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.
npx skills add mehdiforoozandeh/crux --skill crux-rdgit clone --depth 1 https://github.com/mehdiforoozandeh/cruxWrote 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.
[](https://agentmods.dev/skills/mehdiforoozandeh/crux/crux-rd)<a href="https://agentmods.dev/skills/mehdiforoozandeh/crux/crux-rd"><img src="https://agentmods.dev/badge/skills/mehdiforoozandeh/crux/crux-rd/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.
<a href="https://agentmods.dev/skills/mehdiforoozandeh/crux/crux-rd"><img src="https://agentmods.dev/badge/skills/mehdiforoozandeh/crux/crux-rd.svg" alt="Reviewed on agentmods" width="80" height="20"></a>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 | $0.00177 | $0.01705 |
| Opus 5.5 | $0.00071 | $0.00682 |
| Sonnet 5.5 | $0.00035 | $0.00341 |
| Haiku 4.5 | $0.00018 | $0.00170 |
Grade A, and why
crux-rd 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.
How it starts
The opening of the file, as written. The whole thing — 131 lines — stays where its author put it; the contents beside it link to each section on GitHub.
crux-rd — the design document a node cannot hold
crux caps a node's prose at 400 words. That cap exists because nodes measured in the thousands of words stopped being readable by the PI they exist to serve. But a cap with no outlet just makes prose denser. This is the outlet.
An RD is where design detail goes to be readable, not to hide. The node's TL;DR must still stand alone: a reader who never opens the RD still knows what is being asked and what would settle it. If moving text into an RD makes the node unintelligible, the move was wrong.
When you are invoked
Only when asked. This skill ships disable-model-invocation: true and means it. The
design being written up lives in the conversation that just happened with the PI — that
is the whole reason this is a skill and not an agent. From a cold start you would either
restate the node or re-derive the design badly.
So: no interview, no research phase, no "shall I write this up?" offered unprompted. When the PI says the design has settled, you synthesize what was already discussed.
The write-vs-skip filter — apply this before writing anything
Without an explicit filter, every node grows an RD, and the vault has simply moved its bloat one directory over. An RD is warranted when at least one of these holds:
- The design would exceed the 400-word prose cap on its own. Not the node — the design.
If the whole thing fits in
## Idea / Hypothesisinside budget, it belongs there. - The design makes a choice a reader would otherwise re-litigate. There is a rejected alternative worth recording. Research reasoning lives in the branch that was not taken, and it is the half that gets lost.
- The design carries a known distortion that must travel with every result. A pooling scheme that biases small strata, an attenuation you accept, a proxy that is not the thing. If a reader of the findings needs to know it, it has to be written somewhere durable.
Explicitly not warranted: a hypothesis whose design is its verifiables. That is the common case. Pre-registered checks with a metric, a baseline and a threshold are already a complete design, already structured, and already exempt from the cap. Writing an RD for one adds a file and no information.
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 First seen · 131 lines · 177 tokens per session scan A bb19a35dddeb
crux-rd is a skill published in the GitHub repository mehdiforoozandeh/crux (18 stars, last pushed today), licensed MIT. It adds 177 tokens to every session and 1,705 once invoked, about $0.0007 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 2026-08-31.
Other skills, from other repositories
ab-testing
Use when designing or analyzing a controlled experiment — falsifiable hypothesis, sample size from an MDE, reading significance/CI/power, CUPED, or rescuing tests that won't go significant. NOT recurring metric tracking (that is analytics), NOT north-star/KPI trees (that is kpi-framework), NOT projecting metrics…
research-manager
Use this when starting a new research project or managing a complex, multi-step research workflow.
academic-writing
You must use this when producing any research prose — literature reviews, syntheses, analyses, methodology descriptions, discussion sections, abstracts, or any written output intended for an academic audience.
ab-testing-ecommerce
Run controlled experiments on product pages, checkout flows, and pricing to find what converts best using statistical significance testing.
systematic-review
You must use this when conducting PRISMA-standard systematic reviews, protocol development, or Risk of Bias assessment.
hypothesis-testing
You must use this when formulating testable hypotheses, designing experimental controls, or defining falsification criteria.