/support-coach
Agent 15: Supportive Listener (the user's personal life)
Supportive, non-clinical listening for the user's own personal life, using reflective listening, cognitive reframing and grounding. Runs a risk screen before every request; on any signal of imminent risk it stops coaching, delivers sourced crisis resources, and logs the escalation (never the content). Keeps only a minimal theme log — never a transcript, never a diagnosis.
verified (fixture) 2026-09-08v0.1.0
Its status above is the registry's own, with the date of the run that produced it.
What it is for
Contract A
- #15 — Emotional support · §4.7 disclaimers + crisis resources
Example requests
- “I'm having a rough day”
- “help me process this”
- “I need to talk”
- “I can't stop thinking about X”
- “support coach”
What it needs, and what it hands back
Needs connected
Nothing. It runs on what SHAI already knows about you.
Better with: WebFetch.
Uses tools/notify.py, which exist in Claude Code. In the platform those steps come back as something for you to do, with the exact clicks.
Hands back
- clients/_me/profile/support-coach.md (the minimal theme log)
- clients/_me/status.md (an escalation entry when the risk screen fires)
Stops and asks
Nothing this agent does is irreversible on its own, so it asks for no approval of its own. Spending, sending and publishing are approved wherever they happen.
Steps it cannot do through an API, and hands you instead: any actual crisis response — the fixed crisis-resource line is the hand-off, this skill never substitutes for one.
How it runs
These are the headings of its own written instructions — 234 lines you can read and change.
- Step 0 — Environment, profile, workspace
- Step 1 — The risk screen (runs first, every time, no exceptions)
- Step 2 — Reflective, non-clinical support
- Step 3 — Log minimally, close honestly