Entity Memory Management
Persist named entities—campaigns, metrics, decisions, and preferences—across agent sessions so your assistant recalls specifics instead of vague summaries.
Overview
Entity Memory Management is a journey-wide agent skill that extracts, stores, and retrieves named entities across sessions—usable whenever a solo builder needs persistent, queryable context before committing to agent ans
Install
npx skills add https://github.com/itallstartedwithaidea/agent-skills --skill entity-memory-managementWhat is this skill?
- Granular entity tracking (campaigns, metrics, decisions, pain points) vs generic session summaries
- Production architecture used by Buddy at googleadsagent.ai across thousands of entities
- Queryable recall of concrete facts such as strategy changes on named campaigns with dates
- Typed entities with trajectories, budgets, KPI history, and decision rationales
- Transforms transient chats into persistent relationship context for support and ops workflows
- Production system tracks thousands of named entities across hundreds of user sessions (per skill description)
Adoption & trust: 1 installs on skills.sh; 18 GitHub stars; 3/3 security scanners passed (skills.sh audits); trending (+100% hot-view momentum).
What problem does it solve?
Your agent forgets named campaigns, metrics, and past decisions, so every session feels like starting over with vague paraphrases instead of factual recall.
Who is it for?
Builders running longitudinal agents for ads, analytics, CRM, or ops where entity-level memory beats chat summaries.
Skip if: Stateless one-shot codegen tasks or apps with no need to remember user-specific named objects between sessions.
When should I use this skill?
You need systematic extraction, persistence, and retrieval of named entities across agent sessions for long-term contextual awareness.
What do I get? / Deliverables
Sessions reference discrete, typed entities with histories so the agent can cite specific changes, KPIs, and user preferences on demand.
- Entity type model and persistence/retrieval workflow
- Session queries answered with named-entity facts
- Architecture pattern for granular vs summary memory
Recommended Skills
Journey fit
Useful at every journey phase - explore requirements and options before committing to a direction.
Where it fits
Design entity types and storage hooks before shipping a support bot that remembers account-specific campaigns.
Recall prior budget and bidding decisions when a returning user asks why performance shifted last week.
Demo persistent recall in a vertical prototype so stakeholders see differentiated agent UX.
Document entity-memory capabilities in positioning for AI-assisted products that promise continuity.
How it compares
Entity-typed persistent memory architecture, not a single-session context window or generic note-taking skill.
Common Questions / FAQ
Who is entity-memory-management for?
Solo and indie developers building production assistants that must remember campaigns, metrics, decisions, and user context across many return visits.
When should I use entity-memory-management?
During Build when designing agent-tooling, during Grow when lifecycle support needs account history, during Operate when monitoring decisions and incidents per named asset, and during Validate when prototyping personalized agent demos.
Is entity-memory-management safe to install?
It implies storing user and business data persistently; review the Security Audits panel on this page and align retention with your privacy policy before enabling in production.
SKILL.md
READMESKILL.md - Entity Memory Management
# Entity Memory Management Part of [Agent Skills™](https://github.com/itallstartedwithaidea/agent-skills) by [googleadsagent.ai™](https://googleadsagent.ai) ## Description Entity Memory Management is the systematic extraction, persistence, and retrieval of named entities across agent sessions, enabling long-term contextual awareness that transforms transient conversations into persistent relationships. While generic memory systems store session summaries, entity memory operates at a granular level — tracking specific campaigns, metrics, decisions, user pain points, and preferences as discrete, queryable objects. This precision enables agents to recall that "Campaign Alpha switched to Target CPA last Tuesday" rather than merely "we discussed bidding changes recently." This skill is the production entity memory architecture powering Buddy™ at [googleadsagent.ai™](https://googleadsagent.ai), where entity memory tracks thousands of named entities across hundreds of user sessions. The system maintains entities of defined types — campaigns (with their strategies, budgets, and performance trajectories), metrics (specific KPIs and their historical values), decisions (changes made and their rationale), and pain points (recurring issues the user has expressed frustration about). Each entity is extracted, typed, scored for importance, stored in Cloudflare KV, and retrieved contextually in future sessions. The result is an agent that genuinely "remembers" its users — not through conversation replay, but through structured, typed knowledge that persists and evolves. This creates a fundamentally different user experience where the agent builds on prior context rather than starting fresh. ## Use When - The agent serves repeat users who expect continuity across sessions - Domain entities (campaigns, accounts, projects) need tracking over time - Decisions made in past sessions should inform future recommendations - User preferences and pain points should persist without re-explanation - You need to detect trends across sessions (improving metrics, recurring issues) - Compliance or audit requirements demand a structured record of agent-user interactions ## How It Works ```mermaid graph TD A[Conversation Stream] --> B[Entity Extractor] B --> C{Entity Type} C -->|Campaign| D[Campaign Schema] C -->|Metric| E[Metric Schema] C -->|Decision| F[Decision Schema] C -->|Pain Point| G[Pain Point Schema] C -->|Preference| H[Preference Schema] D --> I[Importance Scorer] E --> I F --> I G --> I H --> I I --> J[KV Store Writer] J --> K[Entity Index] L[New Session Start] --> M[Context Builder] M --> N[Entity Retriever] N --> K K --> O[Relevance Filter] O --> P[Context Injection] ``` The entity extraction pipeline processes the conversation stream through a type-aware extractor that identifies named entities and classifies them into predefined schemas. Each extracted entity receives an importance score based on conversational emphasis, actionability, and novelty (new information scores higher than repeated information). Scored entities are written to Cloudflare KV with structured keys that enable efficient retrieval. At the start of each new session, the context builder queries the entity index for the current user, filters by relevance to the new task, and injects the most relevant entities into the agent's context. ## Implementation **Entity Type Schemas:** ```typescript interface BaseEntity { id: string; userId: string; type: EntityType; name: string; content: string; importance: number; createdAt: number; updatedAt: number; sessionId: string; version: number;