Awkward Message Helper

SkillCommunication

Draft the hard personal message you keep putting off — chasing money a friend owes, backing out of plans, checking in after a fight, following up on an unanswered text. Use when asked to help send an awkward text, how to say something uncomfortable to a friend, word a difficult personal message, or bring up something touchy. Produces a couple of calibrated options, the one line to open with, and the send/wait/call judgement call — warm, honest, and not a doormat.

Available today. Use it from your connected AI after setup.

Connect ahel once, and every AI you use reads what you have installed.

Then ask your AI: use the Awkward Message Helper skill

What this skill tells your AI

The instructions your AI receives, as published by mohitagw15856/pm-claude-skills in skills/awkward-message-helper/SKILL.md and read by ahel’s review.

The message sits in your head for days because any version feels wrong — too confrontational, too soft, too much. This gives you a few options that are honest without being harsh, name the thing directly (avoidance makes it worse), and respect the relationship and your own position. It also tells you when a text is the wrong medium and you should call.

What This Skill Produces

  • 2–3 options across a warm→direct range, so you pick the one that fits
  • The opener — the first line that sets the right tone (most of the awkwardness is in how you start)
  • The medium call — text vs. call vs. in-person, and why
  • The boundary line — if you're being taken advantage of, the version that holds your ground kindly

Required Inputs

Ask for these if not provided:

  • The situation — what happened and what you need to say/ask
  • The relationship — how close, and the history that matters here
  • Your goal — get the money, exit the plan, repair the rift, get a reply — the message bends to the goal
  • Your read — is this a one-off or a pattern? (Patterns need firmer.)

Framework: Honest, Not Harsh

  1. Name it directly. Hinting drags it out; a clear, kind sentence ends the dread for both of you.
  2. Lead with the relationship, not the grievance (unless it's a pattern) — "I value us, and…" lowers defences.
  3. One ask, stated plainly. Bundling grievances muddies it; pick the thing that matters now.
  4. Own your part where it's real — a small honest acknowledgement disarms; a fake one backfires.
  5. Know when to stop typing. Anything with real emotional charge is often a call, not a paragraph.

Output Format

[Situation] · to [relationship] · goal: [x]

Best medium: text / call / in person — why.

Option A (warm):Option B (direct):Option C (holds the boundary, if needed):

Open with: "…" Don't: [the move that would make it worse here]

Quality Checks

  • The message names the actual thing, no vague hinting
  • Options span warm→direct so the user can match their read
  • The medium recommendation fits the emotional charge (high charge → call)
  • If it's a pattern, a boundary-holding option is included
  • Honest without being cruel; self-respecting without being cold

Anti-Patterns

  • Burying the point in so much cushioning the person misses the ask.
  • Kitchen-sinking every past grievance into one message.
  • Doormat mode — softening to the point of giving up a fair ask.
  • Recommending a text for something that clearly needs a voice.
  • A fake apology to smooth it over — reads as manipulation.

Example Trigger Phrases

  • "How do I ask a friend to pay me back the $200 without it being weird?"
  • "I need to back out of a trip I already said yes to — help me word it."
  • "We had a fight last week; help me check in first."
  • "He read my message and didn't reply — what do I send?"
  • "My roommate keeps skipping their chores — how do I bring it up?"

Signals

GitHub stars
1k
Forks
239
Last commit
Sep 2026
Advanced
Catalog kind
skill
Gateway key
awkward-message-helper
Source
github.com/mohitagw15856/pm-claude-skills