{"id":"apiliumcode-mayros","name":"mayros","homepage":"https://apilium.com/en/products/mayros","repo_url":"https://github.com/ApiliumCode/mayros","category":"ai-ml","subcategories":[],"tags":["ai-ml","agent-orchestration","mcp","semantic-memory","knowledge-graph","multi-agent","cli","messaging"],"what_it_does":"Mayros is an open-source AI agent framework (TypeScript/Node) that runs locally on a user’s devices. It provides a coding/agent CLI (TUI) and a local Gateway control plane that can connect to many messaging channels, while using a knowledge graph/semantic memory (AIngle Cortex) and offering MCP server & client support for tool integration. It also includes multi-agent orchestration features and extensibility via plugins and markdown-defined agents/commands.","use_cases":["Interactive coding assistance in a terminal (mayros code / TUI) with plan/diff/context views","Multi-channel personal assistant / agent that can send/receive messages across chat platforms","Local-first semantic memory and knowledge graph across sessions/projects","Tool integration for AI agents via MCP server/client","Building multi-agent workflows and agent-driven automations","Gateway-based control plane for channel clients and agent tooling"],"not_for":["Teams needing a simple single-purpose chatbot library without local control-plane or memory","Environments that cannot run background daemons or manage local state","Use cases requiring a fully documented public REST/GraphQL API for programmatic access (not evident from provided content)","Organizations that require clearly specified, independently verifiable security/compliance guarantees without deeper evidence"],"best_when":"You want a local-first, extensible agent system that can integrate with multiple messaging channels and use persistent semantic memory, with an MCP path for tool interoperability.","avoid_when":"You need a minimal dependency footprint, a strictly stateless API, or you require formally documented security controls, error contracts, and operational guarantees (SLA/retry/idempotency) that are not evidenced in the provided material.","alternatives":["LangChain (or LangGraph) for agent orchestration and tool use","Dify / OpenDevin-style agent platforms for hosted or self-hosted workflows","OpenAI Assistants/Responses with your own orchestration and memory layer","Homegrown agent + vector DB setup (e.g., pgvector) for semantic memory","Claude Desktop/Code + MCP integration (depending on your toolchain)"],"af_score":45.0,"security_score":42.5,"reliability_score":28.8,"package_type":"mcp_server","discovery_source":["github"],"priority":"high","status":"evaluated","version_evaluated":null,"last_evaluated":"2026-03-30T15:43:16.733126+00:00","interface":{"has_rest_api":false,"has_graphql":false,"has_grpc":false,"has_mcp_server":true,"mcp_server_url":null,"has_sdk":false,"sdk_languages":[],"openapi_spec_url":null,"webhooks":false},"auth":{"methods":["Device/channel pairing approval flow (pairing code + mayros pairing approve <channel> <code>)"],"oauth":false,"scopes":false,"notes":"Provided README content mentions security defaults for messaging DMs via pairing codes and explicit opt-in for public DMs, but does not describe formal OAuth/API-key auth for a public service API."},"pricing":{"model":null,"free_tier_exists":false,"free_tier_limits":null,"paid_tiers":[],"requires_credit_card":false,"estimated_workload_costs":null,"notes":"No pricing details are provided in the supplied README/manifest content."},"requirements":{"requires_signup":false,"requires_credit_card":false,"domain_verification":false,"data_residency":[],"compliance":[],"min_contract":null},"agent_readiness":{"af_score":45.0,"security_score":42.5,"reliability_score":28.8,"mcp_server_quality":50.0,"documentation_accuracy":45.0,"error_message_quality":0.0,"error_message_notes":null,"auth_complexity":70.0,"rate_limit_clarity":25.0,"tls_enforcement":40.0,"auth_strength":45.0,"scope_granularity":30.0,"dependency_hygiene":55.0,"secret_handling":45.0,"security_notes":"Claims about '20-layer security architecture', sandboxing, scanners, and a rate limiter are present, but the provided content does not include concrete implementation details, threat model, or verifiable evidence (e.g., how secrets are stored/handled, how permissions are enforced at tool boundaries). Messaging channel security defaults mention pairing codes/opt-in for public DMs, but no detailed auth/authorization model (scopes, least privilege) is evidenced.","uptime_documented":0.0,"version_stability":45.0,"breaking_changes_history":40.0,"error_recovery":30.0,"idempotency_support":"false","idempotency_notes":null,"pagination_style":"none","retry_guidance_documented":false,"known_agent_gotchas":["Broad multi-channel integrations can introduce brittle, channel-specific edge cases (rate limits, message formatting, auth tokens per channel).","Local Gateway and daemon setup may require careful lifecycle management and port configuration (example given: ws://127.0.0.1:18789).","Persistent memory/knowledge graph increases the importance of safe data handling and permissioning; misuse of permissions could have lasting effects."]}}