A repeatability check that runs the same analytics task in several fresh trials and measures how much the results agree. It measures consistency, not whether the results are correct.

Use it to compare repeated analyses, report exact or tolerance-based agreement, and investigate changes between runs.

It shows whether a system gives stable answers or changes from run to run, while keeping failed and unusable trials separate. This prevents a single run from hiding variation.

reliability — Skill for Claude Code by ai-analyst-lab/ai-analyst

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent

npx skills add ai-analyst-lab/ai-analyst --skill reliability

Clone the repo

git clone --depth 1 https://github.com/ai-analyst-lab/ai-analyst

Made for: Claude Code.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

README.md

[![agentmods](https://agentmods.dev/badge/skills/ai-analyst-lab/ai-analyst/reliability/github.svg)](https://agentmods.dev/skills/ai-analyst-lab/ai-analyst/reliability)

Your own site

<a href="https://agentmods.dev/skills/ai-analyst-lab/ai-analyst/reliability"><img src="https://agentmods.dev/badge/skills/ai-analyst-lab/ai-analyst/reliability/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

Your own site · 80×15

<a href="https://agentmods.dev/skills/ai-analyst-lab/ai-analyst/reliability"><img src="https://agentmods.dev/badge/skills/ai-analyst-lab/ai-analyst/reliability.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 41 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 730 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.

Origin original No closer match found in the catalogue.

Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

Model	Per session	Once invoked
Fable 5.1	$0.00041	$0.00730
Opus 5.5	$0.00016	$0.00292
Sonnet 5.5	$0.00008	$0.00146
Haiku 4.5	$0.00004	$0.00073

Measured 1mo ago against content hash c35b126a26a5, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

reliability scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.claude/skills/reliability/SKILL.md · 48 lines

How it starts

The opening of the file, as written. The whole thing — 48 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Reliability

What this answers

Does this named system configuration behave consistently on this task?

It does not answer whether the result is correct. A wrong analysis can repeat perfectly.

Run the evaluation

Record the task, model, active data, available tools, system fingerprint, and the tolerance that matters for the decision.
Use five fresh trials unless the user chooses another count. The trials must not share answers.
Preserve completed, failed, blocked, error, and unparseable trials separately.
Save the raw output from every trial.
Use the repository's deterministic reliability runner. It creates five independent workspaces, uses the active data source, preserves every result, and calculates the comparison.
Report exact agreement and agreement within the named tolerance separately.
State what changed if this is a comparison with an earlier run.

The deterministic CLI is available at python3 -m helpers.evals.cli run-reliability. Use it when the task can be executed noninteractively. For the standard run, explicitly pass --trials 5 --parallelism 5 --model claude-opus-4-6. Add --allow-code when the task requires a query. Do not lower parallelism preemptively or infer an account limit. Lower it only after the five-way run returns an actual concurrency or rate-limit error, and tell the user what failed before retrying.

Run the reliability command as one foreground task. Read its progress messages as trials finish. Do not create separate sleep commands or polling shells merely to wait for it.

The runner automatically gives each trial a clean view of the current analytical system. It excludes old outputs, test fixtures, future course examples, and inactive dataset packages. Connection credentials remain outside the workspace and are passed through the process environment. Do not ask the user to manage excluded paths or trial workspaces.

If the current system already contains a reviewed definition for the question, report that fact. Do not hide current context merely to manufacture variation.

Read the full file on GitHub · 48 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

1mo ago Changed · +4 lines c35b126a26a5
1mo ago Changed · +4 lines 05f3c3448cec
1mo ago Changed e45a9a44dfd0
1mo ago Changed · +2 lines fb6c135a99d2
1mo ago First seen · 38 lines · 41 tokens per session scan A e5c4e497b80a

Subscribe to this mod's changes

reliability is a skill published in the GitHub repository ai-analyst-lab/ai-analyst (322 stars, last pushed 1mo ago), licensed MIT. It adds 41 tokens to every session and 730 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-12.

Analytics and metrics

reliability next to 6 skills from other repositories
Mod	Type	Stars	Last push	Licence	Scan	Tokens	Works with
reliabilityai-analyst-lab/ai-analyst	Skill	322	1mo ago	MIT	A	41	Claude Code
imagegenopenai/codex	Skill	133k	yesterday	Apache-2.0	A	113	Codex
playwright-climicrosoft/playwright	Skill	99k	yesterday	Apache-2.0	A	19	Claude Code
local-ai-agentsmicrosoft/ai-agents-for-beginners	Skill	79k	4d ago	MIT	A	200	Claude Code, Codex
systematic-debuggingobra/superpowers	Skill	313k	3d ago	MIT	A	21	Claude Code
next-cache-components-adoptionvercel/next.js	Skill	143k	yesterday	MIT	A	95	Claude Code
next-partial-prefetching-adoptionvercel/next.js	Skill	143k	yesterday	MIT	A	103	Claude Code