Impact previews
Shows the expected consequence of a command, including affected files, PostgreSQL rows and dependent tables, or commits a Git force push would lose.
Termaxa
Freemium
A cooperative gate that previews, evaluates, and records shell commands run by AI coding agents.
Termaxaとは？
Termaxa is a developer tool that acts as a cooperative gate for shell commands and selected file operations run by AI coding agents. It previews the likely consequence of an action, applies a policy, and returns an allow, ask, or deny decision before execution. Its examples include affected files, database rows, and commits that a force push would remove.
The product is intended to work inside agent workflows rather than beside them. It provides hooks for Claude Code, Codex, Cursor, and Copilot, and records decisions in a location outside the project repository. For selected destructive operations, it can create backups before the action runs.
Termaxaでできること
Policy-based decisions
Evaluates commands against project rules and returns an allow, ask, or deny decision. The default policy blocks examples such as recursive force deletion.
Backups before selected operations
Can copy affected paths to `.termaxa/backups` before deletion and use database dumps before PostgreSQL changes, providing rollback support for supported workflows.
Agent and native-tool hooks
Integrates with Claude Code, Codex, Cursor, and Copilot. Supported event handling includes shell commands and, in relevant integrations, Write, Edit, Delete, and patch file headers.
Receipts and replay
Keeps decision records outside the repository in `~/.termaxa` and provides `termaxa replay` to evaluate commands agents have already run.
利用シーン
“Review destructive file commands”
Use the gate to inspect commands such as `rm -rf` before execution. The decision can include the target path, file count, and the reason a policy blocked the operation.
“Protect Git history”
Preview what a force push would remove before sending it to a remote, rather than relying only on the command text or a generic approval prompt.
“Stage database changes”
Run a PostgreSQL change through a workflow that previews affected rows and dependent tables and creates a `pg_dump` before the operation when supported.
“Inspect infrastructure changes”
Place Termaxa in front of Terraform apply workflows so the plan and policy decision are available before infrastructure changes are executed.
よくある質問
Which agent tools does Termaxa integrate with?
The site lists hooks for Claude Code, Codex, Cursor, and Copilot. Integration behavior differs: for example, Codex is documented as honoring deny decisions through its hook, while Claude Code can also be reached through a configured shell wrapper.
Can Termaxa inspect commands without running them?
Yes. `termaxa check "<command>"` reads a command and returns a decision, reason, and contextual impact without executing it.
Where are Termaxa’s records stored?
The site says state was moved out of the repository to `~/.termaxa`, so a repository reset cannot remove the audit records. The records contain the gate’s decisions and related context.
Does Termaxa automatically allow every command?
No. Commands are evaluated against policy and may be allowed, asked about, or denied. The default policy example denies `rm -rf ./scratch` because it detects the recursive force-delete flags.
クイック情報
- Category
- Developer Tool
- Primary users
- Teams and individuals using AI coding agents
- Integrations
- Claude Code, Codex, Cursor, and Copilot
- Installation evidence
- Prebuilt binaries; the site documents Homebrew installation and project commands such as `termaxa init` and `termaxa doctor`
- Platforms listed
- Linux x86_64, macOS arm64, macOS x86_64, and Windows
- Command-line checks
- `termaxa check` evaluates a command; `termaxa replay` reviews commands already run
Termaxa のトラフィック分析
トラフィックデータは参考情報としてご利用ください。
- ドメイン評価
- 3
Termaxaの代替品
Dezycro
app.dezycro.ai
Dezycro is an operations layer for AI coding agents that provides product context, steers work, verifies behavior against requirements, and records decisions. It is designed for developers, product teams, and organizations that want to use agents without giving Dezycro access to source code.
Agentic Integrity Foundation
harden.run
Agentic Integrity Foundation is a local endpoint-security product for coding agents. It evaluates supported tool calls before execution and can allow, approve, make safe, block, or record actions on a developer’s machine.
Guardrail by NeatProxy
www.neatproxy.com
Guardrail is a local spend firewall for AI coding agents. It records usage metadata and, on supported proxy integrations, checks budget policies before forwarding requests to providers.
Relace
relace.ai
コーディングエージェント向けに、コード検索・ファイルマージ・ソース管理ワークフローを支援するAIモデルおよびインフラストラクチャプラットフォーム
Zoku
heyzoku.com
Zoku is a native macOS app for running and coordinating multiple coding-agent CLIs on one canvas. It lets developers spawn, redirect, monitor, and interrupt named agents by voice while using the subscriptions and command-line tools already installed on their Mac.
HAR HQ
harproject.dev
HAR HQ is a team governance and observability layer for organizations running coding agents with HAR. It standardizes verification, tracks AI-related work and spend, and helps engineering teams review and improve agent workflows.