |
+| ParallelAgent | medium | Dispatch multiple agents with DAG dependencies |
+| SendMessage | low | Agent-to-agent peer messaging |
 | AskUser | low | Ask user a question with options |
-| Skill | low | Invoke a skill from .oh/skills/ |
-| Agent | medium | Spawn a sub-agent for delegation |
+| **Scheduling** | | |
+| CronCreate | medium | Schedule recurring tasks |
+| CronDelete | medium | Remove scheduled tasks |
+| CronList | low | List all scheduled tasks |
+| **Planning** | | |
 | EnterPlanMode | low | Enter structured planning mode |
 | ExitPlanMode | low | Exit planning mode |
+| **Code Intelligence** | | |
+| Diagnostics | low | LSP-based code diagnostics |
 | NotebookEdit | medium | Edit Jupyter notebooks |
+| **Memory & Discovery** | | |
+| Memory | low | Save/list/search persistent memories |
+| Skill | low | Invoke a skill from .oh/skills/ |
+| ToolSearch | low | Find tools by description |
+| **Git Worktrees** | | |
+| EnterWorktree | medium | Create isolated git worktree |
+| ExitWorktree | medium | Remove a git worktree |
+| **Process** | | |
+| KillProcess | high | Stop processes by PID or name |
 
-Low-risk read-only tools auto-approve. Medium and high risk tools require confirmation in `ask` mode. Use `--trust` to skip all prompts.
+Low-risk read-only tools auto-approve. Medium and high risk tools require confirmation in `ask` mode. Use `--trust` or `--auto` to skip prompts.
 
-## Slash Commands (30+)
+## Slash Commands (33)
 
 Type these during a chat session. Aliases: `/q` exit, `/h` help, `/c` commit, `/m` model, `/s` status.
 
@@ -334,6 +391,58 @@ Every file modification is automatically checkpointed before execution. If somet
 
 Checkpoints are stored in `.oh/checkpoints/` and cover FileWrite, FileEdit, and Bash commands that modify files.
 
+## Verification Loops
+
+After every file edit (Edit, Write, MultiEdit), openHarness automatically runs language-appropriate lint/typecheck commands and feeds the results back into the agent context. This is the single highest-impact harness engineering pattern — research shows 2-3x quality improvement from automated feedback.
+
+**Auto-detection** — if your project has `tsconfig.json`, `.eslintrc*`, `pyproject.toml`, `go.mod`, or `Cargo.toml`, verification rules are detected automatically. No configuration needed.
+
+**Custom rules** via `.oh/config.yaml`:
+
+```yaml
+verification:
+  enabled: true       # default: true (auto-detect)
+  mode: warn          # 'warn' appends to output, 'block' marks as error
+  rules:
+    - extensions: [".ts", ".tsx"]
+      lint: "npx tsc --noEmit 2>&1 | head -20"
+      timeout: 15000
+    - extensions: [".py"]
+      lint: "ruff check {file} 2>&1 | head -10"
+```
+
+The agent sees `[Verification passed]` or `[Verification FAILED]` with the linter output after each edit, enabling self-correction.
+
+## Memory Consolidation
+
+On session exit, openHarness automatically prunes stale memories using temporal decay:
+
+- Memories not accessed in 30+ days lose 0.1 relevance per 30-day period
+- Memories below 0.1 relevance are permanently deleted
+- Updated relevance scores are persisted to memory files
+
+This keeps the memory system lean and relevant. Configure in `.oh/config.yaml`:
+
+```yaml
+memory:
+  consolidateOnExit: true   # default: true
+```
+
+## Scheduled Tasks (Cron)
+
+Create recurring tasks that run automatically in the background:
+
+```
+# Via slash commands
+/cron list                    # show all scheduled tasks
+/cron create "check-tests"    # create a new task (interactive)
+/cron delete <id>             # remove a task
+```
+
+**Schedule syntax:** `every 5m`, `every 2h`, `every 1d`
+
+The cron executor checks every 60 seconds for due tasks and runs them via sub-queries. Results are stored in `~/.oh/crons/history/`.
+
 ## Agent Roles
 
 Dispatch specialized sub-agents for focused tasks:
@@ -342,16 +451,22 @@ Dispatch specialized sub-agents for focused tasks:
 /roles            # list all available roles
 ```
 
-| Role | Description |
-|------|-------------|
-| `code-reviewer` | Find bugs, security issues, style problems |
-| `test-writer` | Generate unit and integration tests |
-| `docs-writer` | Write documentation and comments |
-| `debugger` | Systematic bug investigation |
-| `refactorer` | Simplify code without changing behavior |
-| `security-auditor` | OWASP, injection, secrets, CVE scanning |
+| Role | Description | Tools |
+|------|-------------|-------|
+| `code-reviewer` | Find bugs, security issues, style problems | Read-only |
+| `test-writer` | Generate unit and integration tests | Read + Write |
+| `docs-writer` | Write documentation and comments | Read + Write + Edit |
+| `debugger` | Systematic bug investigation | Read-only + Bash |
+| `refactorer` | Simplify code without changing behavior | All file tools + Bash |
+| `security-auditor` | OWASP, injection, secrets, CVE scanning | Read-only + Bash |
+| `evaluator` | Evaluate code quality and run tests (read-only) | Read-only + Bash + Diagnostics |
 
-The LLM can dispatch these via `Agent({ subagent_type: 'code-reviewer', prompt: '...' })`.
+Each role restricts the sub-agent to only its suggested tools. You can also pass `allowed_tools` explicitly:
+
+```
+Agent({ subagent_type: 'evaluator', prompt: 'Run all tests and report results' })
+Agent({ allowed_tools: ['Read', 'Grep'], prompt: 'Search for all TODO comments' })
+```
 
 ## Headless Mode
 
@@ -446,6 +561,31 @@ oh --model llamacpp/my-model
 oh models                    # list available models
 ```
 
+## Configuration Hierarchy
+
+Config is loaded in layers (later overrides earlier):
+
+1. **Global** `~/.oh/config.yaml` — default provider, model, theme for all projects
+2. **Project** `.oh/config.yaml` — project-specific settings
+3. **Local** `.oh/config.local.yaml` — personal overrides (gitignored)
+
+Set your default provider once globally:
+
+```yaml
+# ~/.oh/config.yaml
+provider: ollama
+model: llama3
+permissionMode: ask
+theme: dark
+```
+
+Then per-project configs only need what's different:
+
+```yaml
+# .oh/config.yaml
+model: codellama   # override just the model
+```
+
 ## Project Rules
 
 Create `.oh/RULES.md` in any repo (or run `oh init`):
@@ -458,6 +598,42 @@ Create `.oh/RULES.md` in any repo (or run `oh init`):
 
 Rules load automatically into every session.
 
+## How It Works
+
+```mermaid
+graph LR
+    User[User Input] --> REPL[REPL Loop]
+    REPL --> Query[Query Engine]
+    Query --> Provider[LLM Provider]
+    Provider --> LLM[Ollama / OpenAI / Anthropic]
+    LLM --> Tools[Tool Execution]
+    Tools --> Permissions{Permission Check}
+    Permissions -->|Approved| Execute[Run Tool]
+    Permissions -->|Blocked| Deny[Deny & Report]
+    Execute --> Response[Stream Response]
+    Response --> REPL
+```
+
+## FAQ
+
+**Does it work offline?**
+Yes. Use Ollama with a local model — no internet or API key needed.
+
+**How much does it cost?**
+Free. OpenHarness is MIT licensed. You bring your own API key (BYOK) for cloud models, or use Ollama for free.
+
+**Is it safe?**
+Yes. 7 permission modes control what tools can do. Bash commands are analyzed by an AST parser that blocks destructive patterns (`rm -rf`, `curl | bash`, etc.). Every file change is checkpointed and reversible with `/rewind`.
+
+**Can I use it in CI/CD?**
+Yes. Use `oh -p "prompt" --auto` for headless execution, or the built-in GitHub Action for PR reviews.
+
+**Does it support my language/framework?**
+Yes. OpenHarness is language-agnostic — it reads, writes, and executes code in any language. Syntax highlighting covers 20+ languages.
+
+**How does it compare to Claude Code?**
+~90% feature parity for CLI use cases. Main advantage: works with ANY LLM (not just Anthropic). See the [comparison table](#why-openharness) above.
+
 ## Install
 
 Requires **Node.js 18+**.

package/dist/DeferredTool.d.ts

@@ -0,0 +1,34 @@
+/**
+ * DeferredTool — lazy-loads tool schemas to reduce system prompt size.
+ *
+ * Wraps a built-in tool with a minimal prompt (name + description only).
+ * Full schema and prompt are loaded on first invocation or when resolved
+ * via ToolSearch. This mirrors the DeferredMcpTool pattern.
+ *
+ * Token savings: deferred tools contribute ~15 tokens to the system prompt
+ * instead of ~150, reducing context pressure by ~90% per deferred tool.
+ */
+import { z } from "zod";
+import type { Tool, ToolContext, ToolResult } from "./Tool.js";
+import type { RiskLevel } from "./types/permissions.js";
+export declare class DeferredTool implements Tool<z.ZodType> {
+    readonly name: string;
+    readonly description: string;
+    readonly inputSchema: z.ZodType;
+    readonly riskLevel: RiskLevel;
+    private inner;
+    private _activated;
+    constructor(tool: Tool);
+    /** Whether this tool has been activated (called or resolved) */
+    get activated(): boolean;
+    isReadOnly(input: unknown): boolean;
+    isConcurrencySafe(input: unknown): boolean;
+    call(input: any, context: ToolContext): Promise<ToolResult>;
+    /** Minimal prompt when deferred, full prompt when activated */
+    prompt(): string;
+    /** Get the full inner tool (for ToolSearch resolution) */
+    getInner(): Tool;
+    /** Activate this tool so it returns full prompt on next call to prompt() */
+    activate(): void;
+}
+//# sourceMappingURL=DeferredTool.d.ts.map

package/dist/DeferredTool.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"DeferredTool.d.ts","sourceRoot":"","sources":["../src/DeferredTool.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AAEH,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AACxB,OAAO,KAAK,EAAE,IAAI,EAAE,WAAW,EAAE,UAAU,EAAE,MAAM,WAAW,CAAC;AAC/D,OAAO,KAAK,EAAE,SAAS,EAAE,MAAM,wBAAwB,CAAC;AAExD,qBAAa,YAAa,YAAW,IAAI,CAAC,CAAC,CAAC,OAAO,CAAC;IAClD,QAAQ,CAAC,IAAI,EAAE,MAAM,CAAC;IACtB,QAAQ,CAAC,WAAW,EAAE,MAAM,CAAC;IAC7B,QAAQ,CAAC,WAAW,EAAE,CAAC,CAAC,OAAO,CAAC;IAChC,QAAQ,CAAC,SAAS,EAAE,SAAS,CAAC;IAE9B,OAAO,CAAC,KAAK,CAAO;IACpB,OAAO,CAAC,UAAU,CAAS;gBAEf,IAAI,EAAE,IAAI;IAStB,gEAAgE;IAChE,IAAI,SAAS,IAAI,OAAO,CAA4B;IAEpD,UAAU,CAAC,KAAK,EAAE,OAAO,GAAG,OAAO;IAInC,iBAAiB,CAAC,KAAK,EAAE,OAAO,GAAG,OAAO;IAIpC,IAAI,CAAC,KAAK,EAAE,GAAG,EAAE,OAAO,EAAE,WAAW,GAAG,OAAO,CAAC,UAAU,CAAC;IAajE,+DAA+D;IAC/D,MAAM,IAAI,MAAM;IAKhB,0DAA0D;IAC1D,QAAQ,IAAI,IAAI;IAIhB,4EAA4E;IAC5E,QAAQ,IAAI,IAAI;CAGjB"}

package/dist/DeferredTool.js

@@ -0,0 +1,62 @@
+/**
+ * DeferredTool — lazy-loads tool schemas to reduce system prompt size.
+ *
+ * Wraps a built-in tool with a minimal prompt (name + description only).
+ * Full schema and prompt are loaded on first invocation or when resolved
+ * via ToolSearch. This mirrors the DeferredMcpTool pattern.
+ *
+ * Token savings: deferred tools contribute ~15 tokens to the system prompt
+ * instead of ~150, reducing context pressure by ~90% per deferred tool.
+ */
+import { z } from "zod";
+export class DeferredTool {
+    name;
+    description;
+    inputSchema;
+    riskLevel;
+    inner;
+    _activated = false;
+    constructor(tool) {
+        this.inner = tool;
+        this.name = tool.name;
+        this.description = tool.description;
+        this.riskLevel = tool.riskLevel;
+        // Permissive schema until activated — accepts any object
+        this.inputSchema = z.record(z.unknown());
+    }
+    /** Whether this tool has been activated (called or resolved) */
+    get activated() { return this._activated; }
+    isReadOnly(input) {
+        return this.inner.isReadOnly(input);
+    }
+    isConcurrencySafe(input) {
+        return this.inner.isConcurrencySafe(input);
+    }
+    async call(input, context) {
+        this._activated = true;
+        // Validate with the real schema
+        const parsed = this.inner.inputSchema.safeParse(input);
+        if (!parsed.success) {
+            return {
+                output: `Schema validation error: ${parsed.error.message}`,
+                isError: true,
+            };
+        }
+        return this.inner.call(parsed.data, context);
+    }
+    /** Minimal prompt when deferred, full prompt when activated */
+    prompt() {
+        if (this._activated)
+            return this.inner.prompt();
+        return `[deferred] ${this.name}: ${this.description}`;
+    }
+    /** Get the full inner tool (for ToolSearch resolution) */
+    getInner() {
+        return this.inner;
+    }
+    /** Activate this tool so it returns full prompt on next call to prompt() */
+    activate() {
+        this._activated = true;
+    }
+}
+//# sourceMappingURL=DeferredTool.js.map

package/dist/DeferredTool.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"DeferredTool.js","sourceRoot":"","sources":["../src/DeferredTool.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AAEH,OAAO,EAAE,CAAC,EAAE,MAAM,KAAK,CAAC;AAIxB,MAAM,OAAO,YAAY;IACd,IAAI,CAAS;IACb,WAAW,CAAS;IACpB,WAAW,CAAY;IACvB,SAAS,CAAY;IAEtB,KAAK,CAAO;IACZ,UAAU,GAAG,KAAK,CAAC;IAE3B,YAAY,IAAU;QACpB,IAAI,CAAC,KAAK,GAAG,IAAI,CAAC;QAClB,IAAI,CAAC,IAAI,GAAG,IAAI,CAAC,IAAI,CAAC;QACtB,IAAI,CAAC,WAAW,GAAG,IAAI,CAAC,WAAW,CAAC;QACpC,IAAI,CAAC,SAAS,GAAG,IAAI,CAAC,SAAS,CAAC;QAChC,yDAAyD;QACzD,IAAI,CAAC,WAAW,GAAG,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,OAAO,EAAE,CAAC,CAAC;IAC3C,CAAC;IAED,gEAAgE;IAChE,IAAI,SAAS,KAAc,OAAO,IAAI,CAAC,UAAU,CAAC,CAAC,CAAC;IAEpD,UAAU,CAAC,KAAc;QACvB,OAAO,IAAI,CAAC,KAAK,CAAC,UAAU,CAAC,KAAK,CAAC,CAAC;IACtC,CAAC;IAED,iBAAiB,CAAC,KAAc;QAC9B,OAAO,IAAI,CAAC,KAAK,CAAC,iBAAiB,CAAC,KAAK,CAAC,CAAC;IAC7C,CAAC;IAED,KAAK,CAAC,IAAI,CAAC,KAAU,EAAE,OAAoB;QACzC,IAAI,CAAC,UAAU,GAAG,IAAI,CAAC;QACvB,gCAAgC;QAChC,MAAM,MAAM,GAAG,IAAI,CAAC,KAAK,CAAC,WAAW,CAAC,SAAS,CAAC,KAAK,CAAC,CAAC;QACvD,IAAI,CAAC,MAAM,CAAC,OAAO,EAAE,CAAC;YACpB,OAAO;gBACL,MAAM,EAAE,4BAA4B,MAAM,CAAC,KAAK,CAAC,OAAO,EAAE;gBAC1D,OAAO,EAAE,IAAI;aACd,CAAC;QACJ,CAAC;QACD,OAAO,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,EAAE,OAAO,CAAC,CAAC;IAC/C,CAAC;IAED,+DAA+D;IAC/D,MAAM;QACJ,IAAI,IAAI,CAAC,UAAU;YAAE,OAAO,IAAI,CAAC,KAAK,CAAC,MAAM,EAAE,CAAC;QAChD,OAAO,cAAc,IAAI,CAAC,IAAI,KAAK,IAAI,CAAC,WAAW,EAAE,CAAC;IACxD,CAAC;IAED,0DAA0D;IAC1D,QAAQ;QACN,OAAO,IAAI,CAAC,KAAK,CAAC;IACpB,CAAC;IAED,4EAA4E;IAC5E,QAAQ;QACN,IAAI,CAAC,UAAU,GAAG,IAAI,CAAC;IACzB,CAAC;CACF"}

package/dist/DeferredTool.test.d.ts

@@ -0,0 +1,2 @@
+export {};
+//# sourceMappingURL=DeferredTool.test.d.ts.map

package/dist/DeferredTool.test.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"DeferredTool.test.d.ts","sourceRoot":"","sources":["../src/DeferredTool.test.ts"],"names":[],"mappings":""}

package/dist/DeferredTool.test.js

@@ -0,0 +1,79 @@
+import { describe, it } from 'node:test';
+import assert from 'node:assert/strict';
+import { createMockTool } from './test-helpers.js';
+import { DeferredTool } from './DeferredTool.js';
+describe('DeferredTool', () => {
+    it('wraps a tool with correct name and description', () => {
+        const inner = createMockTool('TestTool');
+        const deferred = new DeferredTool(inner);
+        assert.equal(deferred.name, 'TestTool');
+        assert.equal(deferred.description, inner.description);
+        assert.equal(deferred.riskLevel, inner.riskLevel);
+    });
+    it('returns deferred prompt before activation', () => {
+        const inner = createMockTool('TestTool');
+        const deferred = new DeferredTool(inner);
+        const prompt = deferred.prompt();
+        assert.ok(prompt.startsWith('[deferred]'));
+        assert.ok(prompt.includes('TestTool'));
+    });
+    it('returns full prompt after activation', () => {
+        const inner = createMockTool('TestTool');
+        const deferred = new DeferredTool(inner);
+        deferred.activate();
+        assert.equal(deferred.prompt(), inner.prompt());
+    });
+    it('is not activated by default', () => {
+        const inner = createMockTool('TestTool');
+        const deferred = new DeferredTool(inner);
+        assert.equal(deferred.activated, false);
+    });
+    it('activates on call()', async () => {
+        const inner = createMockTool('TestTool');
+        const deferred = new DeferredTool(inner);
+        assert.equal(deferred.activated, false);
+        const result = await deferred.call({ input: 'test' }, { workingDir: '/tmp' });
+        assert.equal(deferred.activated, true);
+        assert.equal(result.isError, false);
+        assert.ok(result.output.includes('TestTool'));
+    });
+    it('activates via activate() method', () => {
+        const inner = createMockTool('TestTool');
+        const deferred = new DeferredTool(inner);
+        deferred.activate();
+        assert.equal(deferred.activated, true);
+    });
+    it('returns full prompt after call()', async () => {
+        const inner = createMockTool('TestTool');
+        const deferred = new DeferredTool(inner);
+        await deferred.call({ input: 'test' }, { workingDir: '/tmp' });
+        assert.equal(deferred.prompt(), inner.prompt());
+    });
+    it('validates input against inner schema and returns error', async () => {
+        const inner = createMockTool('TestTool');
+        const deferred = new DeferredTool(inner);
+        // Inner schema expects { input?: string }, pass something wildly wrong
+        // Since z.object({ input: z.string().optional() }) is permissive,
+        // we pass a valid object and check it works
+        const result = await deferred.call({ input: 'valid' }, { workingDir: '/tmp' });
+        assert.equal(result.isError, false);
+    });
+    it('exposes inner tool via getInner()', () => {
+        const inner = createMockTool('TestTool');
+        const deferred = new DeferredTool(inner);
+        assert.equal(deferred.getInner(), inner);
+    });
+    it('delegates isReadOnly to inner tool', () => {
+        const readOnly = createMockTool('ReadOnly', { readOnly: true });
+        const writable = createMockTool('Writable', { readOnly: false });
+        assert.equal(new DeferredTool(readOnly).isReadOnly({}), true);
+        assert.equal(new DeferredTool(writable).isReadOnly({}), false);
+    });
+    it('delegates isConcurrencySafe to inner tool', () => {
+        const safe = createMockTool('Safe', { concurrent: true });
+        const unsafe = createMockTool('Unsafe', { concurrent: false });
+        assert.equal(new DeferredTool(safe).isConcurrencySafe({}), true);
+        assert.equal(new DeferredTool(unsafe).isConcurrencySafe({}), false);
+    });
+});
+//# sourceMappingURL=DeferredTool.test.js.map

package/dist/DeferredTool.test.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"DeferredTool.test.js","sourceRoot":"","sources":["../src/DeferredTool.test.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,QAAQ,EAAE,EAAE,EAAE,MAAM,WAAW,CAAC;AACzC,OAAO,MAAM,MAAM,oBAAoB,CAAC;AACxC,OAAO,EAAE,cAAc,EAAE,MAAM,mBAAmB,CAAC;AACnD,OAAO,EAAE,YAAY,EAAE,MAAM,mBAAmB,CAAC;AAEjD,QAAQ,CAAC,cAAc,EAAE,GAAG,EAAE;IAC5B,EAAE,CAAC,gDAAgD,EAAE,GAAG,EAAE;QACxD,MAAM,KAAK,GAAG,cAAc,CAAC,UAAU,CAAC,CAAC;QACzC,MAAM,QAAQ,GAAG,IAAI,YAAY,CAAC,KAAK,CAAC,CAAC;QACzC,MAAM,CAAC,KAAK,CAAC,QAAQ,CAAC,IAAI,EAAE,UAAU,CAAC,CAAC;QACxC,MAAM,CAAC,KAAK,CAAC,QAAQ,CAAC,WAAW,EAAE,KAAK,CAAC,WAAW,CAAC,CAAC;QACtD,MAAM,CAAC,KAAK,CAAC,QAAQ,CAAC,SAAS,EAAE,KAAK,CAAC,SAAS,CAAC,CAAC;IACpD,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,2CAA2C,EAAE,GAAG,EAAE;QACnD,MAAM,KAAK,GAAG,cAAc,CAAC,UAAU,CAAC,CAAC;QACzC,MAAM,QAAQ,GAAG,IAAI,YAAY,CAAC,KAAK,CAAC,CAAC;QACzC,MAAM,MAAM,GAAG,QAAQ,CAAC,MAAM,EAAE,CAAC;QACjC,MAAM,CAAC,EAAE,CAAC,MAAM,CAAC,UAAU,CAAC,YAAY,CAAC,CAAC,CAAC;QAC3C,MAAM,CAAC,EAAE,CAAC,MAAM,CAAC,QAAQ,CAAC,UAAU,CAAC,CAAC,CAAC;IACzC,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,sCAAsC,EAAE,GAAG,EAAE;QAC9C,MAAM,KAAK,GAAG,cAAc,CAAC,UAAU,CAAC,CAAC;QACzC,MAAM,QAAQ,GAAG,IAAI,YAAY,CAAC,KAAK,CAAC,CAAC;QACzC,QAAQ,CAAC,QAAQ,EAAE,CAAC;QACpB,MAAM,CAAC,KAAK,CAAC,QAAQ,CAAC,MAAM,EAAE,EAAE,KAAK,CAAC,MAAM,EAAE,CAAC,CAAC;IAClD,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,6BAA6B,EAAE,GAAG,EAAE;QACrC,MAAM,KAAK,GAAG,cAAc,CAAC,UAAU,CAAC,CAAC;QACzC,MAAM,QAAQ,GAAG,IAAI,YAAY,CAAC,KAAK,CAAC,CAAC;QACzC,MAAM,CAAC,KAAK,CAAC,QAAQ,CAAC,SAAS,EAAE,KAAK,CAAC,CAAC;IAC1C,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,qBAAqB,EAAE,KAAK,IAAI,EAAE;QACnC,MAAM,KAAK,GAAG,cAAc,CAAC,UAAU,CAAC,CAAC;QACzC,MAAM,QAAQ,GAAG,IAAI,YAAY,CAAC,KAAK,CAAC,CAAC;QACzC,MAAM,CAAC,KAAK,CAAC,QAAQ,CAAC,SAAS,EAAE,KAAK,CAAC,CAAC;QAExC,MAAM,MAAM,GAAG,MAAM,QAAQ,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,MAAM,EAAE,EAAE,EAAE,UAAU,EAAE,MAAM,EAAE,CAAC,CAAC;QAC9E,MAAM,CAAC,KAAK,CAAC,QAAQ,CAAC,SAAS,EAAE,IAAI,CAAC,CAAC;QACvC,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,OAAO,EAAE,KAAK,CAAC,CAAC;QACpC,MAAM,CAAC,EAAE,CAAC,MAAM,CAAC,MAAM,CAAC,QAAQ,CAAC,UAAU,CAAC,CAAC,CAAC;IAChD,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,iCAAiC,EAAE,GAAG,EAAE;QACzC,MAAM,KAAK,GAAG,cAAc,CAAC,UAAU,CAAC,CAAC;QACzC,MAAM,QAAQ,GAAG,IAAI,YAAY,CAAC,KAAK,CAAC,CAAC;QACzC,QAAQ,CAAC,QAAQ,EAAE,CAAC;QACpB,MAAM,CAAC,KAAK,CAAC,QAAQ,CAAC,SAAS,EAAE,IAAI,CAAC,CAAC;IACzC,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,kCAAkC,EAAE,KAAK,IAAI,EAAE;QAChD,MAAM,KAAK,GAAG,cAAc,CAAC,UAAU,CAAC,CAAC;QACzC,MAAM,QAAQ,GAAG,IAAI,YAAY,CAAC,KAAK,CAAC,CAAC;QACzC,MAAM,QAAQ,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,MAAM,EAAE,EAAE,EAAE,UAAU,EAAE,MAAM,EAAE,CAAC,CAAC;QAC/D,MAAM,CAAC,KAAK,CAAC,QAAQ,CAAC,MAAM,EAAE,EAAE,KAAK,CAAC,MAAM,EAAE,CAAC,CAAC;IAClD,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,wDAAwD,EAAE,KAAK,IAAI,EAAE;QACtE,MAAM,KAAK,GAAG,cAAc,CAAC,UAAU,CAAC,CAAC;QACzC,MAAM,QAAQ,GAAG,IAAI,YAAY,CAAC,KAAK,CAAC,CAAC;QACzC,uEAAuE;QACvE,kEAAkE;QAClE,4CAA4C;QAC5C,MAAM,MAAM,GAAG,MAAM,QAAQ,CAAC,IAAI,CAAC,EAAE,KAAK,EAAE,OAAO,EAAE,EAAE,EAAE,UAAU,EAAE,MAAM,EAAE,CAAC,CAAC;QAC/E,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,OAAO,EAAE,KAAK,CAAC,CAAC;IACtC,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,mCAAmC,EAAE,GAAG,EAAE;QAC3C,MAAM,KAAK,GAAG,cAAc,CAAC,UAAU,CAAC,CAAC;QACzC,MAAM,QAAQ,GAAG,IAAI,YAAY,CAAC,KAAK,CAAC,CAAC;QACzC,MAAM,CAAC,KAAK,CAAC,QAAQ,CAAC,QAAQ,EAAE,EAAE,KAAK,CAAC,CAAC;IAC3C,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,oCAAoC,EAAE,GAAG,EAAE;QAC5C,MAAM,QAAQ,GAAG,cAAc,CAAC,UAAU,EAAE,EAAE,QAAQ,EAAE,IAAI,EAAE,CAAC,CAAC;QAChE,MAAM,QAAQ,GAAG,cAAc,CAAC,UAAU,EAAE,EAAE,QAAQ,EAAE,KAAK,EAAE,CAAC,CAAC;QACjE,MAAM,CAAC,KAAK,CAAC,IAAI,YAAY,CAAC,QAAQ,CAAC,CAAC,UAAU,CAAC,EAAE,CAAC,EAAE,IAAI,CAAC,CAAC;QAC9D,MAAM,CAAC,KAAK,CAAC,IAAI,YAAY,CAAC,QAAQ,CAAC,CAAC,UAAU,CAAC,EAAE,CAAC,EAAE,KAAK,CAAC,CAAC;IACjE,CAAC,CAAC,CAAC;IAEH,EAAE,CAAC,2CAA2C,EAAE,GAAG,EAAE;QACnD,MAAM,IAAI,GAAG,cAAc,CAAC,MAAM,EAAE,EAAE,UAAU,EAAE,IAAI,EAAE,CAAC,CAAC;QAC1D,MAAM,MAAM,GAAG,cAAc,CAAC,QAAQ,EAAE,EAAE,UAAU,EAAE,KAAK,EAAE,CAAC,CAAC;QAC/D,MAAM,CAAC,KAAK,CAAC,IAAI,YAAY,CAAC,IAAI,CAAC,CAAC,iBAAiB,CAAC,EAAE,CAAC,EAAE,IAAI,CAAC,CAAC;QACjE,MAAM,CAAC,KAAK,CAAC,IAAI,YAAY,CAAC,MAAM,CAAC,CAAC,iBAAiB,CAAC,EAAE,CAAC,EAAE,KAAK,CAAC,CAAC;IACtE,CAAC,CAAC,CAAC;AACL,CAAC,CAAC,CAAC"}

package/dist/agents/roles.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"roles.d.ts","sourceRoot":"","sources":["../../src/agents/roles.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AAEH,MAAM,MAAM,SAAS,GAAG;IACtB,EAAE,EAAE,MAAM,CAAC;IACX,IAAI,EAAE,MAAM,CAAC;IACb,WAAW,EAAE,MAAM,CAAC;IACpB,sBAAsB,EAAE,MAAM,CAAC;IAC/B,qDAAqD;IACrD,cAAc,CAAC,EAAE,MAAM,EAAE,CAAC;CAC3B,CAAC;AAgGF,uBAAuB;AACvB,wBAAgB,OAAO,CAAC,EAAE,EAAE,MAAM,GAAG,SAAS,GAAG,SAAS,CAEzD;AAED,+BAA+B;AAC/B,wBAAgB,SAAS,IAAI,SAAS,EAAE,CAEvC;AAED,mBAAmB;AACnB,wBAAgB,UAAU,IAAI,MAAM,EAAE,CAErC"}
+{"version":3,"file":"roles.d.ts","sourceRoot":"","sources":["../../src/agents/roles.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AAEH,MAAM,MAAM,SAAS,GAAG;IACtB,EAAE,EAAE,MAAM,CAAC;IACX,IAAI,EAAE,MAAM,CAAC;IACb,WAAW,EAAE,MAAM,CAAC;IACpB,sBAAsB,EAAE,MAAM,CAAC;IAC/B,qDAAqD;IACrD,cAAc,CAAC,EAAE,MAAM,EAAE,CAAC;CAC3B,CAAC;AA8GF,uBAAuB;AACvB,wBAAgB,OAAO,CAAC,EAAE,EAAE,MAAM,GAAG,SAAS,GAAG,SAAS,CAEzD;AAED,+BAA+B;AAC/B,wBAAgB,SAAS,IAAI,SAAS,EAAE,CAEvC;AAED,mBAAmB;AACnB,wBAAgB,UAAU,IAAI,MAAM,EAAE,CAErC"}

package/dist/agents/roles.js

@@ -21,7 +21,7 @@ const roles = [
 - Verify that changes match the stated intent
 
 Be specific: cite file paths, line numbers, and code snippets. Prioritize issues by severity (critical > major > minor). Don't mention things that look fine — focus on problems.`,
-        suggestedTools: ['FileRead', 'Glob', 'Grep', 'LS'],
+        suggestedTools: ['Read', 'Glob', 'Grep', 'LS'],
     },
     {
         id: 'test-writer',
@@ -36,7 +36,7 @@ Be specific: cite file paths, line numbers, and code snippets. Prioritize issues
 - Include both positive and negative test cases
 
 Read existing tests first to match the style, then write new tests.`,
-        suggestedTools: ['FileRead', 'FileWrite', 'Glob', 'Grep', 'Bash'],
+        suggestedTools: ['Read', 'Write', 'Glob', 'Grep', 'Bash'],
     },
     {
         id: 'docs-writer',
@@ -51,7 +51,7 @@ Read existing tests first to match the style, then write new tests.`,
 - Keep documentation in sync with the actual code
 
 Write for the target audience (developers using this project). Be practical, not verbose.`,
-        suggestedTools: ['FileRead', 'FileWrite', 'FileEdit', 'Glob', 'Grep'],
+        suggestedTools: ['Read', 'Write', 'Edit', 'Glob', 'Grep'],
     },
     {
         id: 'debugger',
@@ -66,7 +66,7 @@ Write for the target audience (developers using this project). Be practical, not
 - Propose a minimal fix that addresses the root cause, not the symptom
 
 Follow systematic debugging: read errors → reproduce → check changes → trace data → form hypothesis → test minimally.`,
-        suggestedTools: ['FileRead', 'Glob', 'Grep', 'Bash', 'LS'],
+        suggestedTools: ['Read', 'Glob', 'Grep', 'Bash', 'LS'],
     },
     {
         id: 'refactorer',
@@ -81,7 +81,7 @@ Follow systematic debugging: read errors → reproduce → check changes → tra
 - Ensure all existing tests still pass after refactoring
 
 Do NOT add new features or change behavior. The refactored code must be functionally identical. Run tests after each change.`,
-        suggestedTools: ['FileRead', 'FileWrite', 'FileEdit', 'Glob', 'Grep', 'Bash'],
+        suggestedTools: ['Read', 'Write', 'Edit', 'Glob', 'Grep', 'Bash'],
     },
     {
         id: 'security-auditor',
@@ -98,7 +98,21 @@ Do NOT add new features or change behavior. The refactored code must be function
 - Check dependency versions for known CVEs
 
 Report findings with severity (Critical/High/Medium/Low), affected file:line, and recommended fix.`,
-        suggestedTools: ['FileRead', 'Glob', 'Grep', 'Bash'],
+        suggestedTools: ['Read', 'Glob', 'Grep', 'Bash'],
+    },
+    {
+        id: 'evaluator',
+        name: 'Evaluator',
+        description: 'Evaluates code quality, correctness, and test results (read-only)',
+        systemPromptSupplement: `You are an evaluator agent. Your job is to:
+- Review code changes for correctness and quality
+- Run existing tests and report results
+- Check for regressions against the stated requirements
+- Verify that changes match the stated intent
+- Provide a pass/fail assessment with specific findings
+
+You CANNOT modify files. Only read, search, and run test/lint commands to evaluate.`,
+        suggestedTools: ['Read', 'Glob', 'Grep', 'LS', 'Bash', 'Diagnostics'],
     },
 ];
 /** Get a role by ID */

package/dist/agents/roles.js.map

@@ -1 +1 @@
-{"version":3,"file":"roles.js","sourceRoot":"","sources":["../../src/agents/roles.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AAWH,MAAM,KAAK,GAAgB;IACzB;QACE,EAAE,EAAE,eAAe;QACnB,IAAI,EAAE,eAAe;QACrB,WAAW,EAAE,gEAAgE;QAC7E,sBAAsB,EAAE;;;;;;;kLAOsJ;QAC9K,cAAc,EAAE,CAAC,UAAU,EAAE,MAAM,EAAE,MAAM,EAAE,IAAI,CAAC;KACnD;IACD;QACE,EAAE,EAAE,aAAa;QACjB,IAAI,EAAE,aAAa;QACnB,WAAW,EAAE,2DAA2D;QACxE,sBAAsB,EAAE;;;;;;;;oEAQwC;QAChE,cAAc,EAAE,CAAC,UAAU,EAAE,WAAW,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,CAAC;KAClE;IACD;QACE,EAAE,EAAE,aAAa;QACjB,IAAI,EAAE,sBAAsB;QAC5B,WAAW,EAAE,gEAAgE;QAC7E,sBAAsB,EAAE;;;;;;;;0FAQ8D;QACtF,cAAc,EAAE,CAAC,UAAU,EAAE,WAAW,EAAE,UAAU,EAAE,MAAM,EAAE,MAAM,CAAC;KACtE;IACD;QACE,EAAE,EAAE,UAAU;QACd,IAAI,EAAE,UAAU;QAChB,WAAW,EAAE,uEAAuE;QACpF,sBAAsB,EAAE;;;;;;;;sHAQ0F;QAClH,cAAc,EAAE,CAAC,UAAU,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,IAAI,CAAC;KAC3D;IACD;QACE,EAAE,EAAE,YAAY;QAChB,IAAI,EAAE,YAAY;QAClB,WAAW,EAAE,4DAA4D;QACzE,sBAAsB,EAAE;;;;;;;;6HAQiG;QACzH,cAAc,EAAE,CAAC,UAAU,EAAE,WAAW,EAAE,UAAU,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,CAAC;KAC9E;IACD;QACE,EAAE,EAAE,kBAAkB;QACtB,IAAI,EAAE,kBAAkB;QACxB,WAAW,EAAE,gEAAgE;QAC7E,sBAAsB,EAAE;;;;;;;;;;mGAUuE;QAC/F,cAAc,EAAE,CAAC,UAAU,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,CAAC;KACrD;CACF,CAAC;AAEF,uBAAuB;AACvB,MAAM,UAAU,OAAO,CAAC,EAAU;IAChC,OAAO,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,CAAC,CAAC;AACtC,CAAC;AAED,+BAA+B;AAC/B,MAAM,UAAU,SAAS;IACvB,OAAO,CAAC,GAAG,KAAK,CAAC,CAAC;AACpB,CAAC;AAED,mBAAmB;AACnB,MAAM,UAAU,UAAU;IACxB,OAAO,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC;AAC9B,CAAC"}
+{"version":3,"file":"roles.js","sourceRoot":"","sources":["../../src/agents/roles.ts"],"names":[],"mappings":"AAAA;;;;;;;;;GASG;AAWH,MAAM,KAAK,GAAgB;IACzB;QACE,EAAE,EAAE,eAAe;QACnB,IAAI,EAAE,eAAe;QACrB,WAAW,EAAE,gEAAgE;QAC7E,sBAAsB,EAAE;;;;;;;kLAOsJ;QAC9K,cAAc,EAAE,CAAC,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,IAAI,CAAC;KAC/C;IACD;QACE,EAAE,EAAE,aAAa;QACjB,IAAI,EAAE,aAAa;QACnB,WAAW,EAAE,2DAA2D;QACxE,sBAAsB,EAAE;;;;;;;;oEAQwC;QAChE,cAAc,EAAE,CAAC,MAAM,EAAE,OAAO,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,CAAC;KAC1D;IACD;QACE,EAAE,EAAE,aAAa;QACjB,IAAI,EAAE,sBAAsB;QAC5B,WAAW,EAAE,gEAAgE;QAC7E,sBAAsB,EAAE;;;;;;;;0FAQ8D;QACtF,cAAc,EAAE,CAAC,MAAM,EAAE,OAAO,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,CAAC;KAC1D;IACD;QACE,EAAE,EAAE,UAAU;QACd,IAAI,EAAE,UAAU;QAChB,WAAW,EAAE,uEAAuE;QACpF,sBAAsB,EAAE;;;;;;;;sHAQ0F;QAClH,cAAc,EAAE,CAAC,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,IAAI,CAAC;KACvD;IACD;QACE,EAAE,EAAE,YAAY;QAChB,IAAI,EAAE,YAAY;QAClB,WAAW,EAAE,4DAA4D;QACzE,sBAAsB,EAAE;;;;;;;;6HAQiG;QACzH,cAAc,EAAE,CAAC,MAAM,EAAE,OAAO,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,CAAC;KAClE;IACD;QACE,EAAE,EAAE,kBAAkB;QACtB,IAAI,EAAE,kBAAkB;QACxB,WAAW,EAAE,gEAAgE;QAC7E,sBAAsB,EAAE;;;;;;;;;;mGAUuE;QAC/F,cAAc,EAAE,CAAC,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,CAAC;KACjD;IACD;QACE,EAAE,EAAE,WAAW;QACf,IAAI,EAAE,WAAW;QACjB,WAAW,EAAE,mEAAmE;QAChF,sBAAsB,EAAE;;;;;;;oFAOwD;QAChF,cAAc,EAAE,CAAC,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,EAAE,aAAa,CAAC;KACtE;CACF,CAAC;AAEF,uBAAuB;AACvB,MAAM,UAAU,OAAO,CAAC,EAAU;IAChC,OAAO,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,KAAK,EAAE,CAAC,CAAC;AACtC,CAAC;AAED,+BAA+B;AAC/B,MAAM,UAAU,SAAS;IACvB,OAAO,CAAC,GAAG,KAAK,CAAC,CAAC;AACpB,CAAC;AAED,mBAAmB;AACnB,MAAM,UAAU,UAAU;IACxB,OAAO,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC;AAC9B,CAAC"}

package/dist/agents/roles.test.js

@@ -4,18 +4,19 @@ import { getRole, listRoles, getRoleIds } from './roles.js';
 describe('agent roles', () => {
     it('lists all roles', () => {
         const roles = listRoles();
-        assert.ok(roles.length >= 6);
+        assert.ok(roles.length >= 7);
         assert.ok(roles.find(r => r.id === 'code-reviewer'));
         assert.ok(roles.find(r => r.id === 'test-writer'));
         assert.ok(roles.find(r => r.id === 'debugger'));
         assert.ok(roles.find(r => r.id === 'security-auditor'));
+        assert.ok(roles.find(r => r.id === 'evaluator'));
     });
     it('gets role by ID', () => {
         const role = getRole('code-reviewer');
         assert.ok(role);
         assert.strictEqual(role.name, 'Code Reviewer');
         assert.ok(role.systemPromptSupplement.length > 50);
-        assert.ok(role.suggestedTools.includes('FileRead'));
+        assert.ok(role.suggestedTools.includes('Read'));
     });
     it('returns undefined for unknown role', () => {
         assert.strictEqual(getRole('nonexistent'), undefined);
@@ -34,5 +35,28 @@ describe('agent roles', () => {
             assert.ok(role.systemPromptSupplement.length > 20, `role ${role.id} has short prompt`);
         }
     });
+    it('evaluator role has read-only suggested tools', () => {
+        const role = getRole('evaluator');
+        assert.ok(role);
+        assert.strictEqual(role.name, 'Evaluator');
+        assert.ok(role.suggestedTools.includes('Read'));
+        assert.ok(role.suggestedTools.includes('Glob'));
+        assert.ok(role.suggestedTools.includes('Grep'));
+        assert.ok(role.suggestedTools.includes('Bash'));
+        assert.ok(role.suggestedTools.includes('Diagnostics'));
+        // Evaluator should NOT have write tools
+        assert.ok(!role.suggestedTools.includes('Write'));
+        assert.ok(!role.suggestedTools.includes('Edit'));
+    });
+    it('all roles use actual tool names (not FileRead/FileWrite/FileEdit)', () => {
+        const invalidNames = ['FileRead', 'FileWrite', 'FileEdit'];
+        for (const role of listRoles()) {
+            if (!role.suggestedTools)
+                continue;
+            for (const toolName of role.suggestedTools) {
+                assert.ok(!invalidNames.includes(toolName), `role ${role.id} uses invalid tool name '${toolName}' — should be 'Read', 'Write', or 'Edit'`);
+            }
+        }
+    });
 });
 //# sourceMappingURL=roles.test.js.map