CodeScope
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@CodeScopeInspect current Python project for reuse candidates."
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
CodeScope
CodeScope is a local-first MCP preflight system for developer tools. Its intended workflow helps GPT-5.6 and Codex inspect an existing Python repository before generating code, so they can make evidence-backed REUSE, EXTEND, or CREATE decisions and reduce duplicate logic.
Current implementation status
OpenAI Build Week Phase 1 is in progress. The repository currently provides:
a Python 3.12 package and version command;
immutable, validated TOML configuration;
immutable public data models and stable domain error codes;
Python-only language and extension validation;
centralized repository-file and future reset-target path guards;
focused unit and path-security tests.
The functional indexing and MCP preflight workflow are not implemented yet. In particular, this phase does not parse source code, create chunks or embeddings, persist an index, search code, expose MCP tools, or delete runtime data.
Related MCP server: Graft
Requirements
Python 3.12
A platform supported by the locked Python dependencies
Phase 1 has been validated in the current Linux development environment. Broader supported-platform claims and clean-clone verification are deferred until the functional MVP exists.
Setup
Clone the repository, then install the locked development environment:
uv syncThe default configuration is codescope.toml. Configuration paths are resolved relative to that file. Its indexing root must already exist; its storage directory may be absent until a later runtime phase.
Current operation
The only functional CLI behavior implemented so far is package version reporting:
uv run codescope versionThis is not yet the final judge testing path. Installation, indexing, MCP configuration, sample-data, and end-to-end judge instructions will be added only after those workflows are implemented and verified.
Testing
Run the current checks with:
uv run pytest tests/unit -q
uv run pytest tests/security -q
uv run ruff check .
uv run ruff format --check .
uv run mypy src/codescope
uv run codescope versionPhase-specific commands and observed results are recorded in BUILD_WEEK_CHANGELOG.md. Hackathon submission requirements are tracked in docs/HACKATHON_COMPLIANCE.md.
Sample data
No sample repository or indexed sample data is included yet because repository indexing is outside Phase 1. A safe, documented sample path will be added before submission and will not require judges to inspect unrelated host files.
Current limitations
Python repositories only; supported source extensions are
.pyand.pyi.No Tree-sitter parsing, symbol extraction, chunking, embedding, Chroma persistence, repository indexing, or search.
No MCP tools or MCP server workflow are operational yet.
No dashboard, remote hosting, authentication, deployment, or file watching.
Path validation reduces traversal and symlink risk, but filesystem state can change after validation; future read and deletion operations must validate immediately before use.
No benchmark or performance claim has been established.
Built During OpenAI Build Week
The repository distinguishes pre-existing planning from Build Week implementation through dated Git history and BUILD_WEEK_CHANGELOG.md. Work completed through Phase 1 comprises the package foundation plus validated configuration, public models, domain exceptions, Python language validation, path-security guards, and their focused tests. Later MVP functionality remains unimplemented and must not be inferred from planning documents.
How Codex and GPT-5.6 Were Used
Codex with GPT-5.6 was used in the primary implementation thread to inspect the Build Master and repository constraints, consult version-matched documentation, implement the current Phase 1 modules and tests, run validation, and review the working-tree security diff. The owner supplied and approved the product positioning, phase boundaries, architecture, security policies, evidence rules, and implementation contract.
This section records only completed work. The final acceleration narrative, demo claims, contribution summary, and /feedback Session ID remain pending until most core functionality is built. The Session ID will be obtained by the user running /feedback in the primary implementation thread; no value is invented here.
License
CodeScope is distributed under the repository's LICENSE. Third-party dependency licenses must be reviewed and documented before submission.
This server cannot be installed
Maintenance
Resources
Unclaimed servers have limited discoverability.
Looking for Admin?
If you are the server author, to access and configure the admin panel.
Latest Blog Posts
- Your AI Chatbot Just Exposed Your CEO's Salary to an InternBy Om-Shree-0709 on .Agent IdentityMCP SecurityOAuth Delegation
- Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)By Om-Shree-0709 on .Agentic AiPrompt InjectionWebAssembly
MCP directory API
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/Ibadat-Ali86/codescope-mcp-preflight'
If you have feedback or need assistance with the MCP directory API, please join our Discord server