SkillFoxx

Prismor CLI

Runtime security for Claude Code, Codex and Cursor: tool call policies, observe and enforce modes and a local dashboard.

  • 487 ★ on the repository
  • Last commit
  • Apache-2.0
CLI

Extended access

Why this level

  • Prismor attaches hooks to every agent call and edits its configuration.
  • In enforce mode it can block calls and interrupt the agent.
All reasons and checks

prismorsec/prismorAuthor:PrismorSec

Install

In your terminal, with SkillFoxx CLI

npx skillfoxx add cli/prismor

Detects the agents on your machine, shows the access level and pins the version.

Other ways to install

Access and checks

How we review

Extended accessThe tool runs code or makes network calls and only reads external data.

  • Prismor attaches hooks to every agent call and edits its configuration.
  • In enforce mode it can block calls and interrupt the agent.

This is third-party code. Review the repository files before installing.

What it does

Prismor installs hooks into an agent and evaluates every tool call against a policy: shell commands, secret reads, network egress, package installs. In observe mode it only records activity; in enforce mode it blocks a call or asks a human to approve it. Capabilities listed in the README include a skill and MCP server scanner, an MCP gateway that checks every `tools/call`, secret leak prevention, network isolation, rules for MCP tools and a local session dashboard. Supported agents are listed in `AGENT_INTEGRATIONS.md`.

Who it is for. Security teams and developers who give agents access to a shell, secrets and MCP and want to see and limit what they do.

Good fit when

  • An agent works with `.env` and keys and you must intercept secrets leaving the machine.
  • Before running an unfamiliar skill or MCP server you want to scan it.
  • Some MCP tools should run only after human approval.
  • An organization collects a session-by-session log of agent actions.

Not a fit when

  • You need kernel-level protection: Prismor judges agent calls, not OS processes.
  • The agent's own permissions and manual command approval are enough.
  • The agent runs on a platform where hooks cannot be attached.

Example request

Install Prismor, run `prismor setup` in observe mode and show which calls the agent made in the last session.

Limitations

Prismor intercepts calls through agent hooks and the model proxy, so depth of protection depends on what a given agent supports; see the matrix in `AGENT_INTEGRATIONS.md`. Enforce mode blocks nothing at first, and you pick the blocking rules yourself. Optional layers such as the LLM semantic guard, PII detection and the enterprise control plane are enabled separately. An agent may edit policy only inside the `prismor unlock` window, which is password-gated and time-boxed.

How to disable. Run `prismor pause` to stop blocking, or remove the Prismor hooks from the agent configuration and the `prismor` package from the Python environment.

FAQ

How do I start Prismor quickly?

Run `pip install prismor` and `prismor setup`. The README promises a 30-second start.

How does observe differ from enforce?

In observe Prismor records activity and blocks nothing. In enforce it blocks calls by the rules you select or asks for approval.

What does the MCP gateway do?

One MCP connector fronts all other servers. Each `tools/call` is policy-checked and each response is injection-scanned before the model sees it.

Does Prismor work with Codex and Cursor?

Yes, the README lists Claude Code, Codex CLI, Gemini CLI, Cursor, GitHub Copilot, OpenCode and other agents.

How do I stop blocking temporarily?

With `prismor pause`: blocking stops while activity logging continues.

Discussion

No questions or reviews yet.

CLI tools

An MCP memory server for agents plus a free gate that stops dangerous commands before they run

MCP serverBroad access660

A guard on an AI agent's execution path: every tool call gets a PASS, AUTH or BLOCK verdict before it runs

CLIBroad access243

A policy gateway for AI agents: it checks tool calls before they run, holds risky ones for approval and writes an audit trail

CLIExtended access523

Vigils sits between an AI agent and its tools: it hides secrets, asks for approval and keeps a tamper-evident log of calls

CLIBroad access74

AgentShield audits the .claude folder for plaintext keys, broad permissions, dangerous hooks and risky MCP servers and grades the setup from A to F

CLIExtended access1.4K
Official

Gryph hooks into Claude Code, Cursor, Codex and other agents, logs every action to a local SQLite database and can block risky commands.

CLIBroad access183

README in short

The README outlines the threats to agents, a quick start and a capability list linking to a document for each.

README badge

Report a problem