Skip to content
Back to skills

Ref Signal Identification

ASecurity

Signal identification reference — the parse-time classifier with categories existing_sensor/instrument/proxy, the sensor economics gate, and the signal.eval.v1 output object the align-signal-assessor (no live agent definition today) and #2372 consume.

  • 3 stars
  • 0 votes
  • 0 copies
  • 4 views
  • Added September 3, 2026
ai-agentsgonode

Works with

  • cli

Security analysis

A100/100

Scanned September 3, 2026

npx -y skills add natb1/commons.systems --skill ref-signal-identification --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Ref Signal Identification?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Ref Signal Identification
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/natb1-ref-signal-identification/badge)](https://www.skillsdirectory.com/skills/natb1-ref-signal-identification)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: ref-signal-identification
description: Signal identification reference — the parse-time classifier with categories existing_sensor/instrument/proxy, the sensor economics gate, and the signal.eval.v1 output object the align-signal-assessor (no live agent definition today) and #2372 consume.
---
# Signal Identification Reference

`.claude/docs/signal-identification.md` is the single source of truth for the
signal-identification contract. This skill is the entry point an align-signal-assessor
(#2371 structural role, Unit 4; no live agent definition today) loads before
classifying an intention node. Everything
that defines categories, the economics inequality, field shapes, and invariants lives in
that doc — this skill points at it; it does not restate it.

## When to use

Invoke when deciding what observable says an intention is met and how to read it, and
whether building or wiring a sensor is worth the cost. The align-signal-assessor role
(no live agent definition today) loads this skill before each classification; the
resulting `signal.eval.v1` output is consumed downstream by #2372, which wires the
signal arm into the align loop.

## The contract

Read `.claude/docs/signal-identification.md` before classifying. If the doc cannot be
read, state so explicitly — do not proceed from memory.

Apply the steps in order:

1. Run the [signal-identification classifier](../../docs/signal-identification.md#signal-identification-classifier)
   — test `existing_sensor` first, then `instrument`, else `proxy`. Emit a `category`
   and a mandatory `rationale` naming which test decided it. For `existing_sensor` or
   `proxy`, set `economics_verdict: "n/a"` and `economics_rationale: null`, then skip
   step 2 and proceed to step 3.
2. When `category == instrument`, run the
   [WORTH economics gate](../../docs/signal-identification.md#worth-economics-sensor-roi)
   — evaluate the sensor ROI inequality and emit `economics_verdict` (`instrument` or
   `decline`) plus `economics_rationale`.
3. Emit the [output-contract](../../docs/signal-identification.md#output-contract)
   `signal.eval.v1` object with all required fields. An object missing any always-present
   field is invalid — do not coerce defaults; re-run instead.

## Perspectives

Two perspectives feed the economics estimate, each formerly a deleted `.claude/agents/`
agent definition; the verbatim pre-deletion source is at `origin/main` commit
`44493733` under `.claude/agents/`. The doc defines exactly how each contributes to
the ROI inequality; this skill only names them:

- **FINANCIAL** — formerly the deleted `align-financial` agent definition; estimates
  `build`, `run`, `decision_value`, and `decision_frequency`.
- **TECHNICAL** — formerly the deleted `align-technical` agent definition; estimates
  `maintenance` and holds a "safe to abandon" concern on a sensor that would rot if
  left untouched.

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.