"Actual Budget" matching MCP servers:

MCP Guard Server - Budget control, approval workflows and audit logging for AI agents (Claude Code, Cursor, ChatGPT)

Enables AI agents to ask what changed in a time window and get back a short, ranked answer across metrics, logs and traces, with the service where a failure started identified and its echoes in other services labelled as such. Read-only and deterministic, it scores changes against a baseline and trims results to a token budget, so agents avoid paging through raw queries and blaming the loudest symptom.

Persistent memory engine for AI coding agents. Single Go binary, zero runtime dependencies, MCP-native. Stores, searches, and deduplicates memories across sessions using embedded SQLite with hybrid FTS + semantic search, memory decay, relation graph, and token-budget context assembly.

Control-plane proxy that sits in front of MCP servers, enforcing per-identity policy (YAML/OPA/Cedar), budget limits, and an audit trail on every call. Includes real-time anomaly detection that auto-blocks a compromised agent without a human in the loop.

Enables AI clients to look up real Danbooru tags — exact names, aliases, categories, post counts, co-occurrence, wiki text, and the tag lists of actual posts — before composing image prompts. This tag verification noticeably improves prompt accuracy for local AI image generation.

Stop runaway agents. One gateway in front of all your agent traffic — MCP tools and LLM models — one budget, one kill switch. Sits between AI clients (Cursor, Claude Desktop, Windsurf) and remote MCP servers / OpenAI-compatible model APIs; hard token budgets; a single static Go binary.