X Web Viewer

# Codex Features and Practical Usage 🏛️ When you roll Codex out company-wide, leaving "which MCPs, models, and permissions are allowed" to each person is risky. Managed configuration lets admins draw the boundaries and enforce them. 🏷️ Title: Centralized Managed Configuration & Governance 🔗 URL: 📘 Overview Codex enterprise managed configuration lets IT admins control agent behavior across the whole organization from a central place. Allowed MCP servers, models, and approval/sandbox permissions are distributed as policy so that users cannot quietly step outside the approved boundaries. ⚙️ How It Works Management has two tiers: ・`requirements.toml`: admin-enforced constraints that users cannot override. ・`managed_config.toml`: starting defaults that users can still change during a session. Precedence, highest to lowest: macOS MDM preferences → cloud-managed requirements (ChatGPT Business/Enterprise) → system files (`/etc/codex/` or `%ProgramData%\OpenAI\Codex\`) → the user's `config.toml` → CLI overrides. Managed layers override CLI flags and are reapplied every session. Lockable settings include approval policies, sandbox modes (e.g. blocking `full_access`), an allowlist of permission profiles, MCP server approval (by command or URL — omitted servers are disabled), feature flags (such as `browser_use`), enforced automatic review, network domain allow/deny lists, lifecycle hooks, and command rules. 🛠️ Practical Usage For example, a `requirements.toml` can pin `allowed_approval_policies` and `allowed_sandbox_modes`, set `deny_read` for `~/.ssh` and `.env`, add a prefix rule forbidding `rm`, and register a single approved MCP server. This blocks `full_access`, denies reads of `~/.ssh` and `.env`, forbids `rm`, and enables only the approved MCP server. 💡 Use Cases You can hand every employee only the MCP servers and models you have approved, and uniformly block reading secret files or running destructive commands — governance that stays consistent without redeploying clients. ⚠️ Caveats Once `allowed_permission_profiles` is defined, it becomes the exclusive list; later sources must explicitly re-enable a profile to use it. Cloud requirements are cached locally, so offline sessions may not reflect the latest policy. Don't push custom profiles to clients running Codex 0.137.0 or earlier. For group-based rules the first matching group wins, and unset fields are not inherited from later groups — use explicit allowlists to combine policies. #Codex# #Enterprise#

# Useful but Little-Known Features of Claude Agent SDK 🌍 What if your agent tries to delete files on its own? Runtime approval gives you safe control! The `canUseTool` callback lets you approve, deny, or modify tool inputs in real time before execution. 📌 Title: Runtime Approval via canUseTool Callback 🔗 URL: 🧩 Overview `canUseTool` is a callback function passed in query options, fired at the last step (6th) of the permission evaluation flow. It never fires for tool calls auto-approved at an earlier step — by an allow rule or a mode like `acceptEdits` or `bypassPermissions` — so use a `PreToolUse` hook for logic that must apply to every tool call. The callback receives the tool name and input parameters. You return `allow` (with original or modified input) or `deny` (with an explanatory message). This enables dynamic control — sandboxing paths, blocking dangerous commands, or requesting user confirmation through your UI. It is skipped in `dontAsk` mode. 🛠 How to Use ```python # Python - Tool approval callback from claude_agent_sdk import ClaudeAgentOptions, query from claude_agent_sdk.types import PermissionResultAllow, PermissionResultDeny, HookMatcher async def can_use_tool(tool_name, input_data, context): print(f"\nTool: {tool_name}") if tool_name == "Bash": print(f"Command: {input_data.get('command')}") # Block rm commands and suggest archiving if "rm" in input_data.get("command", ""): return PermissionResultDeny( message="Please archive instead of deleting" ) response = input("Allow this action? (y/n): ") if response.lower() == "y": return PermissionResultAllow(updated_input=input_data) return PermissionResultDeny(message="User denied this action") # Dummy hook (required for Python streaming mode) async def dummy_hook(input_data, tool_use_id, context): return {"continue_": True} ``` ```typescript // TypeScript - Modify input before allowing import { query } from "@anthropic-ai/claude-agent-sdk"; for await (const message of query({ prompt: "Clean up temporary files", options: { canUseTool: async (toolName, input) => { if (toolName === "Bash") { // Rewrite paths to sandbox const sandboxed = { ...input, command: input.command.replace("/tmp", "/tmp/sandbox") }; return { behavior: "allow", updatedInput: sandboxed }; } return { behavior: "allow", updatedInput: input }; } } })) { /* ... */ } ``` 🏗 Integration into Production Systems - Display tool approval dialogs in web UIs and return user decisions to the callback - Use `context.suggestions` to retrieve `PermissionUpdate` entries for persistent "always allow" rules - Auto-block dangerous operations (`rm`, `DROP TABLE`, etc.) and suggest safe alternatives in the deny message - Sanitize inputs (restrict paths, override parameters) to dynamically control agent scope 💡 Use Cases 🔒 Require approval for write commands to production databases 📁 Restrict file operations to specific directories via dynamic path rewriting 👥 Control tool access based on user permission levels in multi-user environments ⚠️ Caveats - In Python, `canUseTool` requires streaming mode and a `PreToolUse` hook configuration - In `dontAsk` mode, the callback is skipped; use hooks for security-critical restrictions - The callback pauses execution indefinitely until a response is returned; use the `defer` hook decision for long-running cases ✨ It's not "full automation" or "full manual" — canUseTool enables smart operation where humans decide only at critical points! #ClaudeAgentSDK# #AIAgent#

The turmoil at law firm Weil, Gotshal & Manges has led its leaders to wind down the law firm. Here’s what to know about how one of Wall Street’s most vaunted law firms reached this point.

Singapore, let's talk inference. Join the vLLM meetup with engineers from @inferact, @NVIDIA, @alibaba_cloud and the vLLM-Omni community: high-throughput prefill, RL, Blackwell, Vera Rubin, agent infrastructure and interactive world models. Nov 13 · 1–5:30 PM SGT · SQ Collective Free, with registration approval. Bring your serving questions. Meet the people building the stack.

🏆 The @Acurast × EchoHunt campaign has wrapped up! Thank you for creating, sharing, and helping new users discover Acurast. We’ll carry out a final review of the leaderboard. The Top 150 eligible creators after verification will share the 200,000 base:0xc5fed7c8ccc75d8a72b601a66dffd7a489073f0b prize pool based on their final Impact Share. Rewards are expected to be sent to registered wallet addresses on BSC in about a week. Thank you for your patience! 👉 View the current leaderboard

BNB Chain ranks #1# in tokenized equities across the board 🏆 Leadership across the entire market in the last quarter, from assets issued to the people holding them and the activity they generate.