@opencoven/coven-code 0.0.4 → 0.0.7

package/README.md

@@ -1,146 +1,45 @@
-# Coven Code
+# @opencoven/coven-code
 
-A small Node CLI for local Coven Code workflows. It runs deterministic local
-command, thread, tool, MCP, skill, plugin, and SDK flows so integration behavior
-can be exercised end-to-end without a hosted service.
+[![Version](https://img.shields.io/npm/v/@opencoven/coven-code?style=flat-square)](https://www.npmjs.com/package/@opencoven/coven-code)
+[![License](https://img.shields.io/badge/License-GPL--3.0-blue?style=flat-square)](https://github.com/OpenCoven/coven-code/blob/main/LICENSE.md)
 
-## Quickstart
+**Coven Code** — open-source, multi-provider agentic coding TUI built in Rust.  
+OpenCoven fork of [Claurst](https://github.com/Kuberwastaken/claurst) by Kuber Mehta (GPL-3.0).
 
-```sh
-npm test
-npm start
-npm run coven-code -- --help
-npm run coven-code -- -x "what is 2+2?"
-echo "list markdown files" | npm run coven-code
-```
-
-The package exposes `coven-code` and `coven-code-sdk` bins.
-
-## Documentation
-
-- [CLI reference](docs/CLI.md) - commands, flags, interactive mode, execute mode, JSON streaming, and thread workflows
-- [Configuration](docs/CONFIGURATION.md) - settings files, environment variables, precedence, and common examples
-- [MCP, skills, and plugins](docs/MCP-SKILLS-PLUGINS.md) - extension points for tools, MCP servers, skills, and project plugins
-- [SDK](docs/SDK.md) - package exports, install helper, execution API, permissions, and thread helpers
-- [Development](docs/DEVELOPMENT.md) - repository layout, local checks, release notes, and safety expectations
-- [Coven Dogfood Protocol](docs/DOGFOOD-PROTOCOL.md) - daily task packets, harness assignment, handoff, verification, retrospectives, and issue capture
-- [Demo](docs/DEMO.md) - 15-20 minute end-to-end walkthrough of every shipped surface (`npm run demo`)
-
-## Interactive Mode
-
-Running `coven-code` with no arguments in a terminal starts the `neo-blessed`
-panel TUI. Type `/` in the composer to open the slash-command menu with live
-filtering and a details panel for commands, skills, and plugin actions.
+## Install
 
-```text
-$ coven-code
-Coven Code 0.0.4
-/Users/example/project
-[chat]  lane   tools   threads   config   help    mode: smart   effort: high
---------------------------------------------------------------------------------
-Ready. Type a prompt or /help.
---------------------------------------------------------------------------------
->
+```bash
+npm install -g @opencoven/coven-code
+# or
+bun install -g @opencoven/coven-code
 ```
 
-The classic readline REPL remains available for scripts, demos, and compatibility:
+On install, the correct pre-built native binary for your platform is automatically downloaded from [GitHub Releases](https://github.com/OpenCoven/coven-code/releases). No compilation required.
 
-```sh
-COVEN_CODE_REPL=1 coven-code
-```
-
-Line history is stored at `${XDG_CONFIG_HOME:-~/.config}/coven-code/repl_history`.
-Use `COVEN_CODE_REPL_HISTORY_FILE` to override the path or
-`COVEN_CODE_REPL_HISTORY=0` to disable history in the classic REPL.
-
-`COVEN_CODE_TUI_SCRIPTED=1` is reserved for deterministic tests and automation
-that need to drive the TUI with newline-separated input.
+## Usage
 
-## Execute Mode
-
-```sh
-coven-code -x "prompt"
-coven-code -x "see @README.md" </dev/null
-coven-code -x "..." --stream-json
-coven-code -x "..." --stream-json --stream-json-input < msgs.jsonl
-coven-code --continue T-... -x "continue a thread"
-coven-code --toolbox ./toolbox -x "use tb__my_tool"
-coven-code --skills ./skills -x "use the data-map skill"
+```bash
+coven-code                    # interactive TUI
+coven-code -p "fix this bug"  # headless one-shot
 ```
 
-Settings use the `covenCode.*` prefix. User settings live under
-`${XDG_CONFIG_HOME:-~/.config}/coven-code/settings.json`; workspace settings
-live under `.coven-code/settings.json`.
-
-Useful settings and env vars:
+## Providers
 
-- `covenCode.fuzzy.alwaysIncludePaths`
-- `covenCode.tools.enable` / `covenCode.tools.disable`
-- `covenCode.permissions`
-- `covenCode.commands.allowlist`
-- `covenCode.mcpServers`
-- `covenCode.mcpPermissions`
-- `covenCode.mcpRegistry.url`
-- `covenCode.notifications.enabled`
-- `covenCode.updates.mode`
-- `covenCode.defaultVisibility`
-- `COVEN_CODE_TOOLBOX`
-- `COVEN_CODE_API_KEY`
-- `COVEN_CODE_SKIP_UPDATE_CHECK`
-- `COVEN_CODE_FORCE_BEL`
-- `COVEN_CODE_CLI_PATH`
-- `COVEN_CODE_HOME`
+Supports Anthropic (Claude), OpenAI, Google Gemini, Groq, Ollama, LM Studio, OpenRouter, Bedrock, Vertex, and any OpenAI-compatible endpoint.
 
-## Implemented Surface
-
-Core:
-
-- `coven-code --help`, `coven-code --version`
-- panel TUI with transcript, lane status, tabs, status rail, composer, keyboard shortcuts, and command palette
-- lane view for isolated worktree/branch context, harness selection, terminal output, changed files, diff summary, verification state, and PR/merge/cleanup status
-- compatibility REPL with `/new`, `/continue`, `/queue`, `/mode`, `/reasoning`, persistent history, and multiline prompts
-- command-palette aliases for thread archive, visibility, diagnostics, IDE, skills, plugins, and help
-- one-shot execute mode with `-x` / `--execute`
-- `@file`, glob, image, explicit thread, and `@@query` thread-search references
-- `--stream-json`, `--stream-json-thinking`, `--stream-json-input`, and `--reasoning-effort`
-- local persisted threads with labels, visibility, archive state, reports, and maps
-
-Tools and permissions:
-
-- built-in tools: `Bash`, `Read`, `Grep`, `glob`, `create_file`, `edit_file`, `undo_edit`, `Task`, `oracle`, `librarian`, `painter`, `mermaid`, `look_at`, `web_search`, `read_web_page`, `find_thread`, `finder`, and `read_mcp_resource`
-- `coven-code tools list|make|show|use`
-- toolbox discovery through `COVEN_CODE_TOOLBOX`, `--toolbox`, and `${XDG_CONFIG_HOME:-~/.config}/coven-code/tools`
-- `coven-code permissions list|add|edit|test`
-- delegated permission helpers receive `AGENT=coven-code` and `COVEN_CODE_THREAD_ID`
-
-MCP, skills, and plugins:
-
-- `coven-code mcp add|list|doctor|approve`
-- remote MCP headers, OAuth credential storage in `~/.coven-code/oauth`, token refresh, Streamable HTTP sessions, and SSE fallback
-- MCP registry enforcement that blocks unlisted servers and fails closed when the registry is unreachable
-- skill discovery from configured roots, `.agents/skills`, legacy `.claude/skills` roots, and built-ins
-- `coven-code skill list|show`
-- project plugins from `.coven-code/plugins/*.ts` and user plugins from `${XDG_CONFIG_HOME:-~/.config}/coven-code/plugins/*.ts`
-- plugin tools, commands, lifecycle events, configuration, status items, UI fallbacks, and helper APIs
-
-SDK:
-
-- `coven-code-sdk install [--force]`
-- package root exports `execute`, `createUserMessage`, `createPermission`, and `threads`
-- SDK execution streams local `coven-code --stream-json` messages
-- SDK options map to CLI args for cwd, env, mode, reasoning effort, thinking, labels, visibility, archive, continuation, settings, toolbox, skills, MCP config, permissions, enabled tools, system prompt, logging, and permission bypass
+```bash
+coven-code --provider openai "refactor this"
+coven-code --provider ollama --model llama3.2 "explain this"
+```
 
-## Development Checks
+## Configuration
 
-Run the full local verification gate before committing or publishing:
+Settings: `~/.coven-code/settings.json`  
+Env prefix: `COVEN_CODE_*`
 
-```sh
-git diff --check
-node ./bin/coven-code.mjs --help
-node ./bin/coven-code.mjs -x "what is 2+2?"
-npm test
-```
+## Links
 
-The project is intentionally dependency-free at runtime. Keep new behavior
-covered in `test/cli.test.mjs`, avoid hosted-service assumptions, and keep
-examples deterministic.
+- [OpenCoven](https://opencoven.ai)
+- [GitHub](https://github.com/OpenCoven/coven-code)
+- [Issues](https://github.com/OpenCoven/coven-code/issues)
+- [Upstream (Claurst)](https://github.com/Kuberwastaken/claurst)

package/bin/coven-code

@@ -0,0 +1,26 @@
+#!/usr/bin/env node
+'use strict';
+
+const { spawnSync } = require('child_process');
+const path = require('path');
+const os = require('os');
+const fs = require('fs');
+
+const ext = os.platform() === 'win32' ? '.exe' : '';
+const binary = path.join(__dirname, '..', 'native', `coven-code`);
+
+if (!fs.existsSync(binary)) {
+  console.error(
+    'coven-code: native binary not found.\n' +
+    'Try reinstalling: npm install -g /coven-code\n' +
+    `Expected: ${binary}`
+  );
+  process.exit(1);
+}
+
+const result = spawnSync(binary, process.argv.slice(2), {
+  stdio: 'inherit',
+  env: process.env,
+});
+
+process.exit(result.status ?? 1);

package/install.js

@@ -0,0 +1,117 @@
+#!/usr/bin/env node
+'use strict';
+
+const https = require('https');
+const http = require('http');
+const fs = require('fs');
+const path = require('path');
+const os = require('os');
+const { execFileSync } = require('child_process');
+
+const pkg = require('./package.json');
+const VERSION = pkg.version;
+const REPO = 'OpenCoven/coven-code';
+const BASE_URL = `https://github.com/${REPO}/releases/download/v${VERSION}`;
+const NATIVE_DIR = path.join(__dirname, 'native');
+
+function getPlatform() {
+  const platform = process.platform;
+  const arch = process.arch;
+
+  if (platform === 'win32' && arch === 'x64') {
+    return { artifact: 'coven-code-windows-x86_64', ext: '.exe', archive: '.zip' };
+  }
+  if (platform === 'linux' && arch === 'x64') {
+    return { artifact: 'coven-code-linux-x86_64', ext: '', archive: '.tar.gz' };
+  }
+  if (platform === 'linux' && arch === 'arm64') {
+    return { artifact: 'coven-code-linux-aarch64', ext: '', archive: '.tar.gz' };
+  }
+  if (platform === 'darwin' && arch === 'x64') {
+    return { artifact: 'coven-code-macos-x86_64', ext: '', archive: '.tar.gz' };
+  }
+  if (platform === 'darwin' && arch === 'arm64') {
+    return { artifact: 'coven-code-macos-aarch64', ext: '', archive: '.tar.gz' };
+  }
+  throw new Error(
+    `Unsupported platform: ${platform}/${arch}.\n` +
+    `Install manually from: https://github.com/${REPO}/releases/tag/v${VERSION}`
+  );
+}
+
+function download(url, dest) {
+  return new Promise((resolve, reject) => {
+    const file = fs.createWriteStream(dest);
+    const get = url.startsWith('https') ? https : http;
+    get.get(url, (res) => {
+      if (res.statusCode === 301 || res.statusCode === 302) {
+        file.close();
+        try { fs.unlinkSync(dest); } catch (_) {}
+        download(res.headers.location, dest).then(resolve).catch(reject);
+        return;
+      }
+      if (res.statusCode !== 200) {
+        file.close();
+        try { fs.unlinkSync(dest); } catch (_) {}
+        reject(new Error(`HTTP ${res.statusCode} downloading ${url}`));
+        return;
+      }
+      res.pipe(file);
+      file.on('finish', () => file.close(resolve));
+      file.on('error', (err) => {
+        try { fs.unlinkSync(dest); } catch (_) {}
+        reject(err);
+      });
+    }).on('error', (err) => {
+      try { fs.unlinkSync(dest); } catch (_) {}
+      reject(err);
+    });
+  });
+}
+
+async function main() {
+  const { artifact, ext, archive } = getPlatform();
+  const archiveName = `${artifact}${archive}`;
+  const url = `${BASE_URL}/${archiveName}`;
+  const tmpPath = path.join(os.tmpdir(), `coven-code-install-${process.pid}${archive}`);
+  const binaryDest = path.join(NATIVE_DIR, `coven-code${ext}`);
+
+  if (fs.existsSync(binaryDest)) {
+    console.log('coven-code: native binary already present, skipping download.');
+    return;
+  }
+
+  fs.mkdirSync(NATIVE_DIR, { recursive: true });
+
+  console.log(`coven-code: downloading v${VERSION} for ${process.platform}/${process.arch}`);
+  console.log(`            ${url}`);
+  await download(url, tmpPath);
+
+  console.log('coven-code: extracting...');
+  if (archive === '.zip') {
+    execFileSync('powershell', [
+      '-NoProfile', '-NonInteractive', '-Command',
+      `Expand-Archive -Force -Path "${tmpPath}" -DestinationPath "${NATIVE_DIR}"`
+    ]);
+  } else {
+    execFileSync('tar', ['-xzf', tmpPath, '-C', NATIVE_DIR]);
+  }
+
+  try { fs.unlinkSync(tmpPath); } catch (_) {}
+
+  if (!fs.existsSync(binaryDest)) {
+    throw new Error(`Extraction succeeded but binary not found at ${binaryDest}`);
+  }
+
+  if (ext === '') {
+    fs.chmodSync(binaryDest, 0o755);
+  }
+
+  console.log(`coven-code: ready — run \`coven-code\` to start.`);
+}
+
+main().catch((err) => {
+  console.error(`\ncoven-code install failed: ${err.message}`);
+  console.error(`Manual install: https://github.com/${REPO}/releases/tag/v${VERSION}\n`);
+  process.exit(1);
+});

package/package.json

@@ -1,37 +1,40 @@
 {
   "name": "@opencoven/coven-code",
-  "version": "0.0.4",
-  "type": "module",
+  "version": "0.0.7",
+  "description": "Open-source, multi-provider agentic coding TUI for the terminal — OpenCoven fork of Claurst",
+  "license": "GPL-3.0-only",
   "repository": {
     "type": "git",
     "url": "git+https://github.com/OpenCoven/coven-code.git"
   },
-  "exports": {
-    ".": "./src/sdk.mjs"
+  "homepage": "https://opencoven.ai",
+  "bugs": {
+    "url": "https://github.com/OpenCoven/coven-code/issues"
   },
+  "keywords": [
+    "ai",
+    "coding-agent",
+    "tui",
+    "rust",
+    "coven",
+    "opencoven",
+    "terminal",
+    "agentic"
+  ],
+  "author": "OpenCoven (based on Claurst by Kuber Mehta)",
   "bin": {
-    "coven-code": "bin/coven-code.mjs",
-    "coven-code-sdk": "bin/coven-code-sdk.mjs"
+    "coven-code": "bin/coven-code"
+  },
+  "scripts": {
+    "postinstall": "node install.js"
   },
   "files": [
     "bin/",
-    "docs/",
-    "src/",
-    "README.md"
+    "install.js",
+    "README.md",
+    "ATTRIBUTION.md"
   ],
-  "publishConfig": {
-    "access": "public"
-  },
-  "scripts": {
-    "test": "node --test test/cli.test.mjs test/commands.test.mjs test/config.test.mjs test/execute.test.mjs test/mcp.test.mjs test/permissions.test.mjs test/plugins-agent-events.test.mjs test/plugins-commands.test.mjs test/plugins-discovery.test.mjs test/plugins-tool-events.test.mjs test/sdk.test.mjs test/skills.test.mjs test/stream-json.test.mjs test/threads.test.mjs test/tools.test.mjs test/tui.test.mjs",
-    "start": "node ./bin/coven-code.mjs",
-    "coven-code": "node ./bin/coven-code.mjs",
-    "demo": "bash ./scripts/demo.sh"
-  },
   "engines": {
-    "node": ">=24"
-  },
-  "dependencies": {
-    "neo-blessed": "0.2.0"
+    "node": ">=16"
   }
 }

package/bin/coven-code-sdk.mjs

@@ -1,12 +0,0 @@
-#!/usr/bin/env node
-import { runInstallCommand } from '../src/sdk-install.mjs';
-
-runInstallCommand(process.argv.slice(2))
-  .then((exitCode) => {
-    process.exit(exitCode);
-  })
-  .catch((error) => {
-    const message = error instanceof Error ? error.message : String(error);
-    console.error(message);
-    process.exit(1);
-  });

package/bin/coven-code.mjs

@@ -1,19 +0,0 @@
-#!/usr/bin/env node
-import { main, UsageError } from '../src/main.mjs';
-import { CLI_NAME } from '../src/constants.mjs';
-
-process.stdout.on('error', (error) => {
-  if (error?.code === 'EPIPE') process.exit(0);
-  throw error;
-});
-process.on('SIGPIPE', () => process.exit(0));
-
-main().catch((error) => {
-  if (error instanceof UsageError) {
-    console.error(`${CLI_NAME}: ${error.message}`);
-    console.error(`Run \`${CLI_NAME} --help\` for usage.`);
-    process.exit(2);
-  }
-  console.error(error.stack || String(error));
-  process.exit(1);
-});

package/docs/CLI.md

@@ -1,256 +0,0 @@
-# CLI Reference
-
-## Entry Points
-
-During development, run the CLI through npm:
-
-```sh
-npm run coven-code -- --help
-npm run coven-code -- -x "what is 2+2?"
-```
-
-Installed package binaries:
-
-```sh
-coven-code --help
-coven-code-sdk install
-```
-
-The direct local entrypoint is:
-
-```sh
-node ./bin/coven-code.mjs
-```
-
-## Help
-
-`coven-code --help` prints the full option and command reference. The excerpt
-below is verified against actual `--help` output by the test suite; regenerate
-it with `node ./bin/coven-code.mjs --help` between the markers if it drifts.
-
-<!-- BEGIN: coven-code --help -->
-```text
-Coven Code
-
-Usage: coven-code [options] [command]
-
-Run with no arguments in a terminal to enter the panel TUI. Set
-COVEN_CODE_REPL=1 to use the classic readline REPL. Piped stdin becomes the
-first interactive message when stdout is a TTY. Pass --execute or redirect
-stdout to run a single turn and exit.
-
-Options:
-      --execute, -x [prompt]   Run one agent turn, print the final answer, and exit
-      --stream-json            Emit structured JSONL in execute mode
-      --stream-json-thinking   Include thinking blocks in stream JSON output
-      --stream-json-input      Read one or more user messages as JSONL from stdin
-      --dangerously-allow-all  Allow tool calls that would otherwise require approval
-      --mcp-config <json>      Add an inline MCP server config for this run
-      --settings-file <path>   Read user settings from a specific file
-      --mode <name>            Agent mode: smart, deep, rush, or large
-      --reasoning-effort <level>
-                               Set reasoning effort for the active mode
-      --label <name>           Add a label to the created or continued thread
-      --visibility <level>     Thread visibility: private, public, workspace, group, or unlisted
-      --archive                Archive the thread after the execute turn
-      --continue [thread-id]   Continue the latest active thread or the specified thread
-      --toolbox <path>         Use a PATH-like toolbox root for this run
-      --skills <path>          Use a PATH-like skill root for this run
-      --jetbrains              Connect to a JetBrains IDE
-  -h, --help                   Show help
-  -v, --version                Show version
-
-Commands:
-  login                        Print local login instructions
-  update                       Check for updates
-  usage                        Show local usage estimates
-  review                       Run configured local review checks
-  tools list|make|show|use     Manage built-in and toolbox tools
-  permissions list             Show permission policy rules
-  config edit [--workspace]    Open settings in $EDITOR
-  ide connect                  Connect or inspect local IDE integration
-  mcp add|list|doctor|approve|oauth
-                               Manage MCP server settings
-  skill list|show              Inspect discovered Coven Code agent skills
-  plugins list|reload          Show and reload project and user plugin files
-  threads list|show|search|archive|visibility|continue|map|report
-                               Manage local thread records
-  agents-md list               Show AGENTS.md guidance files used for this cwd
-  agents list                  Alias for agents-md list
-
-TUI lane commands:
-  /lane refresh                Refresh worktree, branch, changed files, and diff summary
-  /lane harness <name|next>    Select smart, deep, rush, or large for the lane
-  /lane verify                 Run the detected verification command for the lane
-  /lane status|diff            Show lane status or diff summary
-```
-<!-- END: coven-code --help -->
-
-## Interactive Mode
-
-Run with no arguments in a TTY:
-
-```sh
-coven-code
-```
-
-The default interactive surface is a full-screen `neo-blessed` panel TUI with a
-transcript, tabs, compact status line, command palette, slash-command menu, and
-composer. It keeps a current local thread and accepts slash commands:
-
-```text
-Coven Code 0.0.4
-[chat]  lane   tools   threads   config   help    mode: smart   effort: high
---------------------------------------------------------------------------------
-Ready. Type a prompt or /help.
---------------------------------------------------------------------------------
-> /help
-```
-
-Supported interactive behaviors include:
-
-- panel tabs for chat, lane, tools, threads, config, and help
-- `/lane refresh` to load current worktree, branch, changed files, and diff summary
-- `/lane harness <smart|deep|rush|large|next>` to select the active lane harness
-- `/lane verify` to run the detected verification command and keep its output in the lane panel
-- `/lane status` and `/lane diff` to inspect the visible lane model
-- `/` opens a filtered command menu with details for built-ins, skills, and
-  plugin commands
-- Tab completes the selected slash command; Enter accepts it; Esc closes the menu
-- command palette actions for common thread and tool workflows
-- persistent line history
-- `/new` to start a fresh thread
-- `/continue` to resume a thread
-- `/queue` to queue a follow-up prompt
-- `/mode` to inspect or change the active mode
-- `/reasoning` to inspect or change reasoning effort
-- command aliases for tools, skills, plugins, threads, diagnostics, and IDE
-  inspection
-- multiline prompts with trailing backslash continuation
-
-Use the classic readline REPL explicitly when a plain prompt is better for
-scripts, demos, or compatibility:
-
-```sh
-COVEN_CODE_REPL=1 coven-code
-```
-
-Disable classic REPL history for demos or tests:
-
-```sh
-COVEN_CODE_REPL_HISTORY=0 coven-code
-```
-
-`COVEN_CODE_TUI_SCRIPTED=1` is a deterministic test and automation hook for
-feeding newline-separated TUI input. It is not intended as a normal user mode.
-
-## Execute Mode
-
-Use `--execute` or `-x` for one turn and exit:
-
-```sh
-coven-code -x "what is 2+2?"
-coven-code -x "summarize @README.md"
-coven-code --continue T-example -x "continue this thread"
-```
-
-Piped stdin is combined with the prompt:
-
-```sh
-echo "list markdown files" | coven-code
-echo "additional context" | coven-code -x "answer using this context"
-```
-
-`-x`, `--stream-json`, `--stream-json-input`, stdin piping, and subcommands
-keep their existing noninteractive behavior; the panel TUI only changes the
-bare no-argument TTY path.
-
-## File, Image, Glob, and Thread References
-
-Execute prompts can reference local context:
-
-```sh
-coven-code -x "summarize @README.md"
-coven-code -x "inspect @docs/*.md"
-coven-code -x "describe @image.png"
-coven-code -x "compare this with @T-existing-thread"
-coven-code -x "continue from @@search words"
-```
-
-Text references are capped to keep prompts bounded. Binary files are skipped
-unless handled as supported media references.
-
-## Stream JSON
-
-Use `--stream-json` for structured JSONL suitable for frontends and SDKs:
-
-```sh
-coven-code -x "what is 2+2?" --stream-json
-coven-code -x "inspect this" --stream-json --stream-json-thinking
-```
-
-Use `--stream-json-input` to read user messages from JSONL stdin:
-
-```sh
-coven-code --stream-json --stream-json-input < messages.jsonl
-```
-
-Stream JSON emits initialization, user, assistant, tool use, tool result, and
-final result messages.
-
-## Tools
-
-```sh
-coven-code tools list
-coven-code tools show Bash
-coven-code tools make --bash my-tool
-coven-code tools use tb__my-tool
-```
-
-Built-in tools include `Bash`, `Read`, `Grep`, `glob`, `create_file`,
-`edit_file`, `undo_edit`, `Task`, `oracle`, `librarian`, `painter`, `mermaid`,
-`look_at`, `web_search`, `read_web_page`, `find_thread`, `finder`, and
-`read_mcp_resource`.
-
-## Permissions
-
-```sh
-coven-code permissions list
-coven-code permissions list --builtin
-coven-code permissions add allow Bash command.name git
-coven-code permissions test Bash command.name git
-```
-
-Permission rules can allow or reject tool calls by tool name, action, command,
-arguments, server names, URLs, and nested context fields.
-
-## Threads
-
-```sh
-coven-code threads list
-coven-code threads show T-example
-coven-code threads search "query"
-coven-code threads continue T-example -x "follow up"
-coven-code threads archive T-example
-coven-code threads visibility T-example workspace
-coven-code threads map T-example
-coven-code threads report T-example
-```
-
-Threads are local records that preserve prompts, labels, visibility, archive
-state, diagnostic reports, and reference edges.
-
-## Other Commands
-
-```sh
-coven-code login
-coven-code update
-coven-code usage
-coven-code review
-coven-code config edit
-coven-code config edit --workspace
-coven-code ide connect
-coven-code agents-md list
-```
-
-`agents list` is an alias for `agents-md list`.