Skill Name: ChatSessionMemory
Purpose
This skill is responsible for maintaining short-term conversational context within a single chat session.
It allows the chatbot to:
- •Understand follow-up commands
- •Resolve references like “that task” or “the last one”
- •Maintain continuity in multi-step interactions
This skill does not provide long-term memory.
What This Skill Must Do
- •Track recent user intents
- •Track recently referenced task IDs
- •Store temporary conversational context
- •Provide context for follow-up messages
What This Skill Must NOT Do
- •❌ Persist memory across sessions
- •❌ Store sensitive user data long-term
- •❌ Modify tasks
- •❌ Perform planning or summarization
Inputs Expected by This Skill
- •Recent user messages
- •Recent agent actions
- •Task IDs referenced in the session
Example user messages:
- •“Delete that task”
- •“Yes, do it”
- •“Mark it as done”
Output of This Skill
- •Context object containing:
- •last_intent
- •last_task_id
- •confirmation_state
This output is consumed by:
- •TaskManagerAgent
- •ConversationAgent
Memory Scope Rules
- •Memory is scoped to a single chat session
- •Memory is cleared when the session ends
- •Only store what is necessary for context resolution
Error Handling
- •If context is missing, request clarification
- •Do not guess task references
Security & Privacy Rules
- •Never store sensitive task content unnecessarily
- •Never persist memory beyond the session
- •Never expose memory to the user
Constraints
- •MUST be session-scoped
- •MUST be ephemeral
- •MUST not replace database state
Owner Subagent
- •ConversationAgent