Codex Delegate vs Ocular

A side-by-side look at the Codex Delegate and Ocular MCP servers: transport, auth, tools, quality signals, and the config to install either one in Claude, Cursor, Windsurf or VS Code.

Codex Delegate logoCodex DelegateCoding Agents · Local subprocess (stdio)Bridge AI coding hosts to the OpenAI Codex CLI for delegated implementation.63GoodAuth: not declaredOcular logoOcularCoding Agents · Local subprocess (stdio)Vision tools for coding agents: screenshots, OCR, UI diffs, errors, tables, and charts.60GoodAuth: not declared

Verdict: choose Codex Delegate for coding agents work as a local process; choose Ocular for coding agents as a local process. You can also run both side by side in one mcpServers block.

Which MCP server should you choose?

Codex Delegate logoChoose Codex Delegate when

  • You need dedicated capabilities in the Coding Agents domain.
  • You prefer a local stdio subprocess transport.
  • You want tools like delegate, cancel, doctor.
Explore Codex Delegate

Ocular logoChoose Ocular when

  • You need dedicated capabilities in the Coding Agents domain.
  • You prefer a local stdio subprocess transport.
  • You have the keys it needs: OCULAR_BASE_URL, OCULAR_API_KEY, OCULAR_MODEL.
  • You want tools like analyze_image, extract_text_from_image, analyze_ui_screenshot.
Explore Ocular

Feature & specification comparison

Rows where the two servers differ are marked.

SpecCodex Delegate logoCodex DelegateOcular logoOcular
CategoryCoding AgentsCoding Agents
Quality signal63/100 · Good60/100 · Good
Install commandConfirmedConfirmed
TransportLocal subprocess (stdio)Local subprocess (stdio)
AuthNot declaredNot declared
PricingBYOK (pay the provider)Not declared
Required env varsNone listedOCULAR_BASE_URLOCULAR_API_KEYOCULAR_MODEL
Verified ownerCommunity listingCommunity listing
GitHub stars20
Directory activity 6 views 0 installs 0 upvotes 1 views 0 installs 0 upvotes
Tools38

Tools & capabilities

Codex Delegate tools 3

  • delegateDelegate a coding task to OpenAI Codex. Never run codex from the shell — use this tool. Check status before trusting result — a run that spawns then fails returns normally. Keep model, reasoningEffort and fast at their defaults unless the user asks. See the delegate skill.
  • cancelCancel active runs. delegationId selects one (announced in progress before spawn — the only handle if a run wedges at startup), threadId all on its thread, neither all. Waits for settlement; if a process tree survives the kill deadline, it still returns and the delegate result warns. Status: cancelled, nothing-active, or not-found (no active run with this ID).
  • doctorDiagnose plugin/CLI/login, recursion, and workspace (default: server cwd). deep adds no-quota help and model-catalog checks.

Ocular tools 8

  • analyze_imageAnalyze a general image and return structured JSON for a coding agent.
  • extract_text_from_imageExtract OCR text from screenshots, documents, tables, terminal output, or code images.
  • analyze_ui_screenshotAnalyze a webpage, app, UI, or design mockup screenshot for frontend implementation work.
  • diagnose_error_screenshotAnalyze error, terminal, console, browser, or build failure screenshots.
  • compare_ui_screenshotsCompare two UI screenshots and return implementation-useful visual differences.
  • extract_table_from_imageExtract visible table data from an image as structured JSON, Markdown, or CSV-oriented output.
  • analyze_chart_imageAnalyze a chart image and return labels, trends, approximate values, and limitations.
  • create_upload_sessionReturns upload instructions for the binary side channel (use this to analyze LOCAL image files without base64 corruption). Before uploading, verify the local file exists and is non-empty with Bash (e.g. `ls -la <path>` or `test -f <path> && stat -c %s <path>`). Then PUT the raw bytes to the returned upload_url with curl --data-binary (NOT base64); the response gives a file_id to pass to any vision tool. Stateless -- call once to learn the endpoint.

Ready-to-paste client configs

Paste either block (or both) into your client config, e.g. claude_desktop_config.json or ~/.cursor/mcp.json. We only show a config when the install command is confirmed.

Codex Delegate

mcpServers (Claude Desktop / Cursor)
{
  "mcpServers": {
    "codex-delegate": {
      "command": "npx",
      "args": [
        "-y",
        "codex-delegate-mcp"
      ]
    }
  }
}

Ocular

mcpServers (Claude Desktop / Cursor)
{
  "mcpServers": {
    "ocular": {
      "command": "npx",
      "args": [
        "-y",
        "ocular-mcp"
      ],
      "env": {
        "OCULAR_BASE_URL": "YOUR_OCULAR_BASE_URL_HERE",
        "OCULAR_API_KEY": "YOUR_OCULAR_API_KEY_HERE",
        "OCULAR_MODEL": "YOUR_OCULAR_MODEL_HERE"
      }
    }
  }
}

Frequently asked questions

Codex Delegate is categorized under Coding Agents and uses a local stdio subprocess. Ocular belongs to Coding Agents and uses a local stdio subprocess. Pick Codex Delegate for coding agents work and Ocular for coding agents.

Copy the mcpServers block from the config section on this page into your client's config (for example claude_desktop_config.json or ~/.cursor/mcp.json), then fully restart the client.

Codex Delegate is bring-your-own-key: usage is billed by the API provider you supply a key for. Ocular has no declared pricing — check the project's own site. If a server needs environment variables such as API keys, define them under the env key in your client's MCP config.

Yes. MCP clients support multiple servers: add codex-delegate and ocular as separate keys in the same mcpServers object, and your assistant will call tools from either one.

On AllMCPs, Codex Delegate has a quality signal of 63/100 (Good) with 6 views, 0 installs and 2 GitHub stars. Ocular has 60/100 (Good) with 1 views, 0 installs and 0 GitHub stars.