beast-agent 2.15.0 → 2.16.0

package/src/agent/opencode/systemprompt.js

@@ -0,0 +1,157 @@
+'use strict';
+
+/* ---------- opencode session/system.ts + reminders.ts portu (birebir) ----------
+   Beast Code (bcCode) oturumlarında system prompt artık TEK elle yazılmış
+   metin değil, opencode birleşimidir (prompt.ts:1257-1269):
+
+     system = [ ...environment(model),            → bu modül
+                ...instructions(AGENTS.md),        → engine._projectInstructions
+                ...(skills kataloğu),             → bu modül (skill/skill.ts fmt)
+                ...ajan tanımının promptu ]       → agents.js (explore vb.)
+
+   Ana prompt MODEL BAZLI seçilir (system.ts:27-49): gpt-4 ailesi, o1, o3 → beast.txt,
+   gpt* → gpt.txt (codex* → codex.txt), gemini-* → gemini.txt, claude →
+   anthropic.txt, kimi → kimi.txt, muse → meta.txt, trinity → trinity.txt,
+   diğer → default.txt.
+
+   Reminders (reminders.ts:15-90): plan ajanındaysa son user mesajına plan.txt
+   synthetic part'ı; geçmişte plan koşmuşsa ve şimdi build'se build-switch.txt
+   eklenir. Belleği kirletmemek için payload ÖNCESİ kopya mesajlara uygulanır
+   (her tur taze — opencode her turda DB'den yeniden yükleyip uygular). */
+
+const fs = require('fs');
+const path = require('path');
+const Permission = require('./permission');
+
+const PROMPT_DIR = path.join(__dirname, 'prompts');
+function loadPrompt(name) {
+  try {
+    let t = fs.readFileSync(path.join(PROMPT_DIR, name), 'utf8');
+    if (t.charCodeAt(0) === 0xfeff) t = t.slice(1); // BOM-strip
+    return t.replace(/\s+$/, '');
+  } catch {
+    return '';
+  }
+}
+
+const PROMPTS = {
+  beast: loadPrompt('beast.txt'),
+  gpt: loadPrompt('gpt.txt'),
+  codex: loadPrompt('codex.txt'),
+  anthropic: loadPrompt('anthropic.txt'),
+  gemini: loadPrompt('gemini.txt'),
+  kimi: loadPrompt('kimi.txt'),
+  meta: loadPrompt('meta.txt'),
+  trinity: loadPrompt('trinity.txt'),
+  default: loadPrompt('default.txt'),
+  plan: loadPrompt('plan.txt'),
+  buildSwitch: loadPrompt('build-switch.txt'),
+  explore: loadPrompt('explore.txt'),
+};
+
+/* opencode system.ts:27-49 — model-bazlı prompt seçimi (birebir) */
+function provider(modelId) {
+  const id = String(modelId || '');
+  if (id.includes('muse')) {
+    const name = id.includes('muse-glimmer') ? 'Muse Glimmer' : 'Muse Spark';
+    return [PROMPTS.meta.replaceAll('{{MODEL_NAME}}', name)];
+  }
+  if (id.includes('gpt-4') || id.includes('o1') || id.includes('o3')) return [PROMPTS.beast];
+  if (id.includes('gpt')) {
+    if (id.includes('codex')) return [PROMPTS.codex];
+    return [PROMPTS.gpt];
+  }
+  if (id.includes('gemini-')) return [PROMPTS.gemini];
+  if (id.includes('claude')) return [PROMPTS.anthropic];
+  if (id.toLowerCase().includes('trinity')) return [PROMPTS.trinity];
+  if (id.toLowerCase().includes('kimi')) return [PROMPTS.kimi];
+  return [PROMPTS.default];
+}
+
+/* opencode SystemPrompt.environment — model kimliği + <env> bloğu */
+function environment({ modelId, providerId, directory, worktree }) {
+  const isGit = (() => {
+    try {
+      return fs.existsSync(path.join(String(directory || ''), '.git')) ? 'yes' : 'no';
+    } catch {
+      return 'no';
+    }
+  })();
+  return [
+    [
+      `You are powered by the model named ${modelId}. The exact model ID is ${providerId}/${modelId}`,
+      `Here is some useful information about the environment you are running in:`,
+      `<env>`,
+      `  Working directory: ${directory}`,
+      `  Workspace root folder: ${worktree}`,
+      `  Is directory a git repo: ${isGit}`,
+      `  Platform: ${process.platform}`,
+      `  Today's date: ${new Date().toDateString()}`,
+      `</env>`,
+    ].join('\n'),
+  ];
+}
+
+/* opencode skill/skill.ts fmt (verbose) — skills.js scan() listesini
+   opencode katalog biçimine basar */
+function skillsCatalog(skills) {
+  const described = (skills || []).filter((s) => s && s.description);
+  if (!described.length) return undefined;
+  return [
+    'Skills provide specialized instructions and workflows for specific tasks.',
+    'Use the skill tool to load a skill when a task matches its description.',
+    '<available_skills>',
+    ...described
+      .slice()
+      .sort((a, b) => String(a.name).localeCompare(String(b.name)))
+      .flatMap((s) => [
+        '  <skill>',
+        `    <name>${s.name}</name>`,
+        `    <description>${s.description}</description>`,
+        `    <location>${s.path}</location>`,
+        '  </skill>',
+      ]),
+    '</available_skills>',
+  ].join('\n');
+}
+
+/* opencode SystemPrompt.skills — ajanın skill izni yoksa katalog gösterilmez */
+function skills(ruleset, scanFn) {
+  if (Permission.disabled(['skill'], ruleset).has('skill')) return undefined;
+  try {
+    return skillsCatalog(scanFn());
+  } catch {
+    return undefined;
+  }
+}
+
+/* opencode SessionReminders.apply — synthetic part yerine son user mesajının
+   KOPIASINA metin ekler; session.messages asla değişmez (her tur taze uygulanır).
+   wasPlan: bu oturumda daha önce plan ajanı koştu mu (engine takip eder). */
+function applyReminders(messages, agent, wasPlan) {
+  const msgs = (messages || []).slice();
+  const lastUserIdx = msgs.findLastIndex ? msgs.findLastIndex((m) => m.role === 'user') : -1;
+  if (lastUserIdx < 0 || !agent) return msgs;
+  const adds = [];
+  if (agent.name === 'plan') adds.push(PROMPTS.plan);
+  if (wasPlan && agent.name === 'build') adds.push(PROMPTS.buildSwitch);
+  if (!adds.length) return msgs;
+  const base = msgs[lastUserIdx];
+  const baseText =
+    typeof base.content === 'string' ? base.content : '';
+  msgs[lastUserIdx] = {
+    ...base,
+    content: (baseText ? baseText + '\n\n' : '') + adds.join('\n\n'),
+  };
+  return msgs;
+}
+
+module.exports = {
+  PROMPTS,
+  provider,
+  environment,
+  skills,
+  skillsCatalog,
+  applyReminders,
+  loadPrompt,
+};

package/src/agent/opencode/toolmap.js

@@ -0,0 +1,316 @@
+'use strict';
+
+/* ---------- opencode tool seti → Beast araç köprüsü ----------
+   Beast Code (bcCode) oturumlarında model opencode'un ARAÇ DİLİYLE konuşur:
+   read · edit · write · bash · glob · grep · list · fetch · websearch ·
+   todowrite · task · skill  (opencode builtin registry sırası)
+
+   Tanımlar opencode-dev'den birebir: açıklama metinleri tool/*.txt,
+   parametre şemaları opencode Schema.Struct karşılıkları (JSON Schema).
+   execMap() çağrıyı Beast'in yerleşik aracına çevirir — implementasyon
+   değişmez, modelin gördüğü dünya opencode olur. */
+
+const fs = require('fs');
+const path = require('path');
+
+const PROMPT_DIR = path.join(__dirname, 'prompts');
+function loadPrompt(name) {
+  try {
+    let t = fs.readFileSync(path.join(PROMPT_DIR, name), 'utf8');
+    if (t.charCodeAt(0) === 0xfeff) t = t.slice(1); // BOM-strip
+    return t.replace(/\s+$/, '') + '\n';
+  } catch {
+    return '';
+  }
+}
+
+/* opencode tool/shell/prompt.ts Windows (PowerShell) açıklaması — Beast
+   değerleriyle: varsayılan timeout 120s, 2000 satır / 50KB tavan, tam
+   çıktı geçici dosyaya düşer (Beast run_command zaten bu disiplinde) */
+function bashDescription() {
+  const defaultTimeoutMs = 120000;
+  const maxLines = 2000;
+  const maxBytes = '50KB';
+  return `Executes a given PowerShell command in a persistent shell session.
+
+## Command Execution
+
+Before executing the command, please follow these steps:
+  1. If the command will create new directories or files, first use \`Test-Path -LiteralPath <parent>\` to verify the parent directory exists and is the correct location
+  2. If the command will modify or delete files, first verify the paths are correct
+  3. For commands expected to create new directories or files, verify the expected file/folder location is inside the working directory
+
+After ensuring proper quoting, execute the command.
+  - Capture the output of the command.
+
+## Usage Notes
+
+  - The command argument is required.
+  - You can specify an optional timeout in milliseconds. If not specified, commands will time out after ${defaultTimeoutMs}ms.
+  - If the output exceeds ${maxLines} lines or ${maxBytes} bytes, it will be truncated and the full output will be written to a file. You can use Read with offset/limit to read specific sections or Grep to search the full content. Do NOT use \`Select-Object -First\`, \`Select-Object -Last\`, or other truncation commands to limit output; the full output will already be captured to a file for more precise searching.
+  - Avoid using Shell with PowerShell file/content cmdlets unless explicitly instructed or when these cmdlets are truly necessary for the task. Instead, always prefer using the dedicated tools for reading, searching, and editing files:
+    - Read: Reading and analyzing file contents (with line numbers)
+    - Edit: Making targeted changes to specific sections of files
+    - Write: Creating new files from scratch
+    - Grep: Searching file contents with regex patterns
+    - Glob: Finding files by patterns (e.g., "src/**/*.ts")
+  - When issuing multiple commands:
+    - If the commands are independent and can run in parallel, make multiple bash tool calls in a single message. For example, if you need to run "git status" and "git diff", send a single message with two bash tool calls in parallel.
+    - Use ';' only when you need to run commands sequentially but don't care if earlier commands fail
+    - If the commands depend on each other and must run sequentially, avoid '&&' in this shell because Windows PowerShell (5.1) does not support it. Use PowerShell conditionals such as \`cmd1; if ($?) { cmd2 }\` when later commands must depend on earlier success.
+    - DO NOT use newlines to separate commands (newlines are ok in quoted strings)
+  - The shell is PERSISTENT: working directory and environment variables are preserved between calls.
+  - Commands run in Windows PowerShell (5.1).`;
+}
+
+/* opencode tool şemaları — Schema.Struct karşılığı JSON Schema'lar */
+const DEFS = () => [
+  {
+    name: 'bash',
+    description: bashDescription(),
+    parameters: {
+      type: 'object',
+      properties: {
+        command: { type: 'string', description: 'The command to execute' },
+        timeout: { type: 'integer', description: 'Optional timeout in milliseconds' },
+      },
+      required: ['command'],
+    },
+  },
+  {
+    name: 'read',
+    description: loadPrompt('read.txt'),
+    parameters: {
+      type: 'object',
+      properties: {
+        filePath: { type: 'string', description: 'The absolute path to the file or directory to read' },
+        offset: { type: 'integer', description: 'The line number to start reading from (1-indexed)' },
+        limit: { type: 'integer', description: 'The maximum number of lines to read (defaults to 2000)' },
+      },
+      required: ['filePath'],
+    },
+  },
+  {
+    name: 'edit',
+    description: loadPrompt('edit.txt'),
+    parameters: {
+      type: 'object',
+      properties: {
+        filePath: { type: 'string', description: 'The absolute path to the file to modify' },
+        oldString: { type: 'string', description: 'The text to replace' },
+        newString: { type: 'string', description: 'The text to replace it with (must be different from oldString)' },
+        replaceAll: { type: 'boolean', description: 'Replace all occurrences of oldString (default false)' },
+      },
+      required: ['filePath', 'oldString', 'newString'],
+    },
+  },
+  {
+    name: 'write',
+    description: loadPrompt('write.txt'),
+    parameters: {
+      type: 'object',
+      properties: {
+        filePath: { type: 'string', description: 'The absolute path to the file to write (will create the file if it does not exist, will overwrite if it does)' },
+        content: { type: 'string', description: 'The content to write' },
+      },
+      required: ['filePath', 'content'],
+    },
+  },
+  {
+    name: 'glob',
+    description: loadPrompt('glob.txt'),
+    parameters: {
+      type: 'object',
+      properties: {
+        pattern: { type: 'string', description: 'The glob pattern to match files against' },
+        path: { type: 'string', description: 'The directory to search in. If not specified, the current working directory will be used. IMPORTANT: Omit this field to use the default directory. DO NOT enter "undefined" or "null" - simply omit it for the default behavior. Must be a valid directory path if provided.' },
+      },
+      required: ['pattern'],
+    },
+  },
+  {
+    name: 'grep',
+    description: loadPrompt('grep.txt'),
+    parameters: {
+      type: 'object',
+      properties: {
+        pattern: { type: 'string', description: 'The regex pattern to search for in file contents' },
+        path: { type: 'string', description: 'The directory to search in. Defaults to the current working directory.' },
+        include: { type: 'string', description: 'File pattern to include in the search (e.g. "*.js", "*.{ts,tsx}")' },
+      },
+      required: ['pattern'],
+    },
+  },
+  {
+    name: 'list',
+    description: 'List files and directories in a given path. Returns entries sorted by name (directories first, then files), with type indicator and size.',
+    parameters: {
+      type: 'object',
+      properties: {
+        path: { type: 'string', description: 'The directory path to list. Defaults to the current working directory.' },
+      },
+    },
+  },
+  {
+    name: 'webfetch',
+    description: loadPrompt('webfetch.txt'),
+    parameters: {
+      type: 'object',
+      properties: {
+        url: { type: 'string', description: 'The URL to fetch' },
+        format: { type: 'string', enum: ['text', 'markdown', 'html'], description: 'The format to return the content in (text, markdown, or html). Defaults to markdown.' },
+        timeout: { type: 'integer', description: 'Optional timeout in seconds (max 120)' },
+        max_chars: { type: 'integer', description: 'Optional maximum characters to return' },
+      },
+      required: ['url'],
+    },
+  },
+  {
+    name: 'websearch',
+    description: loadPrompt('websearch.txt'),
+    parameters: {
+      type: 'object',
+      properties: {
+        query: { type: 'string', description: 'Websearch query' },
+        numResults: { type: 'integer', description: 'Number of search results to return (default: 8)' },
+        type: { type: 'string', enum: ['auto', 'fast', 'deep'], description: "Search type - 'auto': balanced search (default), 'fast': quick results, 'deep': comprehensive search" },
+      },
+      required: ['query'],
+    },
+  },
+  {
+    name: 'todowrite',
+    description: loadPrompt('todowrite.txt'),
+    parameters: {
+      type: 'object',
+      properties: {
+        todos: {
+          type: 'array',
+          description: 'The updated todo list',
+          items: {
+            type: 'object',
+            properties: {
+              content: { type: 'string', description: 'Brief description of the task' },
+              status: { type: 'string', enum: ['pending', 'in_progress', 'completed', 'cancelled'], description: 'Current status of the task' },
+              priority: { type: 'string', enum: ['high', 'medium', 'low'], description: 'Priority level of the task' },
+            },
+            required: ['content', 'status'],
+          },
+        },
+      },
+      required: ['todos'],
+    },
+  },
+  {
+    name: 'task',
+    description: loadPrompt('task.txt'),
+    parameters: {
+      type: 'object',
+      properties: {
+        description: { type: 'string', description: 'A short description of the task (max 80 chars)' },
+        prompt: { type: 'string', description: 'The full task description for the subagent to execute' },
+        subagent_type: { type: 'string', enum: ['general', 'explore'], description: 'Type of subagent to launch (options: explore, general)' },
+      },
+      required: ['description', 'prompt', 'subagent_type'],
+    },
+  },
+  {
+    name: 'skill',
+    description: loadPrompt('skill.txt'),
+    parameters: {
+      type: 'object',
+      properties: {
+        name: { type: 'string', description: 'The name of the skill from available_skills' },
+      },
+      required: ['name'],
+    },
+  },
+  {
+    /* Beast Sandbox'a özgü: UZUN SÜRELİ süreçleri (dev server vb.) ÇALIŞTIR
+       paneline devreder — tur BLOKLAMAZ. panel_run izni yalnız sandbox
+       ajanında allow edilir (agents.js); BC build/plan'da gizlidir. */
+    name: 'panel_run',
+    description:
+      'Run a LONG-RUNNING process (dev server, background service) in the managed RUN panel instead of the blocking shell. The turn does not wait for it to finish; output streams live in the panel and can be stopped by the user. Use this instead of starting servers with bash — a blocking bash call would stall the whole job. Only ONE process runs at a time.',
+    parameters: {
+      type: 'object',
+      properties: {
+        command: { type: 'string', description: 'The command to run in the panel' },
+      },
+      required: ['command'],
+    },
+  },
+];
+
+/* OpenAI function-call formatına çevirir (engine TOOLS formatı) */
+function definitions() {
+  return DEFS().map((d) => ({ type: 'function', function: d }));
+}
+
+const TO_BUILTIN = {
+  bash: 'run_command',
+  read: 'read_file',
+  edit: 'edit_file',
+  write: 'write_file',
+  glob: 'glob',
+  grep: 'grep',
+  list: 'list_dir',
+  webfetch: 'webfetch',
+  websearch: 'web_search',
+  todowrite: 'todo_write',
+  task: 'delegate_task',
+  panel_run: 'panel_run',
+};
+
+/* opencode parametreleri → Beast araç parametreleri */
+function execMap(name, args) {
+  const a = args || {};
+  switch (name) {
+    case 'read':
+      return { name: 'read_file', args: { path: a.filePath, offset: a.offset, limit: a.limit } };
+    case 'edit':
+      return {
+        name: 'edit_file',
+        args: { path: a.filePath, old_string: a.oldString, new_string: a.newString, replace_all: a.replaceAll === true },
+      };
+    case 'write':
+      return { name: 'write_file', args: { path: a.filePath, content: a.content } };
+    case 'bash':
+      return { name: 'run_command', args: { command: a.command, timeout_ms: a.timeout } };
+    case 'list':
+      return { name: 'list_dir', args: { path: a.path } };
+    case 'webfetch':
+      return { name: 'webfetch', args: { url: a.url, format: a.format, timeout: a.timeout, max_chars: a.max_chars } };
+    case 'websearch':
+      return { name: 'web_search', args: { query: a.query, max_results: a.numResults || a.max_results } };
+    case 'todowrite':
+      return { name: 'todo_write', args: { items: (a.todos || []).map(opencodeTodoToItem) } };
+    case 'task':
+      return {
+        name: 'delegate_task',
+        args: { task: a.prompt || a.description, context: a.description || '', subagent_type: a.subagent_type },
+      };
+    default:
+      return { name: TO_BUILTIN[name] || name, args: a };
+  }
+}
+
+/* opencode Todo.Info {content, status, priority} → Beast iç depo {title, status}
+   opencode durum makinesi: pending → in_progress → completed | cancelled
+   Beast iç durumu: pending | active | done */
+const OC_TODO_STATUS = { pending: 'pending', in_progress: 'active', completed: 'done', cancelled: 'done' };
+function opencodeTodoToItem(t) {
+  const it = t || {};
+  return {
+    title: String(it.content ?? '').replace(/\s+/g, ' ').trim().slice(0, 120),
+    status: OC_TODO_STATUS[String(it.status || 'pending')] || 'pending',
+  };
+}
+
+/* BC oturumunda modele sunulan opencode araç adları (permission filtresinden
+   önce tam set — opencode registry builtin sırası) */
+function names() {
+  return DEFS().map((d) => d.name);
+}
+
+module.exports = { definitions, names, execMap, TO_BUILTIN, opencodeTodoToItem };