CodePrism reads source code, stores a graph of it on your machine, and exposes that graph to AI agents through an MCP server. We take reports about it seriously.

CodePrism is pre-1.0 and moves quickly. Only the latest release on PyPI receives security fixes. If you are on an older version, upgrade first (pip install -U codeprism-ai) and check whether the problem still exists.

Once a 1.0 release exists, this table will list the supported minor versions.

Please do not open a public GitHub issue, discussion or pull request for a security problem.

Report it privately by either of these routes:

1. GitHub private vulnerability reporting: the Report a vulnerability button on the Security tab, if it is shown.
2. Email: krishnatyagibest321@gmail.com with the subject CodePrism security: <short summary>.

Please include:

• the CodePrism version (pip show codeprism-ai), Python version and operating system;
• a description of the problem and its impact;
• steps to reproduce, ideally a minimal script or sample repository;
• any suggested fix or mitigation, if you have one.

Do not include real secrets or private source code in a report; a synthetic example is enough.

This is a volunteer-maintained project, so these are targets, not guarantees. If you hear nothing after the acknowledgement window, please send a reminder.

If the report is accepted, we will fix it in private, release a patched version, publish a GitHub security advisory and credit you in the release notes unless you prefer to stay anonymous. If it is declined (for example because it is out of scope or not reproducible), we will explain why.

Please give us a reasonable chance to release a fix before you share details publicly. We ask for up to 90 days from your report, and we will tell you if we need longer. We are happy to agree a shorter or earlier date with you when a fix is ready sooner, or when the issue is already being exploited.

We will not take legal action against anyone who reports a vulnerability in good faith, follows this policy, and avoids harming other people's data or systems.

In scope

• The codeprism-ai package: the CLI, the Python API, the parsers and indexer, the query layer.
• The MCP server (codeprism serve, stdio and SSE transports) and its tools.
• The security gate and scanner (for example a way to make a BLOCK finding pass undetected is a vulnerability in a feature we advertise).
• codeprism setup and the configuration files it writes.
• Our release and CI pipeline (.github/workflows), including how the package is published.

Out of scope

• Vulnerabilities in third-party dependencies, unless CodePrism uses them in a way that makes them exploitable. Please report those upstream (we still welcome a heads-up).
• Findings that need an attacker who already has local access to your account or can already edit your CodePrism data directory or .codeprism.toml.
• Missing security detectors, false positives and false negatives in the scanner. These are bugs, not vulnerabilities; please open a normal issue.
• Denial of service from indexing an enormous or adversarial repository on your own machine.
• Reports from automated scanners without a demonstrated impact.

These are documented design choices, not vulnerabilities, so you do not need to report them. They are worth knowing about when you deploy CodePrism.

• SSE transport has no authentication and binds to 127.0.0.1 only. Do not expose it to a network without a reverse proxy that authenticates requests. Remote binding and authentication are tracked in #45.
• The index is a local database of your code's structure. It is stored in your user data directory, not encrypted, with the permissions of that directory. Treat it like the source it was built from.
• codeprism setup edits agent and editor configuration files in your home or project directory. Some editors' files are currently not handled safely; see #43.
• MCP tools run with your user's permissions. An agent connected to CodePrism can read anything the index covers and can ask it to record writes. Only connect agents you trust.

Testing against your own installation or a local test project is fine. Please do not test against other people's systems, and do not run load tests against infrastructure you do not own.

For general questions about this policy, open a regular discussion or issue that does not contain vulnerability details.