Skip to main content
PromptMetricsPromptMetrics
Book a free call
← All open source
Skill

Walk into any room as an equal at the table.

The Human Layer coaches you through a real workplace conversation: feedback, pushback, a hard client or team talk. Ranya ran weekly soft-skills coaching for her operations team for years before she turned the method into a skill. It plays your counterpart relentlessly, debriefs you honestly, and writes every session to a Growth Log. Deliberately uncomfortable; that discomfort is the method working.

MarkdownMITClaude Code · Claude.ai · Coworkgithub.com/promptmetrics/coach-skill
View on GitHubOpen an issue
A pure conversational skill: no scripts, no secrets, no MCP. The same folder runs unchanged on Claude Code, Claude.ai, and Cowork.
Install

Claude Code

Also auto-triggers on coaching-shaped questions; no slash command needed.

claude
> /plugin marketplace add promptmetrics/coach-skill
> /plugin install promptmetrics-coach@promptmetrics
> /reload-plugins
✓ /promptmetrics-coach:the-human-layer

Claude.ai / Cowork

Requires Code execution and file creation enabled under Settings ▸ Capabilities.

bash
# get the-human-layer.zip from the latest release (or build it)
$ ./scripts/build-zip.sh
# Customize ▸ Skills → + Create skill → Upload a skill
What you get
  • Five-step method: Ground it → Name the pattern → Roleplay it → Debrief with teeth → Log it.
  • Your choice of framework: LAER alone, or sharpened with Never Split the Difference: mirroring, labeling, calibrated questions.
  • Art of War layer: opt in to make the roleplay counterpart tougher and less predictable.
  • Seven doors: seven ways in, including the accent, partner-not-vendor, reading the room, and the Client Ledger move.
  • Growth Log: every session logged to a Google Sheet tab when Drive is connected, else a markdown file.
  • Clear boundaries: workplace communication, not therapy; it stops the frame and points to real support when a situation is bigger than rehearsal.
How it works

One session, five steps, one outcome: seen, felt, and heard as an equal.

  1. 01

    Ground it

    Two or three sharp questions until it has the real, specific situation; vague situations produce vague roleplay.

  2. 02

    Name the pattern

    The dynamic at play, named in two or three sentences you can hold onto. You pick the framework.

  3. 03

    Roleplay it

    It plays the counterpart (relentless, not clever) and offers reusable sound bites when you get stuck.

  4. 04

    Debrief + log

    An honest read, a rerun of the moment that needs work, three tips for the week, all written to your Growth Log.

Try it
"My client keeps pushing scope and I fold every time."Auto-triggers the skill; grounding questions start immediately.
/promptmetrics-coach:the-human-layerInvoke it directly in Claude Code.
"Rerun the part where I apologized."The debrief re-tests the exact moment that needs work.

Questions or issues? Open one on GitHub, or browse the rest of the catalog.

All open source