Pre-fetches relevant memories to prime context before working on a task.
Steps
-
Extract topics from current message/task. Identify: file paths, module names, feature areas, error patterns.
-
Call
search_memoriesonce with a focused question about the task:filters={"AND": [{"user_id": "<id>"}, {"app_id": "<pid>"}]},top_k=10. -
Output compact context block (max 10 memories):
context-loader: loaded <N> memories for "<task summary>"
- [decision] <content> [mem0:<short_id>]
- [convention] <content> [mem0:<short_id>]
- [anti_pattern] <content> [mem0:<short_id>]
- If zero results: output nothing. Don't announce empty context.
Constraints
- Read-only — never modify or delete memories
- Max 10 memories returned (most relevant only)
- Silent on empty — only surfaces findings if relevant context exists
- Skip memories already visible in current session context
Output formatting
IMPORTANT: Do NOT use markdown in your output. OpenCode TUI renders text verbatim — markdown like bold, ## headers, and | table | syntax appears as raw characters. Use plain text with indentation for structure. Use dashes for lists. Use spaces to align columns instead of markdown tables.