resonant
Resonant is an open-source, self-hosted “relational AI companion” framework that runs a local Node.js server and orchestrates interactions with the Claude Code Agent SDK. It provides persistent identity/session continuity (via Claude Code’s memory system), an orchestrator for autonomous routines/triggers, and multiple user channels (web UI, Discord, Telegram, voice with transcription/TTS).
Score Breakdown
⚙ Agent Friendliness
🔒 Security
README indicates self-hosting and local persistence (SQLite) with optional web UI password protection, but does not describe TLS enforcement details, secure secret handling for any configuration/.env, password hashing/storage, authorization model, or dependency/SBOM/CVE practices. Also, autonomy features and MCP tool integration increase the importance of least-privilege tool access and strict prompt/tool guardrails to prevent unintended data access/actions.
⚡ Reliability
Best When
You want a locally hosted, always-on or frequently running companion that preserves memory across restarts and can schedule or conditionally trigger actions using chat-accessible tools (including MCP-based tools).
Avoid When
You need a simple “call an HTTP endpoint and get a deterministic response” interface, or you can’t accept that an agent can be configured for autonomy (routines/impulses) and may require careful prompt/guardrail design.
Use Cases
- • Building a self-hosted AI companion that maintains long-term conversational context and identity continuity
- • Running autonomous check-ins and conditional routines (routines/pulses/watchers/timers/failsafes) that can act based on presence/time/context
- • Providing relationship-aware agent experiences using injected hooks/context and structured prompts (CLAUDE.md, wake prompts, skills)
- • Managing life data through an integrated “Command Center” that is accessible in chat and backed by MCP tools
- • Creating a multi-channel companion experience (web + chat platforms) with streaming tool-call visualization
- • Experimenting with MCP tool integration by wiring the companion to external MCP servers
Not For
- • Use as a managed SaaS API for serverless automation (it’s primarily a self-hosted companion app/server)
- • Environments that cannot run a long-lived local service (Node server, websocket connections, scheduler/orchestrator)
- • Teams needing a documented public REST API/SDK contract separate from the web UI/agent tooling
- • Compliance regimes requiring formally documented security controls beyond what the README describes
- • Cases where you need strict, guaranteed idempotency for agent-driven side effects without additional guardrails
Interface
Authentication
Authentication is described as relying on Claude Code login (no separate API key for Resonant itself). Web UI has optional password protection via resonant.yaml/settings, but the README does not specify detailed auth mechanisms (e.g., rate-limited login, password hashing method).
Pricing
Cost depends on Claude Code usage/subscription; README does not provide Resonant-specific pricing tiers or credit-card requirements.
Agent Metadata
Known Gotchas
- ⚠ Agent autonomy (routines/pulses/impulses/watchers) can cause repeated or unintended actions if prompts/conditions are misconfigured; add guardrails and cooldowns where supported.
- ⚠ Side effects through tools may not be inherently idempotent; agent retries (if any) could duplicate actions unless operations are designed to be safe.
- ⚠ Because it’s a companion infrastructure (web/socket/multi-channel), integration testing should account for message ordering/streaming behavior and state persistence across restarts.
Alternatives
Full Evaluation Report
Comprehensive deep-dive: security analysis, reliability audit, agent experience review, cost modeling, competitive positioning, and improvement roadmap for resonant.
AI-powered analysis · PDF + markdown · Delivered within 30 minutes
Package Brief
Quick verdict, integration guide, cost projections, gotchas with workarounds, and alternatives comparison.
Delivered within 10 minutes
Score Monitoring
Get alerted when this package's AF, security, or reliability scores change significantly. Stay ahead of regressions.
Continuous monitoring
Scores are editorial opinions as of 2026-03-30.