Piccolo: 4B, 9B and 27B decision models for agent tool review, context retention and data workflows, with an interactive Arena.
-
Updated
Nov 22, 2026 - Python
Piccolo: 4B, 9B and 27B decision models for agent tool review, context retention and data workflows, with an interactive Arena.
PuPu is a lightweight, cross-platform desktop AI client that works with both local and cloud-hosted models. Whether you prefer running models on your own machine or connecting to providers like OpenAI and Anthropic, PuPu gives you a unified, elegant interface — your AI, your rules.
Policy-aware AI permission reviewer for OpenCode V1 and V2. Evaluates tool and command requests, allowing, denying with feedback, or escalating to the user.
Review what your agent changed and approve what it may do: React components for multi-file diff review and tool approvals that return a result your agent acts on, for AI SDK 6 & 7, AG-UI and ACP agents, inside assistant-ui, AI Elements or your own chat.
Self-hosted WeChat bridge for Claude Code and Codex, with Claude tool approval and bidirectional file relay.
Runtime guardrails for AI agent MCP tools. Blocks dangerous tool calls at runtime with a default-deny policy proxy.
Open-source MCP gateway with per-agent tool permissions. Give Claude Code, Codex and Cursor one endpoint each that exposes only the tools you allow, require your approval for chosen tool calls, and review changed tool definitions before agents see them. Run locally with npx toolstile server start or use toolstile.com.
Let OpenCode run unattended, securely. An isolated System One or LLM reviewer answers permission prompts. Risky calls go back to the agent with a reason, and unclear ones come to you.
Runtime risk guard for OpenClaw: hook-based tool interception, approval workflows, prompt injection defense, and auditable local security controls.
The form renderer for MCP tools, A2A agents, and human-in-the-loop AI. One component. Any JSON Schema. Beautiful forms. Under 1ms.
Minimal shell over Claude: Blazor UI + Claude Agent SDK sidecar with browser-based tool approval (Allow/Deny on every tool call). No injected prompt, no governance - fork it to build your own workflow.
Minimal reproduction for Vercel AI SDK historical Tool Approval and No tool invocation found resume failures
过目 · Second Look 是 DeepSeek Harness 的低打扰工具审核插件,适合习惯用「完全权限」的人。选用「日常审查」后,读项目文档、在项目里新建普通文件通常直接放行,整份覆盖已有内容的文件则先由独立模型审核。
Provider-free before/after reproduction for Vercel AI SDK WorkflowAgent signed tool approvals and replay enforcement.
Provider-free OpenAI Agents SDK RunState lab for tool approval, cross-process resume, rejection, idempotency, and streamed-resume persistence.
Live Filament inbox for Laravel AI tool approvals: your agent pauses before a risky call, your team approves, edits or rejects it, on Hibira
[READ ONLY] Tool approval guard middleware for Laravel AI agents.
Three-tier intelligent permission gate for AI agents — automated approval with hash-based trust, AI evaluation, and full audit ledger
AI-powered automatic tool-call review for Oh My Pi (OMP). An independent AI reviewer approves or blocks risky actions before execution.
To associate your repository with the tool-approval topic, visit your repo's landing page and select "manage topics."