What is shellfirm?

shellfirm is a terminal safety tool that intercepts dangerous shell commands and makes you confirm before executing. Think of it as a captcha for your terminal.

How it works

Every command you type passes through shellfirm's check pipeline:

1. Shell hook intercepts the command before execution
2. Pattern matching checks against 100+ regex patterns across 9 ecosystems
3. Context detection evaluates your environment (SSH, root, git branch, k8s)
4. Policy merge applies team rules from .shellfirm.yaml
5. Blast radius computes the real-world impact of the command
6. Challenge prompt asks you to confirm — or blocks the command entirely

Key features

• 9 ecosystems — filesystem, git, Docker, Kubernetes, AWS, Azure, GCP, Terraform, databases
• Context-aware — escalates protection on SSH, root, production branches, k8s clusters
• 3 challenge types — Math, Enter, or Yes (auto-escalated by severity and context)
• Blast radius — shows real impact before you confirm ("Deletes 347 files, 12.4 MB")
• AI agent safety — MCP server for Claude Code, Cursor, and any MCP-compatible agent
• Team policies — .shellfirm.yaml in your repo for shared rules
• Audit trail — JSON-lines log of every command and decision
• Interactive wrapper — protect live database sessions (psql, mysql, redis-cli)
• 8 shells — Zsh, Bash, Fish, Nushell, PowerShell, Elvish, Xonsh, Oils

Who is it for?

• Individual developers — protect yourself from costly mistakes
• DevOps / SRE teams — enforce safety policies across your organization
• AI agent operators — ensure Claude Code, Cursor, and other agents can't run destructive commands unchecked
• Security & compliance teams — audit trail for every command, SOC2-friendly logging