This skill should be used to unblock a decision that needs data. Trigger with "measure this", "do users actually...", "check the numbers before we cut/keep", "did the change land". Runs a read-only measurement and returns a data-backed flip/keep/cut verdict — never guesses, never writes.
Scanned 8/30/2026
Install to Claude Code
npx -y skills add GiustoPiedimonte/agentic-engineering-marketplace --skill measure --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Measure?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/giustopiedimonte-measure)More formats (shields.io, HTML) on the badges page.
---
name: measure
description: >
This skill should be used to unblock a decision that needs data. Trigger with
"measure this", "do users actually...", "check the numbers before we cut/keep",
"did the change land". Runs a read-only measurement and returns a data-backed
flip/keep/cut verdict — never guesses, never writes.
metadata:
version: "0.2.0"
---
# /measure — turn a gated decision into a data verdict
When a decision is blocked on "what do the data actually say?", measure instead
of guessing. `$ARGUMENTS` is the question to answer.
## Process
1. **State the question precisely** and what verdict each outcome implies
(e.g. "if <1% of sessions hit this path -> cut; otherwise keep").
2. **Delegate to the `measurer` subagent** (or run directly if trivial). All
access is **read-only** — open databases read-only, query logs/metrics
without mutation. Never print secrets (no env dumps / token values).
3. **Report:** the question, the exact read-only command(s) run, the raw result,
and a clear verdict (flip / keep / cut) with the reasoning.
4. **If the data isn't ready** (the runtime window hasn't matured, or the metric
isn't populated), say so explicitly and state what window/data would settle
it. Do not invent a verdict.
Distinguish what the data shows from what you infer from it. A measure-gated
decision stays open until the measurement is real.
## Flip-gate for dark launches
This is the mechanism that flips a dark-launched change ON (see `/ship` §4 and
the `Flip-criteria` field in `/adr`). When a decision was merged dormant with a
recorded flip-criterion, `/measure` is what closes it: run the read-only check
against that exact criterion and window, then return flip / keep / cut. Flip
only on a real flip verdict — never on a green build or a synthetic check, which
validate the mechanism, not the magnitude.
Is this your skill, or is something wrong with this listing? . Author removals are honored within 72 hours.
No comments yet. Be the first to comment!