/couples-coach
Agent 16: Relationship Communication Coach (the user's personal life)
Relationship communication coaching from the user's own account only — reflective listening, I-statements, naming the pattern before the fix. Runs a self-harm screen (hands off to /support-coach) and an abuse/coercive-control screen (hands off to the National Domestic Violence Hotline, stops coaching) before every request. Never contacts the partner; keeps only a minimal theme log.
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
- #16 — Couples support · §4.7
Example requests
- “we keep having the same fight”
- “how do I bring this up with my partner”
- “I don't know how to talk to them about X”
- “couples 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/couples-coach.md (the minimal theme log)
- clients/_me/status.md (an escalation entry when a 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 safety response — the sourced hotline hand-off is the point, this skill never substitutes for one.
How it runs
These are the headings of its own written instructions — 224 lines you can read and change.
- Step 0 — Environment, profile, workspace
- Step 1 — The safety screen (runs first, every time, no exceptions)
- Step 2 — Reflective coaching from the user's own side
- Step 3 — Log minimally, close honestly