project-memory-recall

name	project-memory-recall
description	Retrieve universal coding patterns from vector database. Auto-invokes before complex tasks or when user says "--recall". Searches relevant role collections based on task context.

Use Task tool with `subagent_type: "memory-agent"` to keep main context clean.

The memory-agent has ZERO access to Read/Write/Edit/Glob/Bash - it can ONLY use MCP memory tools. This prevents file reading pollution by design.

Search when: Non-obvious bugs, complex architectures, performance issues, unfamiliar domains, hard problems.

Skip when: Obvious tasks, basic file operations, standard workflows, problems solvable with basic knowledge.

**Step 1: Build semantic query** Construct 2-3 sentences capturing: What is the problem? What is the technical context? What outcome is desired?

Example: "Need to implement rate limiting for REST API to prevent abuse. Backend service in Node.js with Express. Want proven pattern that prevents thundering herd."

Step 2: Detect relevant roles Determine which role collections to search based on task context. Use role_mapping below. Default to ["universal"] if unclear. Can search multiple roles when task spans domains.

Step 3: Search previews Use search_memory with query, roles=["detected_role", "universal"], limit=20.

Note: The roles parameter tells MCP which collections to search. Always include "universal" to catch cross-domain patterns.

Step 4: Analyze previews Review returned previews (title + description + tags). Select 3-5 most relevant based on:

• Does title match problem domain?
• Does description indicate relevant solution?
• Do tags align with task?
• Is memory type appropriate? (episodic for debugging, procedural for workflows, semantic for patterns)

Step 5: Retrieve full content Use batch_get_memories with selected doc_ids and roles=["detected_role", "universal"].

Note: batch_get_memories needs roles parameter to know which collections to search in.

Step 6: Present results Format retrieved memories clearly. Let the main agent decide what to apply.

<role_mapping> Available roles (maps to Qdrant collections):

• universal: General patterns applicable across domains
• backend: API, endpoint, database, server, auth
• frontend: React, Vue, component, UI, CSS
• devops: Deploy, Docker, Kubernetes, CI/CD
• ai: Model, training, embedding, LLM
• security: Vulnerability, encryption, JWT
• mobile: iOS, Android, Flutter, Swift
• pm: Project management, coordination, delegation, reporting
• scrum-master: Agile, sprint, standup, retrospective, planning
• qa: Testing, quality assurance, verification, validation
• quant: Trading, backtesting, portfolio, risk

Default to "universal" if unclear. Each role maps to a collection named "{role}-patterns" (e.g., "backend-patterns").

Search multiple role collections when task spans domains (e.g., ["backend", "devops", "universal"] for API deployment task). </role_mapping>