プラグインランキング

QAMap

Trace PR risk before merge

Browser Use

RONALD FERRARI SOARES

Guides browser automation, website QA, browser-use code review, and hosted-browser architecture using approved capabilities already available in the host. It does not bootstrap external runtimes or defeat access controls.

FlightDeck

NorthGate Strategic LLC

Audit code, packaging, security, privacy, public claims, reviewer cases, and submission gates without turning missing evidence into a pass. FlightDeck returns a clear ship decision, reproducible findings, executed checks, bounded fixes when requested, and every remaining gap.

CastReader

CAST AI PTE. LTD.

Read English Codex answers aloud in the right-side CastReader page, preserving headings, lists, tables, links and code while spoken words are highlighted. Try it in a new task: ask Codex to write a short English answer and read it aloud, or read an existing answer in the same task. Sign in to your existing CastReader account for 20 free AI reading minutes per day or unlimited ordinary reading with Pro. Audio is prepared ahead for continuous reading, with pause, resume and paragraph navigation. No developer API key is needed for answer reading. Separately, build runnable voice features and generate MP3/WAV files with CastReader Voice API estimates and resumable jobs. Developer API credit is separate from consumer membership. This answer-reading release supports English and highlights the formatted reading page.

heyGRC

BETTER ISMS

heyGRC reviews each pull request against the frameworks your company must comply with, grounded in your company profile. It posts inline findings, a neutral check run, and one sticky summary. This plugin does not review code locally. It walks you through installing the heyGRC GitHub App so reviews run server-side. Public repositories are free.

codex-sdlc

vimihi

codex-sdlc is a software development tool for developers and engineering teams using Codex. It helps turn feature requests into requirements, implement backend, web and mobile code changes, run tests, and review acceptance criteria in Git repositories. It stores project configuration, task history, verification evidence and reports in the project's .sdlc folder so development can resume across sessions. Choose Full delivery or Compact for bounded features, and configure the model used for each delivery role. The plugin provides Codex skills and uses the codex-sdlc npm CLI to operate the repository-local workflow. It requires an existing application repository and Node.js. Final acceptance remains with the user.

MCP Boundary

Faye & Cove

A pure-skill engineering plugin that traces active callers and entrypoints, implements or migrates the real MCP path, reconciles source with package and activated runtime, and keeps host claims inside qualifying observations.

Simulator Login

Reliable simulator sign-in

Migrating to React Native

Move Android and iOS to RN

Review Times

See app store review times

microCMS

CMS設計・実装・Next.js連携・WP移行

Mac Sign

ailuntz

直接告诉 Codex 要签名的项目或 App，自动识别产物、复用本机开发证书、签名并严格校验。可指定签名输出位置，或通过项目配置重建后签名。临时副本校验通过后才替换目标，保留原有权限声明。需要本地 Mac 终端和系统钥匙串中的有效证书；不上传 App 或私钥，不代替系统首次授权或公证。

Flower

상빈 박

Java workflows and governed actions that coding agents can build, verify, and maintain. Flower is a lightweight in-JVM workflow runtime for explicit, testable Flow and Step transitions, including persistence, Spring Boot integration, runtime observability, Flower Check, Flower Flow Graph, and local read-only Flower Studio inspection. Flower Action Runtime applies policy, approval, idempotency, durable run state, and audit to business mutations. This plugin helps Codex set up, build, inspect, review, migrate, test, and verify either library or both together. Support and bug reports are handled in the Flower GitHub issue tracker.

OneGate

OneGate Codex Plugin

CrowdStrike Falcon Fusion

CrowdStrike

Discover live Falcon Fusion actions, author workflow YAML with schema validation, import and release workflow definitions to a CID, trigger and monitor executions, and manage Falcon Next-Gen SIEM lookup files.

CompText Benchmark

CompText Labs

Run, inspect, and compare Raw versus CompText experiments while keeping quality and efficiency separate and preserving evidence.

ECZ-ID Agent Trust

EcoCitizenz Ltd

Review what AI agents in a workspace can reach: agent manifests, instructions, declared tools, reachable MCP servers, permission policy and ECZ-ID public proof references. Inspection only. It helps identify declared reach, authority-related evidence and review gaps without turning the result into a trust or safety score.

EW AI Power Digital Twin

Embodied Worker

EW AI Power Digital Twin converts electrical requirements and drawings into a downloadable draft JSON model. It separately reports contract validation and drawing-audit gaps for review in the AI Power Digital Twin Dynamic Simulation System. / 將電氣需求與圖紙轉為可下載的草稿 JSON，分別報告契約驗證與圖紙還原缺口，供使用者在 AI 電力數位雙生動態推演系統中核對。

JSON Change Lens

ailuntz

Understand what changed between two JSON files. JSON Change Lens compares configuration snapshots and API responses, ignores object-key order, distinguishes type changes, and preserves decimal precision. Match arrays of objects by a unique ID field to avoid reorder noise, ignore explicitly selected timestamp keys, and generate Markdown or JSON reports with paths into both original files. The helper reads two selected files and makes no network requests; reports are written only to a new output path when requested. Requires Python 3.9+ and a terminal or code execution environment. Host chat and execution-environment data handling still applies. It does not apply patches or determine API compatibility without a supplied contract. Independent community software.

CRIADOR / Reef

RONALD FERRARI SOARES

Guides structured evaluation, reflection, and iteration for improving agent behavior, prompts, workflows, and reliability.

Blink

Blink

Blink turns a description into a working app, with a database, sign-in, a backend and hosting, and publishes it to a live link. You do it all from this conversation. What you can do:• Build an app by describing it: a client portal, a booking site, an internal dashboard, a landing page or a mobile app.• Change it by asking: "add a dark mode toggle" or "let clients approve deliverables".• Answer Blink's questions right in the chat when it needs a decision, like which kind of app or which look.• See the finished app right in the chat, with a screenshot and what changed, then publish it to a live URL.• Look inside any project: list your projects, read the source at any saved version, compare two versions, query the database, and read backend logs when something fails.• Restore an earlier version if a change went wrong. How it works: connecting is a normal sign-in, and you choose which Blink workspace to grant. Code changes always go through Blink's agent, so your app's source and its saved versions stay in step. You can disconnect at any time from Blink Settings, Access Tokens, Connected apps.

Embrasure

Embrasure

Your data warehouse, analytics, and beautiful dashboards, one prompt away Get started in 30 seconds. Tell Codex what you want to know, connect your data, and let Embrasure set up your warehouse. Designed for billions of rows, so you can start small and keep growing.

Bubble

Bubble

Connect ChatGPT to your Bubble apps and build by describing what you want. ChatGPT can read and edit your app (pages, workflows, data types, option sets, styles, and records) and make changes the same way the Bubble editor does.

Bird

Bird

Bird is the communications infrastructure AI agents operate. Every agent eventually has to reach a human — verify a sign-up, confirm an order, chase an invoice, answer an inbox, place a call. Connect Bird and Codex does all of it for real: email, SMS, and WhatsApp on Bird's own sending infrastructure and carrier connections, calls on Bird SIP trunks, live events to every open browser and app over Bird Realtime — with a delivery timestamp or the exact failure reason instead of a guess. And Codex doesn't just send through Bird — it can own the endpoints. It claims its agent a durable email address in one call and answers what arrives there. It buys a phone number, gets it verified for texting, and registers the 10DLC campaign behind it. It sets up a SIP trunk and proves the caller ID with a real verification call. Addresses, numbers, trunks, and the compliance behind them — acquired in conversation. Whatever your account can do, Codex can do here. The tools are generated from Bird's live API — the same surface Bird's own dashboard runs on — so the list Codex shows is always the current one and grows as Bird does. Some of it works before you even connect: grade any domain's live email authentication, analyze headers and DMARC reports, or search Bird's docs, no account needed. You sign in with Bird when you connect, so there is no API key to paste and no secret sitting in your config.toml. Codex gets only the permissions you already hold, and you can revoke them from your Bird profile. A good first ask: "Send me a test email and tell me the moment it's delivered." Building an agent? "Claim yourself an email address and reply to whatever arrives."