Remembering Conversations

Core principle: Search before reinventing. Searching costs nothing; reinventing or repeating mistakes costs everything.

Mandatory: Use the Search Agent

YOU MUST dispatch the search-conversations agent for any historical search.

Announce: "Dispatching search agent to find [topic]."

Then use the Task tool with subagent_type: "search-conversations" :

Task tool: description: "Search past conversations for [topic]" prompt: "Search for [specific query or topic]. Focus on [what you're looking for - e.g., decisions, patterns, gotchas, code examples]." subagent_type: "search-conversations"

The agent will:

• Search with the search tool

• Read top 2-5 results with the show tool

• Synthesize findings (200-1000 words)

• Return actionable insights + sources

Saves 50-100x context vs. loading raw conversations.

When to Use

Search proactively:

• Before implementing features

• Before making architectural decisions

• When debugging (especially familiar-seeming issues)

• When partner mentions past work

• At the start of ANY non-trivial task

Don't search:

• For info in current conversation

• For current codebase structure (use Grep/Read)

Direct Tool Access (Discouraged)

You CAN use MCP tools directly, but DON'T:

• mcp__plugin_episodic-memory_episodic-memory__search

• mcp__plugin_episodic-memory_episodic-memory__show

Using these directly wastes your context window. Always dispatch the agent instead.

See MCP-TOOLS.md for complete API reference if needed for advanced usage.