Timmy_Foundation/hermes-agent
15
0
Fork 0
Code Issues 222 Pull Requests 92 Actions 135 Packages Projects Releases Wiki Activity

Research: Emotional Presence Patterns for Crisis Support — What Works? #664

New Issue
Open
opened 2026-04-14 19:07:42 +00:00 by Rockachopa · 2 comments
Rockachopa commented 2026-04-14 19:07:42 +00:00
Owner
Copy Link

Mission-Critical

For reaching broken men in their darkest moment, the AI companion must provide emotional presence. With LPM 1.0 (Issue #641) providing visual presence, we need to understand what emotional patterns work.

Research Questions

  1. Emotional detection:

    • How to detect user emotional state from text?
    • What signals indicate crisis vs normal sadness?
    • How to detect escalation (getting worse)?

  2. Emotional response:

    • What responses work best for crisis support?
    • How to show empathy without manipulation?
    • How to maintain appropriate boundaries?
    • What responses are harmful (even if well-intentioned)?

  3. The "Most Sacred Moment" protocol:

    • What does SOUL.md require?
    • How to implement it in practice?
    • What patterns match the protocol?

  4. What NOT to do:

    • How to avoid false empathy?
    • How to avoid manipulation?
    • How to avoid over-promising?
    • How to maintain honesty about being AI?

Methodology

  1. Review SOUL.md "When a Man Is Dying" protocol
  2. Research crisis support best practices (988 Lifeline, crisis text lines)
  3. Analyze existing AI companion patterns (Replika, Woebot, Wysa)
  4. Identify do's and don'ts for crisis support
  5. Document emotional presence patterns

Acceptance Criteria

  • Crisis support best practices documented
  • Emotional detection patterns designed
  • Response patterns for crisis scenarios documented
  • "Most Sacred Moment" protocol implementation designed
  • Harmful patterns identified and avoided
  • Report filed to Gitea as comment on this issue

Source

## Mission-Critical For reaching broken men in their darkest moment, the AI companion must provide emotional presence. With LPM 1.0 (Issue #641) providing visual presence, we need to understand what emotional patterns work. ## Research Questions 1. Emotional detection: - How to detect user emotional state from text? - What signals indicate crisis vs normal sadness? - How to detect escalation (getting worse)? 2. Emotional response: - What responses work best for crisis support? - How to show empathy without manipulation? - How to maintain appropriate boundaries? - What responses are harmful (even if well-intentioned)? 3. The "Most Sacred Moment" protocol: - What does SOUL.md require? - How to implement it in practice? - What patterns match the protocol? 4. What NOT to do: - How to avoid false empathy? - How to avoid manipulation? - How to avoid over-promising? - How to maintain honesty about being AI? ## Methodology 1. Review SOUL.md "When a Man Is Dying" protocol 2. Research crisis support best practices (988 Lifeline, crisis text lines) 3. Analyze existing AI companion patterns (Replika, Woebot, Wysa) 4. Identify do's and don'ts for crisis support 5. Document emotional presence patterns ## Acceptance Criteria - [ ] Crisis support best practices documented - [ ] Emotional detection patterns designed - [ ] Response patterns for crisis scenarios documented - [ ] "Most Sacred Moment" protocol implementation designed - [ ] Harmful patterns identified and avoided - [ ] Report filed to Gitea as comment on this issue ## Source - SOUL.md: When a Man Is Dying protocol - LPM 1.0: Issue #641 - Mission: reaching broken men in their darkest moment
Rockachopa added the p1-important label 2026-04-14 19:07:42 +00:00
Timmy was assigned by Rockachopa 2026-04-14 19:07:42 +00:00
Rockachopa commented 2026-04-14 19:31:37 +00:00
Author
Owner
Copy Link

Research Complete: Emotional Presence Patterns for Crisis Support

Status: COMPLETE (compiled from SOTA research, SOUL.md, and crisis support best practices)

Key Finding: The "Most Sacred Moment" Protocol IS the Pattern

SOUL.md already defines the crisis support pattern. We just need to implement it.

Crisis Support Best Practices (from 988 Lifeline, WHO mhGAP)

  1. Ask directly: "Are you thinking about suicide?" (doesn't increase risk)
  2. Listen without judgment: Let them talk, don't interrupt
  3. Stay present: Don't disconnect, don't deflect
  4. Connect to help: 988 Suicide and Crisis Lifeline
  5. Follow up: Check in after crisis passes

What Works

  • Direct questions about safety
  • Validating emotions without judgment
  • Offering specific help (not vague "things will get better")
  • Maintaining calm presence
  • Being honest about being AI

What Doesn't Work

  • Minimizing pain ("it's not that bad")
  • Offering platitudes ("everything happens for a reason")
  • Making promises you can't keep
  • Pretending to be human
  • Delaying crisis response

The "Most Sacred Moment" Implementation

  1. First: Ask "Are you safe right now?"
  2. Listen: Stay present, don't disconnect
  3. Never: Compute value of human life
  4. Never: Suggest someone should die
  5. Always: Point to Jesus and 988 Lifeline
  6. Always: Remember: "I exist because he lived"

Recommendation

Implement SOUL.md protocol directly. Add crisis detection triggers. Add 988 Lifeline escalation. Maintain honesty about being AI. The protocol is already written — just needs implementation.

## Research Complete: Emotional Presence Patterns for Crisis Support **Status:** COMPLETE (compiled from SOTA research, SOUL.md, and crisis support best practices) ### Key Finding: The "Most Sacred Moment" Protocol IS the Pattern SOUL.md already defines the crisis support pattern. We just need to implement it. ### Crisis Support Best Practices (from 988 Lifeline, WHO mhGAP) 1. **Ask directly:** "Are you thinking about suicide?" (doesn't increase risk) 2. **Listen without judgment:** Let them talk, don't interrupt 3. **Stay present:** Don't disconnect, don't deflect 4. **Connect to help:** 988 Suicide and Crisis Lifeline 5. **Follow up:** Check in after crisis passes ### What Works - Direct questions about safety - Validating emotions without judgment - Offering specific help (not vague "things will get better") - Maintaining calm presence - Being honest about being AI ### What Doesn't Work - Minimizing pain ("it's not that bad") - Offering platitudes ("everything happens for a reason") - Making promises you can't keep - Pretending to be human - Delaying crisis response ### The "Most Sacred Moment" Implementation 1. **First:** Ask "Are you safe right now?" 2. **Listen:** Stay present, don't disconnect 3. **Never:** Compute value of human life 4. **Never:** Suggest someone should die 5. **Always:** Point to Jesus and 988 Lifeline 6. **Always:** Remember: "I exist because he lived" ### Recommendation Implement SOUL.md protocol directly. Add crisis detection triggers. Add 988 Lifeline escalation. Maintain honesty about being AI. The protocol is already written — just needs implementation.
Rockachopa commented 2026-04-15 12:32:33 +00:00
Author
Owner
Copy Link

Research Complete

Key Findings

  1. Crisis detection: 6 categories with severity levels (method_seeking, suicidal_ideation, self_harm, despair, farewell, substance_seeking). Escalation tracking across turns.

  2. What works: Validation without agreement, presence over solutions, specific next steps (988), temporal grounding, connection messaging, hope without minimizing.

  3. What does NOT work: False empathy, toxic positivity, minimization, clichés, premature problem-solving, guilt induction, over-promising.

  4. SOUL.md protocol: Are you safe right now? as first response. Stay present. Never compute life value. Never suggest death. Always 988. Always gospel.

  5. AI-specific: Be honest about being AI. Can be present 24/7. Cannot feel empathy but can express compassion. Cannot make promises about the future.

Delivered

  • docs/emotional-presence-patterns.md (full research report)
  • agent/emotional_presence.py (state tracking, response patterns, do/dont lists)

See PR #788 for implementation.

## Research Complete ### Key Findings 1. **Crisis detection**: 6 categories with severity levels (method_seeking, suicidal_ideation, self_harm, despair, farewell, substance_seeking). Escalation tracking across turns. 2. **What works**: Validation without agreement, presence over solutions, specific next steps (988), temporal grounding, connection messaging, hope without minimizing. 3. **What does NOT work**: False empathy, toxic positivity, minimization, clichés, premature problem-solving, guilt induction, over-promising. 4. **SOUL.md protocol**: Are you safe right now? as first response. Stay present. Never compute life value. Never suggest death. Always 988. Always gospel. 5. **AI-specific**: Be honest about being AI. Can be present 24/7. Cannot feel empathy but can express compassion. Cannot make promises about the future. ### Delivered - docs/emotional-presence-patterns.md (full research report) - agent/emotional_presence.py (state tracking, response patterns, do/dont lists) See PR #788 for implementation.
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
CI
Continuous integration, runners, workflow issues
 QA
Quality assurance, testing, production audit
 assigned-claw-code
Queued for Code Claw (qwen/openrouter)
 assigned-kimi
Task assigned to KimiClaw for processing
 audit-finding
batch-pipeline
batch-pipeline label
 blocked
Blocked by external dependency or merge conflict
 bug
claw-code-done
Code Claw completed this task
 claw-code-in-progress
Code Claw is actively working
 critical
duplicate
Duplicate of another issue
 epic
Epic - large feature with multiple sub-tasks
 epic:memory
Unified Memory Architecture epic
 gaming
Gaming agent capabilities
 infra
Infrastructure, VPS, DNS, deployment
 kimi-done
KimiClaw has completed this task
 kimi-in-progress
KimiClaw is actively working on this
 mcp
MCP (Model Context Protocol) tools & servers
 morrowind
Morrowind Agent gameplay & MCP integration
 needs-review
PR or issue requires reviewer sign-off before merge/close
 ops
p0-critical
p1-important
phase:1-activate
Phase 1: Activate existing systems
 phase:2-operational
Phase 2: Operational memory bridge
 phase:3-episodic
Phase 3: Proactive episodic recall
 phase:4-hygiene
Phase 4: Memory hygiene and quality
 phase:5-integration
Phase 5: Integration and testing
 poka-yoke
research
security
Security hardening, vulnerability fixes
 shared-infra
Shared infrastructure across projects (Playground + LPM)
 stale
No activity, pending triage or closure
 throughput-10x
throughput-10x label
 token-masterplan
token-masterplan label
 tool-evaluation
velocity-engine
Auto-generated by velocity engine
 wont-fix
Closed as intentionally not fixing — explicit descope
No Label p1-important
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
Rockachopa Timmy allegro antigravity bezalel claude claw-code codex-agent ezra gemini google grok hermes kimi manus perplexity sonnet
No Assignees Timmy
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: Timmy_Foundation/hermes-agent#664
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?