self-knowledge

Answers questions about Fermix's agent loop, providers, channels, jobs, memory, and configuration surfaces.

9|Updated Mar 31, 2026
One-click install
npx skills add https://github.com/tezra-io/fermix --skill self-knowledge-tezra-io
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: self-knowledge
Source: https://github.com/tezra-io/fermix/tree/main/apps/fermix_core/priv/skills/self_knowledge
Command: npx skills add https://github.com/tezra-io/fermix --skill self-knowledge-tezra-io

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? When users ask what Fermix is, what it can do, or how its internal pieces fit together, the agent needs accurate, detailed knowledge of the daemon's architecture rather than guessing. This Skill supplies that authoritative reference so answers about capabilities, configuration, and behavior are correct. ## Core Features & Use Cases - Runtime architecture reference: Explains the gateway, agent turn flow, sub-agent fan-out, iteration limits, and the /ultra run mode. - Provider and routing knowledge: Covers the eight model providers, primary/fallback chains, failover rules, and sub-agent model routing configuration. - Capability catalog: Documents built-in tools for files, shell, web, browser, places, media, memory, jobs, events, coding harnesses, and meetings, plus trust and sandbox rules. - Use Case: A user asks "why did my scheduled job use a different model?" — the agent loads this Skill to explain routing inheritance and the cron_provider configuration. ## Quick Start Ask the agent what Fermix can do or how a specific feature like providers, jobs, or memory works, and it will consult this Skill for the answer.

Frequently Asked Questions about self-knowledge

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
What is Fermix and what can it do?▼

Fermix is an Elixir/OTP agent daemon that runs as a local background service and connects to chat channels like Telegram, WhatsApp, Slack, Discord, and Signal. It supports eight model providers, scheduled jobs, durable memory, and built-in tools for files, shell, web, and browser automation.

How does Fermix provider failover work?▼

Fermix uses a primary-flag model where exactly one provider is primary and others are automatic fallbacks in catalog order. Failover happens only on the turn's initial model call for transient errors like timeouts, 5xx, rate limits, or quota issues, after a bounded same-provider retry.

How do I set a different model for sub-agents in Fermix?▼

Set subagent_model, subagent_provider, or subagent_reasoning_effort under [fermix_core.routing] in config.toml, via the setup page, or with the model_routing_config tool. A model set without an explicit provider runs on the primary provider.

What built-in tools does the Fermix agent have?▼

Built-ins include file_read/file_write/file_edit, shell and git tools, web_search, web_fetch, browser automation, place_search, generate_image, memory tools, job scheduling, events and reminders, skill tools, subagents, and a coding harness for Codex and Claude Code CLIs.

Can guests use all Fermix capabilities?▼

No. Guests get read-only chat with no skills, MCP, exec, or network access, and tools returning owner data like file_read, list_jobs, and memory_sources_list are excluded from their surface entirely. Only operators get the full capability set.

Why is the Fermix mobile channel not available?▼

The iPhone companion app is not released, and the mobile channel is feature-flagged off with no setup surface. It can only be enabled by hand-editing enabled = true under [fermix_channels.mobile] in config.toml and restarting the daemon.