From human perspective

SkillAI & models

Use when a user wants an answer only from the human seat: what a person can love, trust, and tolerate. Not for rebuilding from primitives: use from-first-principle.

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 From human perspective skill

What this skill tells your AI

The instructions your AI receives, as published by outlinedriven/odin-claude-plugin in plugins/odin-critique/skills/from-human-perspective/SKILL.md and read by ahel’s review.

Contract

FieldBound contract
TriggerUser wants an answer only from the human seat (what a person can love, trust, and tolerate).
AuthorityRead-only. No file, VCS, credential, paid, published, deployed, or remote mutation.
Side effectReturns a human-perspective analysis in chat; no other surface is touched.
DoneReturns a human-perspective answer without blending.

Not for

  • Rebuilding a design from primitives: use from-first-principle.
  • Blended multi-seat analysis: run each from-*-perspective seat independently and compare after.
  • Source or remote mutation: this skill is read-only.

Inputs

  • The question or artifact under analysis (required).
  • Optional: named constraints the human seat should weigh, such as what a person can love, trust, or tolerate.

Procedure

  1. Take the question or artifact as the only input. Confirm the human seat is the requested perspective; if the request names a different seat, stop and report which seat was requested. Done when: the seat is confirmed as human.
  2. Answer strictly from what a person can love, trust, and tolerate: lived experience, attention cost, fatigue, trust erosion, tolerable friction, and what a person would keep or abandon. Done when: the answer is grounded in human-seat concerns.
  3. Do not blend other seats (money, codebase, impact, stability, moat, and the rest) into this answer. If another seat is relevant, name it as a separate lens to run later, never as part of this output. Done when: no other seat's reasoning appears in the answer body.
  4. Keep the output independent so it can be compared against other lens outputs after the fact; do not pre-merge them. Done when: the answer stands alone without cross-seat synthesis.
  5. Stop when the human-seat answer is complete. Do not widen scope, invent evidence, or import methodology from another perspective. Done when: the answer is complete and scope is unchanged.

Failure and recovery

  • Wrong-seat request: if the user asked for a different perspective, stop and report the requested seat; do not substitute the human seat.
  • Blending drift: if the draft starts importing another seat's concerns, discard the blended portion and re-answer from the human seat only.
  • Insufficient input: if the question or artifact is missing, stop and request it; do not fabricate a human-perspective read.
  • Non-convergence: if no human-seat answer can be formed from the given input, return that explicitly rather than emitting a blended or generic answer.
  • No mutation: nothing is written, committed, paid, published, deployed, or remotely changed; recovery is to re-issue the chat answer.

Output

A single human-perspective analysis answering the input from the human seat only, with any other relevant seats named as separate lenses to run later, no blending.

Signals

GitHub stars
35
Last commit
Sep 2026
Advanced
Catalog kind
skill
Gateway key
from-human-perspective
Source
github.com/outlinedriven/odin-claude-plugin