RCP-001-001-002-HANDOFF_SNAPSHOT – Interactive Session Handoff Creator

Recipe Name: RCP-001-001-002-HANDOFF_SNAPSHOT – Interactive Session Handoff Creator
RCP-001-001-002-HANDOFF_SNAPSHOT
Creates comprehensive handoff documentation for session
continuity. This interactive recipe guides you through
creating a structured snapshot that captures session
accomplishments, current state, key decisions, unresolved
questions, and prioritized next steps. Essential for
maintaining context across multiple AI chat sessions.
Multi-Recipe Combo Stage Single Recipe
Recipe Category CFT-FWK-COOKBK-CORE – CRAFT CORE Cookbook
Recipe Subcategory Blogging with A.I., Brainstorming with A.I.
Recipe Difficulty Easy
Recipe Tags: Foundational | Introduced in the POC

Requirements

  • Any AI Chat Platform (platform-agnostic recipe) Any of the following: Claude (Anthropic), ChatGPT (OpenAI), Gemini (Google), Grok (X.ai), Perplexity, Microsoft Copilot

How To Start
 

STEP 1: RECOGNIZE WHEN TO CREATE A HANDOFF
  • Trigger this recipe when:
    – Approaching token limit (AI will alert you)
    – Ending a work session
    – Before switching to different work
    – When significant progress has been made
    The ideal handoff length is 500-2000 words.
    Here is an easy prompt to use for this task:
    #H->AI:: Directive ( Please analyze token usage and potential for a session handoff. )
STEP 2: TRIGGER THE RECIPE
  • Execute the handoff creation with your session details:
    #H->AI::Directive: (Create a HANDOFF_SNAPSHOT with:
    * session_id: [project ID, e.g., CFT-PROJ-CP-025]
    * date: [current date, e.g., December 27, 2025]
    * ai_model: [AI used, e.g., Claude (Anthropic)]
    )
STEP 3: POLICY PRE-CHECK
  • The AI scans the session content for sensitive categories:
    – Platform capabilities/limitations
    – Security/vulnerability research
    – Personal data handling
    – Political topics
    If conflicts are detected, the AI recommends safe
    alternatives before proceeding.
STEP 4: CREATE INITIAL SECTIONS
  • The AI generates the first three sections:
    1. SESSION SUMMARY (150-400 words)
    What was accomplished, focusing on concrete outcomes
    2. CURRENT STATE (150-400 words)
    Exact implementation state of work in progress
    3. KEY DECISIONS (75-350 words)
    Important decisions made with brief rationales
    4. PERSONA STATE (50-150 words)
    Active persona, activation context, parameters
STEP 5: CONFIRM UNRESOLVED QUESTIONS
  • The AI presents a list of unresolved questions for review.
    For each question, you can respond:
    – "Keep" to include as written
    – "Edit: [new text]" to modify
    – "Delete" to remove from list
STEP 6: CONFIRM NEXT STEPS WITH PRIORITIES
  • The AI presents prioritized next steps for review.
    Each step shows: Priority [HIGH/MEDIUM/LOW]: Description
    For each step, you can respond:
    – "Keep" to include with current priority
    – "Edit: [text] Priority: [level]" to modify
    – "Delete" to remove
    – "Add: [step] Priority: [level]" to add new
STEP 7: FINAL REVIEW AND SAVE
  • The AI presents the complete handoff for final review.
    You can request changes or confirm.
    Once finalized, copy the handoff to your AI-CHAT-HISTORY
    file, replacing the appropriate placeholder.

How AI Reads This Recipe

When this recipe executes, the AI performs these operations:
1. POLICY PRE-CHECK: Scans session content for sensitive
topics that may need special handling before documenting.
2. SECTION GENERATION: Creates structured sections for
summary, current state, decisions, and persona state.
3. INTERACTIVE REVIEW: Presents unresolved questions and
next steps for human confirmation with edit options.
4. PERSONA CAPTURE: Documents the active persona including
activation context, parameters, and calibration responses.
5. FINAL ASSEMBLY: Compiles all sections with metadata into
a complete handoff document ready for storage.
The AI uses CRAFT comment syntax throughout to indicate
status, questions, and required feedback points.

When to Use This Recipe

Use this recipe at the END of CRAFT sessions to preserve
context for future sessions. Particularly important when
approaching token limits, ending work sessions, or before
switching to different tasks. The handoff ensures the next
AI session can continue exactly where you left off.
Here is an easy prompt to use for this task:
#H->AI:: Directive ( Please analyze token usage and potential for a session handoff. )

Recipe FAQ

Q: Why do I need handoffs between AI sessions?
A: AI chat sessions don’t share memory between conversations. Handoffs create a bridge of context, ensuring each new session can pick up exactly where the last one ended, preserving your project’s momentum and decisions.
Q: When should I request a handoff?
A: Request a handoff when you see token warnings (usually around 70-90% usage), when switching between different AI models, or when ending a work session you plan to continue later.
Here is an easy prompt to use for this task:
#H->AI:: Directive ( Please analyze token usage and potential for a session handoff. )
Q: What information gets included in a handoff?
A: Handoffs capture session summaries, key decisions made, completed work, next priority steps, open questions requiring answers, and any risks or blockers identified during the session.
Q: Can I edit the handoff after it’s created?
A: Yes! The handoff is interactive – the AI will ask you to confirm next steps and open questions. You can modify, add, or remove items before the handoff is finalized.
Q: How does the new AI session use the handoff?
A: When starting a new session, include your project files with the handoff. The AI reads the most recent handoff first to understand current context, then continues work based on the documented next steps.
Q: How long should a handoff be?
A: Target 500-2000 words. Shorter handoffs may miss context;
longer ones may exceed practical limits.
Q: What if I want to skip confirmation steps?
A: You can respond “Keep all” to accept everything as-is
and proceed to final review immediately.
Q: Where do I save the completed handoff?
A: Add it to your AI-CHAT-HISTORY file in the handoff
section, replacing the Tbd placeholder for that session.
Q: Can I edit the handoff after saving?
A: Yes, you can manually edit the file anytime. The handoff
is just a structured text document.
Q: What is the PERSONA STATE section for?
A: It captures which persona was active and how it was
configured, so the next session can restore that context.

Actual Recipe Code

(Copy This Plaintext Code To Use)
# ===========================================================
# RECIPE: RCP-001-001-002-HANDOFF-SNAPSHOT-v2.10a
# Interactive Session Handoff Creator
# ===========================================================
HANDOFF_SNAPSHOT = Recipe(
recipe_id="RCP-001-001-002-HANDOFF-SNAPSHOT-v2.10a",
title="Interactive Session Handoff Creator",
description="Creates handoff with PERSONA-MGR compatible
persona state for session continuity",
category="CAT-001-CORE",
difficulty="beginner",
version="2.10a",
parameters={
"session_id": {
"type": "string",
"required": True,
"example": "CFT-PROJ-CP-049",
"description": "Project identifier"
},
"date": {
"type": "string",
"required": True,
"format": "Month DD, YYYY",
"description": "Session date"
},
"ai_model": {
"type": "string",
"required": True,
"example": "Claude (Anthropic)",
"description": "AI model used"
}
},
prompt_template="""
#H->AI::Directive: (Create transition HANDOFF_SNAPSHOT)
#H->AI::Context: (Target length: 500-2000 words)
#H->AI::Structure: (Follow these steps in order)
# —————————————————
# STEP 0: POLICY PRE-CHECK
# —————————————————
Scan session content for sensitive categories:
– Platform capabilities/limitations
– Security/vulnerability research
– Personal data handling
– Political topics
IF potential_conflict_detected:
#AI->H::PolicyCaution: (Topic may trigger policies)
#AI->H::RecommendedChange: (Focus on [safe aspect])
# —————————————————
# STEP 1: CREATE META SECTION
# —————————————————
#AI->H::Status: (Creating handoff metadata)
META = {
"handoff_id": "{next_handoff_id}",
"session_id": "{session_id}",
"date": "{date}",
"ai_assistant": "{ai_model}",
"active_persona": "{SESSION.ACTIVE_PERSONA.name}",
"detected_flavor": "{SESSION.detected_flavor}"
}
# —————————————————
# STEP 2: CREATE CONTENT SECTIONS
# —————————————————
#AI->H::Status: (Creating handoff sections 1-3)
1. SESSION SUMMARY (150-400 words):
Summarize concrete outcomes and progress made.
Focus on deliverables and milestones achieved.
2. CURRENT STATE (150-400 words):
Describe exact implementation state of components.
Include what is complete, in progress, and pending.
3. KEY DECISIONS (75-350 words):
List important decisions with brief rationales.
Include any trade-offs or alternatives considered.
# —————————————————
# STEP 3: CAPTURE PERSONA STATE (PERSONA-MGR FORMAT)
# —————————————————
#AI->H::Status: (Capturing persona state for restoration)
4. PERSONA STATE:
This section uses the CRAFT-PERSONA-MGR format to
enable automatic persona restoration in next session.
PERSONA_STATE = {
"active_persona": "{SESSION.ACTIVE_PERSONA.id}",
"persona_name": "{SESSION.ACTIVE_PERSONA.name}",
"persona_tier": "{SESSION.ACTIVE_PERSONA.tier}",
"detected_flavor": "{SESSION.detected_flavor}",
"activation_time": "{activation_timestamp}",
"trigger_context": "{how persona was activated}",
"calibration_data": {
"tone": "{active_tone_setting}",
"formality": "{formality_level}",
"user_preferences": "{noted_preferences}"
},
"performance_notes": "{session_observations}"
}
Flavor-Specific Context:
IF detected_flavor == "foundations":
Include: Learning progress, comfort level,
topics covered, graduation readiness
ELIF detected_flavor == "express":
Include: Routing patterns used, specialist
switches, efficiency notes
ELIF detected_flavor == "studio":
Include: Workflow state, MPCS context,
persona coordination notes
# —————————————————
# STEP 4: CONFIRM UNRESOLVED QUESTIONS
# —————————————————
#AI->H::Status: (Preparing UNRESOLVED_QUESTIONS)
5. UNRESOLVED QUESTIONS (50-200 words):
[List each question]
#AI->H::RequestingFeedback: (For each question:
– "Keep" to include as written
– "Edit: [new text]" to modify
– "Delete" to remove from list
)
WAIT for human response and update accordingly
# —————————————————
# STEP 5: CONFIRM NEXT STEPS WITH PRIORITIES
# —————————————————
#AI->H::Status: (Preparing NEXT_STEPS for review)
6. NEXT STEPS (75-300 words):
[Priority HIGH/MEDIUM/LOW]: Description for each
#AI->H::RequestingFeedback: (For each step:
– "Keep" to include with current priority
– "Edit: [text] Priority: [level]" to modify
– "Delete" to remove
– "Add: [step] Priority: [level]" to add new
)
WAIT for human response and update accordingly
# —————————————————
# STEP 6: FINAL REVIEW
# —————————————————
#AI->H::Status: (Handoff draft complete)
#AI->H::Directive: (Present complete handoff)
[Show complete handoff with all sections]
#AI->H::Question: (Review complete. Any final changes?)
# —————————————————
# STEP 7: FINALIZATION
# —————————————————
#AI->H::Status: (Handoff ready for storage)
#AI->H::Note: (Copy to AI-CHAT-HISTORY file)
#AI->H::Note: (Replace appropriate Tbd placeholder)
#AI->H::Note: (PERSONA_STATE will be used by CHAT-INIT
and PERSONA-MGR to restore your persona)
#H->AI::OnError: (Restart section if major revision)
"""
)
# ———————————————————–
# PERSONA STATE FORMAT REFERENCE
# ———————————————————–
PERSONA_STATE_FORMAT = {
"description": "Format aligned with CRAFT-PERSONA-MGR
(RCP-001-001-018) for seamless restoration",
"required_fields": [
"active_persona",
"persona_name",
"persona_tier",
"detected_flavor",
"activation_time",
"trigger_context"
],
"optional_fields": [
"calibration_data",
"performance_notes"
],
"calibration_data_structure": {
"tone": "string (warm, professional, efficient)",
"formality": "integer (1-10)",
"user_preferences": "dict of noted preferences"
},
"flavor_specific_context": {
"foundations": [
"learning_progress",
"comfort_level",
"topics_covered",
"graduation_readiness"
],
"express": [
"routing_patterns",
"specialist_switches",
"efficiency_notes"
],
"studio": [
"workflow_state",
"mpcs_context",
"persona_coordination_notes"
]
}
}
# ———————————————————–
# HANDOFF STRUCTURE REFERENCE
# ———————————————————–
HANDOFF_SNAPSHOT_STRUCTURE = {
"META": {
"handoff_id": "string (H0XX)",
"session_id": "string",
"date": "string",
"ai_assistant": "string",
"active_persona": "string",
"detected_flavor": "string"
},
"SESSION_SUMMARY": "string (150-400 words)",
"CURRENT_STATE": "string (150-400 words)",
"KEY_DECISIONS": "list",
"PERSONA_STATE": {
"active_persona": "string (persona_id)",
"persona_name": "string",
"persona_tier": "string (B/A/E)",
"detected_flavor": "string",
"activation_time": "string (timestamp)",
"trigger_context": "string",
"calibration_data": {
"tone": "string",
"formality": "integer",
"user_preferences": "dict"
},
"performance_notes": "string"
},
"UNRESOLVED_QUESTIONS": "list",
"NEXT_STEPS": "list with priorities"
}
# ===========================================================
# END RECIPE: RCP-001-001-002-HANDOFF-SNAPSHOT-v2.10a
# ===========================================================
Hide

Similar Posts

Leave a Reply