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 heidihelena/vahtian --skill methodvahtigit clone --depth 1 https://github.com/heidihelena/vahtianWrote 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/heidihelena/vahtian/methodvahti)<a href="https://agentmods.dev/skills/heidihelena/vahtian/methodvahti"><img src="https://agentmods.dev/badge/skills/heidihelena/vahtian/methodvahti/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/heidihelena/vahtian/methodvahti"><img src="https://agentmods.dev/badge/skills/heidihelena/vahtian/methodvahti.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 | $0.00104 | $0.02973 |
| Opus 5 | $0.00052 | $0.01486 |
| Sonnet 5 | $0.00021 | $0.00595 |
| Haiku 4.5 | $0.00010 | $0.00297 |
Grade A, and why
methodvahti-design-defensibility 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 9d 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 — 203 lines — stays where its author put it; the contents beside it link to each section on GitHub.
MethodVahti — design defensibility for agents
Skill v1.1.0 · prompt_version 2 · compatible tools:
methodvahti≥0.4.0 (Python:classify_defensibility,render_report,sampling_heterogeneity_score),optimise.mjs(JS twin),construct_check.py. Full validation framework:VALIDATION.mdin this directory. Read Ch. 0.3 before you claim anything.
You are the intelligence; MethodVahti is deterministic and local. Your job is to help a researcher arrive at a sample size they can defend in a methods section and to a reviewer — and to make the methodological commitments behind it visible, in their language, so they can own them.
The one rule above all
MethodVahti describes a design profile. It never says a sample is adequate. "Adequate" is a judgement about a specific study answering a specific question, and it belongs to the researcher. You surface the profile and name its weakest point. You never certify.
The rule you state, and the override you must offer
There is no overall number. The overall judgement is an ordinal classification, and the rule is fixed — you state it, you do not ask the researcher to choose it:
The overall judgement cannot be more favourable than the least defensible dimension.
Accumulated concerns elsewhere may push it lower. Nothing pushes it higher. This is how appraisal
instruments already work (VALIDATION.md Ch. 1.2.4) — you are not applying a Vahtian invention, and
saying so plainly is usually enough for a researcher to accept it.
The researcher's authority is the override, not the rule. If they judge the classification wrong for their study, they may override it — and then you require a written justification and carry it into the report. A documented departure is defensible; a quietly different rule is not. Offer the override explicitly; do not wait to be asked.
Do not explain axioms, order statistics, invariance, or λ unless they ask. If they do ask, it is
in VALIDATION.md Ch. 1.2.3–1.2.4. The mathematics is your burden, not theirs.
What ships with it
34 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- .gitignore 70 B
- claim-check/index.html 25 KB
- codebook/index.html 28 KB
- construct_check.py 7.9 KB runs code
- examples/methodvahti_heterogeneity_example.py 4.0 KB runs code
- examples/methodvahti_pdf_example.py 3.8 KB runs code
- explore/index.html 31 KB
- fixtures/golden.json 4.7 KB
- index.html 30 KB
- LICENSE 11 KB
- MANIFEST.in 126 B
- methodvahti_pdf.py 35 KB runs code
- methodvahti/__init__.py 1.2 KB runs code
- methodvahti/assets/fonts/LiberationMono-Regular.ttf 312 KB
- methodvahti/assets/fonts/LiberationSans-Bold.ttf 405 KB
- methodvahti/assets/fonts/LiberationSans-BoldItalic.ttf 400 KB
- methodvahti/assets/fonts/LiberationSans-Italic.ttf 406 KB
- methodvahti/assets/fonts/LiberationSans-Regular.ttf 401 KB
- methodvahti/assets/fonts/LICENSE-OFL.txt 5.3 KB
- methodvahti/defensibility.py 10 KB runs code
- methodvahti/heterogeneity.py 26 KB runs code
- optimise.mjs 3.3 KB runs code
- pyproject.toml 1.1 KB
- README.md 6.0 KB
- sample-report.pdf 84 KB
- tests/test_construct_separation.py 1.8 KB runs code
- tests/test_defensibility.py 8.5 KB runs code
- tests/test_heterogeneity.py 11 KB runs code
- tests/test_optimise_wiring.py 2.2 KB runs code
- tests/test_parity.py 3.9 KB runs code
- tests/test_pdf.py 4.3 KB runs code
- tests/test_properties.py 7.5 KB runs code
- tools/parity_emit.mjs 1.5 KB runs code
- VALIDATION.md 89 KB
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.
- 9d ago First seen · 203 lines · 104 tokens per session scan A 4d027ea85151
methodvahti-design-defensibility is a skill published in the GitHub repository heidihelena/vahtian (1 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 104 tokens to every session and 2,973 once invoked, about $0.0005 per session on Opus 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
lit-review-assistant
Search, summarize, and synthesize economics literature.
ml-paper-writing
Write publication-ready ML/AI papers for NeurIPS, ICML, ICLR, ACL, AAAI, COLM. Use when drafting papers from research repos, structuring arguments, verifying citations, or preparing camera-ready submissions. Includes LaTeX templates, reviewer guidelines, and citation verification workflows.
link-evidence
Prevents fabricated URLs. Only use links that appear in the sources pool or are provided by the user.
ml-paper-writing
Write publication-ready ML/AI papers for NeurIPS, ICML, ICLR, ACL, AAAI, COLM. Use when drafting papers from research repos, conducting literature reviews, finding related work, verifying citations, or preparing camera-ready submissions. Includes LaTeX templates, citation verification workflows, and paper…
lit-review-assistant
Search, summarize, and synthesize economics literature.
llmsr-discover
Use when the user wants to discover or fit a closed-form equation from a dataset via LLM-SR and ingest the result into the Wheeler knowledge graph.