Stop most hogging CPU process when CPUs are pegged

Catches situations when a react use(Layout)Effect runs repeatedly in rapid succession

Kill Switch CLI — monitor cloud spending, kill runaway services from the terminal

Supervised process/test runner — caps, culls, and reaps runaway processes (native binary)

Periderm CLI: A pre-launch checklist to catch user-losing bugs, legal liabilities, revenue leaks, runaway cloud bills and so many more before you ship.

Kill Switch for coding agents — stop runaway Claude Code / Cursor / Aider sessions from racking up an LLM bill. Native hook + token-metering proxy with per-session and daily-rolling budgets.

Local-first runtime safety layer for AI agents that blocks runaway costs, loops, retries, and budget overruns before API calls execute.

Track AI agent costs, tokens, runtime and spending. Prevent runaway OpenAI, Anthropic, LangGraph and OpenRouter agents from exceeding budget.

Keep AI coding agents from cooking your laptop — load/thermal guard that right-sizes runaway parallel commands. Pure bash, no deps.

Your command center for Claude Code and GitHub Copilot CLI sessions. Watch every session live, send commands, and stop runaway agents, all from a single browser tab.

Detects expensive code patterns — runaway Firebase reads, missing real-time listener cleanup, polling abuse, and React render loops — before they hit production.

AI cost protection that runs inside your app. Catch runaway loops before they bankrupt you.

YARI (Yet Another Repetition Intervention). Interrupts agent tool loops and runaway/stuck thinking.

A linter and governance layer for AI agent loops. Catches runaway, unverified, or unsupervised agentic loops before they cost you.

PromptOps observability for OpenClaw: per-call cost tracking, runaway loop detection, circuit breaker, and budget enforcement

The task state machine at the core of Sockt's anti-runaway design. A SQLite-backed `TaskStore` implementation, a strict finite state machine that enforces valid task transitions, an LLM-call budget guard, and an atomic claim lock for concurrent agent work

Detect when problem-solving degrades into unproductive loops, over-reflection, or runaway self-criticism, then stop and escalate to the user for help.

Detect repeated agent steps and stop runaway loops.

Free local proxy that prevents AI API cost runaway loops

Harness for a GitHub-issue-driven SDLC agent loop: spawn Claude Code headlessly to implement an issue as a draft PR, with runaway guards, preflight refusal, and verify-comment posting.