Use ChatGPT Developer Mode as a local coding agent for your repo through MCP.
1.2k
Stars
114
Forks
1
Open issues
2
Contributors
AI Analysis
CodexPro is a local MCP server that connects ChatGPT's Developer Mode to a single repository, enabling ChatGPT to read files, search code, make scoped edits, run verification commands, and manage context handoffs—all without proxying through a hosted service. It serves developers who want to use ChatGPT as a coding agent scoped to their local codebase through official API channels, particularly those needing offline-capable or privacy-conscious local development workflows.
Inferred from signals mentioned in the README (tests, CI, type safety) — not a review of the actual code.
AI's overall editorial judgment — not an average of the bars above, can weigh other factors too.
ChatGPT Developer Mode bridge for local repos via MCP; 2-week-old project gaining traction in AI-assisted coding niche
CodexPro wraps ChatGPT's Developer Mode as a local coding agent for individual repositories using the Model Context Protocol (MCP). Users install a CLI, run setup in a repo, and point ChatGPT to a local MCP server to gain bounded access to file operations, code search, git inspection, and bash verification. Built for developers who want ChatGPT to act as a practical local coding assistant without proxying, rate-limit bypassing, or model resale. Adoption appears early-stage but showing momentum: 1,046 stars in 14 days, +30 in the last week, available on npm, with Chinese localization suggesting intentional international reach.
Project created 2026-06-16, extremely recent. Enters a crowded space of AI-assisted coding tools and workspace bridges (codex-proxy, devspace, codex-plusplus already established). README explicitly positions CodexPro as narrower and more opinionated than generic workspace bridges, with ChatGPT Developer Mode as the primary constraint and design center.
30 stars in 7 days on a project 2 weeks old suggests initial wave of interest, likely driven by ChatGPT's expanding Developer Mode adoption and growing demand for local AI coding workflows. Timing aligns with broader movement toward agent-based development. Growth rate is notable but typical for new open-source tools that solve a fresh, specific problem; early adopter velocity should not be extrapolated to sustained growth.
Adoption not verified. No case studies, testimonials, or documented organizational usage provided in README. npm package exists (codexpro), indicating distribution infrastructure is in place. Chinese README and website suggest awareness in non-English markets, but no evidence of production deployments. Early GitHub star velocity may indicate interest rather than adoption.
Based on README, appears to be a Node.js CLI that establishes a local MCP server, exposes bounded tools (file read/write/edit, code search, git inspection, bash execution with safety modes), and connects to ChatGPT Developer Mode via a stable URL (ngrok or Cloudflare tunnel). Likely uses MCP protocol for tool invocation; no details on how repo context (AGENTS.md, .ai-bridge) is parsed or prioritized. Architecture is inferred from README and feature table only; actual implementation not inspectable.
Not documented in README. CI workflow exists (GitHub Actions badge present) but no mention of test suite, coverage percentage, or testing methodology.
Last push 2026-06-30 08:09:43, same as current date analysis, indicating active development continuing into the present. No mention of unresolved issues, open PRs, or response times. Project is 2 weeks old; typical for new projects to show high velocity. Cannot assess long-term maintenance pattern from this timeframe alone.
ADOPT IF: you are a developer using ChatGPT Developer Mode regularly and want to extend it with local repo awareness, bounded file/code operations, and git inspection without a third-party proxy or SaaS layer. AVOID IF: you need production-scale multi-user collaboration, audit trails for compliance, or support for non-OpenAI models; CodexPro is explicitly single-workspace and ChatGPT-centric. MONITOR IF: you use ai-assisted coding but prefer alternative models (Claude, local LLMs) or need multi-agent orchestration — CodexPro is tightly coupled to ChatGPT Developer Mode and will not easily adapt to competing LLM surfaces.
Independent dimensions
Mainstream potential
4/10
Technical importance
6/10
Adoption evidence
2/10
- Tight coupling to ChatGPT Developer Mode API surface: if OpenAI changes app authentication, MCP integration, or deprecates Developer Mode, CodexPro loses its primary use case with little fallback.
- Early-stage immaturity: 2-week-old project with no documented production usage; unknown how it handles edge cases, error recovery, or large repos in real workflows.
- Adoption unverified: star velocity may reflect curiosity rather than genuine use; no case studies or public deployments confirm this tool is actually used at scale.
- Limited to single workspace per setup; no multi-tenant, team collaboration, or shared context; may frustrate teams or developers managing multiple projects.
- Bash execution model introduces security surface; README mentions 'safe verification commands' but does not detail actual restrictions, sandboxing, or how malicious ChatGPT outputs are prevented from causing damage.
If ChatGPT Developer Mode remains a stable, widely-used OpenAI product, CodexPro may consolidate a niche of solo developers seeking local AI coding agents; likely to remain a specialized tool rather than a category leader. If OpenAI deprecates or significantly changes Developer Mode, the project faces existential risk and may need rapid pivot or sunset. Growth trajectory will depend on whether early adopters report genuine productivity gains and recommend it.
Explore similar
Newsletter
Get analyses like this every Monday
Free weekly digest of the most interesting open-source discoveries.
Languages
Information
- Language
- JavaScript
- License
- MIT
- Last updated
- 2d ago
- Created
- 3w ago
- Analyzed with
- anthropic/claude-haiku-4-5
Stars over time
Contributors over time
Top 100 contributors only — repos with more will plateau at 100.
Open issues
Open pull requests
No open pull requests.
Similar repos
icebear0828/codex-proxy
Codex Proxy is a lightweight local proxy that bridges Codex Desktop's API to...
| Repository | Stars | Week Δ | Language | Score | Updated |
|---|---|---|---|---|---|
|
rebel0789/codexpro
this
|
1.2k | +148 | JavaScript | 8/10 | 2d ago |
|
icebear0828/codex-proxy
|
1.4k | — | TypeScript | 7/10 | 3w ago |
 Waishnav/devspace
|
3.1k | — | TypeScript | 8/10 | 3d ago |
 b-nnett/codex-plusplus
|
3.5k | — | JavaScript | 7/10 | 1mo ago |
 openai/codex
|
96.9k | — | Rust | 8/10 | 1h ago |
Also uses ChatGPT/Codex but positioned as a proxy layer. CodexPro is explicitly not a proxy; it bridges MCP directly to a local server. codex-proxy has 370 more stars and ~4x CodexPro's age apparent from fork count, suggesting more established adoption.
Broader workspace automation platform. CodexPro is narrower and ChatGPT-first. devspace likely addresses a wider use case but not specifically for ChatGPT Developer Mode local coding loops.
Similar star tier to CodexPro but 3x larger. Likely an earlier or broader implementation of Codex-based workflow. CodexPro is younger and more opinionated about ChatGPT Developer Mode specificity.
Official Codex reference; orders of magnitude larger because it is OpenAI's own repository. Not a competitor in the traditional sense, but context for ecosystem scale.
README explicitly differentiates CodexPro from 'generic workspace bridges' by claiming tighter ChatGPT integration, explicit safety modes, and repo-backed context. This suggests awareness of a broader category; CodexPro claims to be more opinionated and less flexible.
