Tirith

Tirith: Terminal security software that intercepts commands and pastes to detect homograph attacks, injection, and malicious payloads before execution.

Tirith is a terminal security tool designed to protect developers and AI agents from command-based attacks. It functions by intercepting shell commands and clipboard pastes, analyzing them through a three-tier pipeline that includes regex filtering, entity extraction, and a rule-based engine. The tool monitors for threats such as homograph attacks, pipe-to-shell patterns, ANSI injection, credential exfiltration, and supply-chain vulnerabilities. It also secures AI coding agents by scanning configuration files, plugins, and MCP tools for prompt injection and unauthorized permission changes. Tirith runs entirely locally, ensuring data privacy, and integrates with editors via an LSP server for inline diagnostics.

Tirith at a glance

Best for: Software engineer
Runs on: macOS, Linux, Windows, CLI
Pricing: Community: Free; Team/Enterprise: Contact for pricing.
Released: February 2026
Category: AI coding tool
Website: tirith.sh

Key features

Detection of 200+ threat rules across 29 categories
Real-time interception of shell commands and clipboard pastes
AI configuration scanning for prompt injection and hidden instructions
Local-first execution with no required network calls
LSP integration for inline editor diagnostics
Support for MCP server protocols for AI agents
Audit logging in JSONL and SARIF formats

Use it for

Blocking malicious pipe-to-shell commands
Detecting credential exfiltration from environment variables
Scanning AI configuration files for prompt injection
Verifying plugin and skill safety for AI coding agents
Auditing terminal activity for supply-chain threats

Alternatives to Tirith

ARTEX — Runs autonomous multi-agent penetration testing with a dual-graph architecture.
Tusk — Generates tests from production traffic to catch regressions in PRs.
Promptfoo — It simulates real users to uncover agent and RAG vulnerabilities.
jevgrep — Reduces coding agent costs by providing targeted, relevant repository context.
abtop — Monitor multiple AI agent sessions and detect orphan ports from a single screen.
Vigil — Drives end-to-end security engagements with specialist agents and a shared Kali container.
Stroq — Scans recorded sessions and guards future actions with local hooks.
PentAGI — It autonomously manages penetration testing workflows within a secure Docker sandbox.
Mantis — Automates vulnerability discovery and patching using AI agents.
CyberStrike — Autonomous AI agents execute recon, web, network, and cloud penetration testing.
VulnClaw — It uses LLM agents to autonomously execute end-to-end security testing loops.
Aegis Kernel — Stops prompt injection and tool-hijacking at the OS syscall boundary.

All Tirith alternatives