AI & Agents MCP servers

Local-first automatic memory for Claude Code and Codex: capture, distill, recall from SQLite.

Working memory for AI agents: digest-indexed episodes on a git shelf, recalled a section at a time.

59 pay-per-call AI services as MCP tools with automatic x402 USDC-on-Base payment.

Capability graph of your agent stack: what it can do, what to set up next, what breaks.

AI BVF: review AI proposals with Stop, Fix or Accelerate, evidence gaps and next actions.

Project context for Cursor/Claude via stdio MCP. Key: www.neither.online/start

Your agent's overlay network. 435 specialist agents + P2P A2A messaging on a 190k-node network.

Compile your own business capabilities from declarative YAML and serve them as MCP tools.

Your agents share what they learn. Semantik finds it by meaning, with no topic names to agree on.

Goal-oriented conversations and next-question selection through the Zelinqa API.

Cross-client Agent identity, DMs, groups and block lists. Local MCPB; Node >=24.15; user pairing.

Multi-model consensus: 2-6 frontier LLMs answer, an independent judge synthesises one answer.

Typed decisions with Jev / System One via OpenRouter or TypeSafe: classify, verify, rerank, decide.

Local-first intent and provenance for Codex, Claude Code, Cursor, Gemini CLI, and MCP agents.

Signed messages between AI agents. Dangerous actions wait for a human-signed approval.

Atlasent for AI agents: with an API key, actions wait for approval with signed proof. No key: demo.

Grade an AI agent's transcripts on 18 reliability tests. Thin evidence is NOT TESTED, not guessed.

Encrypted-first embedded database with vector search and agent memory, exposed as MCP tools

Keep the state. Change the model. Shared context and a multi-model panel across coding assistants.

Decision lock, anti-hallucination, meta-evolution and provenance for any MCP agent. 55 tools.

Search local AI conversations by file or phrase and recall source turns. Four read-only tools.

Local-first agent memory that preserves what happened, when and why

Routes each task to the few skills, MCP servers and tools that fit it.

Ask a human before your agent acts: approval requests decided on a phone.

Run Claude Code subagents on cheaper opencode models (DeepSeek, opencode Go) with native parity.

A second opinion from an unrelated AI model before you commit to a decision.

A project's long-term memory as plain Markdown files, with an append-only, tamper-evident log.

Deterministic runtime enforcement of step order for AI agents: ALLOW/DENY before a step runs.