Tag: agent-safety
Pre-action gates for preventing agent mistakes
This skill implements pre-tool-use hooks to intercept and block known bad agent actions, such as repeated hallucinations or destructive calls. It saves tokens and prevents the model from executing mistakes that have been previously identifi…
Pre-action gates for AI agent safety
ThumbGate implements pre-action gates for AI agents, capturing structured thumbs-up/down feedback to generate prevention rules from repeated failure patterns. It intercepts tool calls via the MCP PreToolUse hook, blocking known-bad actions …
Configure agent spending limits and safety controls
This skill allows configuring critical spending guardrails, including per-transaction and daily send limits, for autonomous agent wallets. It provides a kill switch mechanism to lock operations, ensuring robust financial safety before enabl…
Agent Safety Gates and Failure Pattern Prevention
ThumbGate implements pre-action gates for AI agents, capturing structured thumbs-up/down feedback on tool usage. It automatically generates prevention rules from repeated failures, blocking known-bad patterns before execution via the MCP Pr…
Pre-Action Gates for Preventing AI Mistakes
ThumbGate intercepts agent tool calls via PreToolUse hooks, capturing negative feedback to synthesize prevention rules. This mechanism physically blocks known bad actions or repeat hallucinations before they consume tokens or execute destru…
Agent Safety Gates for AI Coding Agents
ThumbGate implements pre-action gates for AI agents, capturing structured feedback (thumbs up/down) to identify failure patterns. It automatically generates prevention rules and blocks known-bad tool calls via the MCP PreToolUse hook, signi…
PII Detection and Redaction Skill
A zero-dependency Python implementation for detecting and redacting personally identifiable information from text files. It replaces sensitive patterns such as SSNs, API keys, and email addresses with standardised placeholders to ensure age…