Configuration

Saved defaults

Stop retyping the same flags by saving defaults:

mentor config --set provider=claude --set level=intermediate --set style=coaching
mentor config                       # show effective config
mentor config --unset style
mentor config --reset

Stored at ~/.config/expert-mentor/config.json. Valid keys:

provider, model, level, language, style, temperature, learner, prompt_format, compact, thinking, reasoning_effort.

Command-line flags always override saved defaults.

Environment variables

Variable	Purpose
`EXPERT_MENTOR_HOME`	base config/data directory (default `~/.config/expert-mentor`)
`ANTHROPIC_API_KEY`, `CLAUDE_API_KEY`	Claude credentials
`OPENAI_API_KEY`, `CHATGPT_API_KEY`	OpenAI credentials
`ANTHROPIC_BASE_URL`, `OPENAI_BASE_URL`	endpoint overrides
`OLLAMA_HOST`, `LLAMACPP_HOST`, `LLAMACPP_API_KEY`	local backends
`NO_COLOR`	disable colored output

A .env file in the current directory or in $EXPERT_MENTOR_HOME/.env is loaded automatically; existing environment variables win.

# ~/.config/expert-mentor/.env
ANTHROPIC_API_KEY=sk-ant-...
OPENAI_API_KEY=sk-...
OLLAMA_HOST=http://localhost:11434

File locations

Everything lives under $EXPERT_MENTOR_HOME (default ~/.config/expert-mentor):

config.json                         saved defaults
.env                                optional credentials
prompts/<name>.md                   saved mentors (mentor save)
models/<name>.Modelfile             generated Ollama models
learners/<name>.json                learner profiles
learners/<name>.cards.json          flashcards
sessions/<timestamp>-<name>.md      session transcripts
sessions/<timestamp>-<name>.json    session metadata

All of it is plain JSON/markdown — inspect or edit by hand, and commit it to your own dotfiles if you like.

Field profiles

Curated field profiles ship in references/fields.md. Add your own in references/fields.json (merged over the built-ins):

{
  "my-field": {
    "key": "Human-readable name",
    "aliases": ["my field", "abbreviation"],
    "mentor_name": "Dr. Example",
    "notes": "What professional competence means here: core reasoning, standards, traps, anchors.",
    "concepts": ["concept 1", "concept 2"],
    "misconceptions": ["wrong model A"],
    "resources": ["Author, Title"],
    "capstone": "One realistic project that proves competence."
  }
}

See references/fields.md for details.

FilesExpand file tree

configuration.md

Latest commit

History