Holdback Experiment Design

What it does

When your agent uses it

• A team needs a long-term counterfactual
• Wants to measure delayed impact after launch
• Is worried about metric degradation over time
• Needs to decide holdback size

Example prompts

• “/holdback-experiment-design”

Requirements

• Compatibility (from SKILL.md): Codex, Claude Code, and other Agent Skills-compatible clients.

Workflow steps

6 steps, taken from the first numbered list in SKILL.md.

1. 1State why short-term A/B evidence is insufficient.
2. 2Choose degradation or long-term cumulative holdback.
3. 3Define who remains withheld, for how long, and from what experience.
4. 4Select long-term metrics and guardrails.
5. 5Estimate the user, business, and ethical cost of withholding.
6. 6Define monitoring cadence, exit criteria, and communication plan.

What it can do on your machine

Read from SKILL.md and the folder at commit 9e7b281. It shows what the files ask for, not the result of running them.

• Tool permissions

  Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

  From allowed-tools in the SKILL.md frontmatter.

• Runs code

  No scripts in the folder and no shell commands in SKILL.md (its code samples are markdown).

  From the folder's file list and the shell code blocks in SKILL.md.

• Network

  No URLs in SKILL.md.

  From URLs in SKILL.md, links to its own repository left out.

• Credentials

  Names no API keys, tokens, secrets or passwords.

  From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

• Compatibility

  Codex, Claude Code, and other Agent Skills-compatible clients.

  From compatibility in the SKILL.md frontmatter.

Context cost

Holdback Experiment Design loads about 736 tokens when it runs, and up to ~2k if it reads all its reference files. Until then it costs about 96 tokens; SKILL.md has 214 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it

~96

When it runs · the whole SKILL.md, loaded when a task matches

~736

With references · SKILL.md plus every file in references/, read only if the agent opens them

~2k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from hashgraph-online/awesome-codex-plugins at commit 9e7b281, republished under its MIT licence (© hashgraph-online). 214 words, ~736 tokens.

name

holdback-experiment-design

description

Design degradation holdbacks and long-term cumulative holdbacks for product experiments and feature rollouts. Use when a team needs a long-term counterfactual, wants to measure delayed impact after launch, is worried about metric degradation over time, needs to decide holdback size or duration, or must weigh the user/business cost of withholding a feature.

compatibility

Codex, Claude Code, and other Agent Skills-compatible clients.

license

MIT

metadata.version

0.1.0

metadata.displayName

Holdback Experiment Design

metadata.category

Product Management

metadata.tags

practical-ab-testing,ab-testing,experimentation,holdbacks,product-analytics

# Holdback Plan

## Purpose
[What long-term question this holdback answers.]

## Holdback Type
[Degradation | Long-term cumulative]

## Population And Duration
- Holdback population:
- Rollout population:
- Duration:
- Removal criteria:

## Metrics
| Metric | Role | Readout Cadence | Concern |
|--------|------|-----------------|---------|

## Cost Of Withholding
- User cost:
- Business cost:
- Ethical or trust concern:

## Decision Rules
- Continue holdback if:
- End holdback if:
- Escalate if:

© hashgraph-online, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 5 other files (references) in plugins/LVTD-LLC/skills/skills/holdback-experiment-design of hashgraph-online/awesome-codex-plugins.

• SKILL.md
• guidelines.md
• references/core/examples.md
• references/core/knowledge.md
• references/core/rules.md
• workflows/design-holdback.md

Open the folder on GitHubat commit 9e7b281