Command-Line Interface

Installing mycontext-ai adds a mycontext console script. Every command runs locally and offline — the only command that ever contacts an LLM is run --execute, and only with your own API key.

mycontext --version
mycontext --help

Commands

Command	What it does
mycontext list	List all 88 cognitive patterns with descriptions
mycontext run	Build a context from a pattern (optionally execute it)
mycontext skills export	Emit progressive-disclosure SKILL.md packages or a plugin
mycontext mcp	Start the local stdio MCP server

list

List every pattern and its one-line description:

mycontext list

audience_adapter                  Adapt a message for a specific audience
code_reviewer                     Security, performance, and maintainability review
decision_framework                Structured decision-making with weighted criteria
...
88 cognitive patterns (all open source).

run

Build a structured context from any pattern. By default it prints the assembled, portable prompt — no LLM call, no API key required:

mycontext run root_cause_analyzer "Why did our API latency spike after the deploy?"

--generic

Print the pre-authored generic prompt for the pattern instead of the full cognitive scaffold (the fastest, zero-dependency option):

mycontext run decision_framework "Which database should we standardize on?" --generic

--execute

Execute the context against an LLM using your own key. Set the provider's API key in your environment first (e.g. OPENAI_API_KEY):

export OPENAI_API_KEY="sk-..."
mycontext run risk_assessor "Assess the risk of migrating to a monorepo" --execute --provider openai

Flag	Default	Description
--generic	off	Emit the pre-authored generic prompt instead of the full scaffold
--execute	off	Run the context through an LLM (requires a provider API key)
--provider	openai	LLM provider used with --execute

Next steps

• Export patterns as agent skills →
• Run the local MCP server →
• Author Requirements-as-Code →