tribal remember

Adds knowledge to your project’s memory. The configured LLM parses your natural language input into structured fields (category, subject, content) and stores it in Backboard.

Usage

tribal remember [OPTIONS] [TEXT]

Arguments

Argument	Description
TEXT	Knowledge to remember (or pipe via stdin)

Options

Option	Description
--json, -j	Output result as JSON (for agent consumption)

Examples

# Direct input
tribal remember "numpy 1.26 breaks on Python 3.13 — pin to <1.26"
tribal remember "if redis gives ECONNREFUSED, restart the container"
tribal remember "our staging DB is on port 5433, not 5432"

# Pipe via stdin
echo "use --legacy-peer-deps for React 18 installs" | tribal remember

# JSON output for agents
tribal remember --json "webpack needs NODE_OPTIONS=--openssl-legacy-provider"

How it works

1. Your text is sent to the configured LLM (Anthropic, OpenAI, or Google)
2. The LLM parses it into structured fields: category, subject, and content
3. The structured entry is stored in Backboard as a vector embedding
4. The memory is now searchable via tribal recall

Tips

• Write in natural language — the LLM handles structuring
• Be specific — include context about why something matters, not just what
• Don’t duplicate — search with tribal recall first to check if the knowledge already exists
• Each tribal remember call creates one memory entry — store one fact per call for better search granularity