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/citevahti --skill citevahti-releasegit clone --depth 1 https://github.com/heidihelena/citevahtiWrote 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/citevahti/citevahti-release)<a href="https://agentmods.dev/skills/heidihelena/citevahti/citevahti-release"><img src="https://agentmods.dev/badge/skills/heidihelena/citevahti/citevahti-release/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/citevahti/citevahti-release"><img src="https://agentmods.dev/badge/skills/heidihelena/citevahti/citevahti-release.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 | $0.00103 | $0.01607 |
| Opus 5 | $0.00051 | $0.00804 |
| Sonnet 5 | $0.00021 | $0.00321 |
| Haiku 4.5 | $0.00010 | $0.00161 |
Grade A, and why
citevahti-release 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 11d 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 — 110 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CiteVahti release — parity and shipping
This skill is the release checklist, not the release mechanics. The mechanics —
safety gate, offline build/test gate, signing, Trusted Publishing — live in the
secure-release skill (.claude/skills/secure-release/SKILL.md) and its canonical
docs (docs/RELEASING.md, desktop-extension/BUILD.md, docs/SAFETY_INVARIANTS.md).
Read secure-release and run its three gates in order; this skill adds the
production-push gates around them. Never reconstruct either from memory.
Triggers
Use when the maintainer asks to: cut/ship/tag a release, bump the version, check surface parity before shipping, write release notes, mint a DOI for a release, or prepare rollback notes.
Do NOT use for: day-to-day development (no release intent), or the accuracy
measurement itself (citevahti-eval — this skill only consumes its verdict).
Release order — gates before mechanics
Gate 0 — the eval gate (production-push addition)
Once acceptance thresholds are pre-registered in validation/claimcheck/, every
release starts by checking the citevahti-eval verdict: a scored ledger at or above
every pre-registered floor, measured on (or unchanged since) the code being released.
- No scored ledger, or any floor missed → stop. No release. File the gap.
- Pass → record the numbers; they ship with the release notes and the eval-results page.
(Until thresholds are first pre-registered, note explicitly in the release notes that the release predates the accuracy gate — don't let the gap ride silently.)
Gates 1–3 — safety, build/test, release/update
These are secure-release §1–§3, unchanged and in order:
- Safety gate — did the change weaken an invariant? (
docs/SAFETY_INVARIANTS.md) - Build & test gate — full offline suite,
pytest -m security, VS Code compile, artifact-contents checks, smoke-run of frozen artifacts. - Release & update gate — version lockstep across all nine tracked
version-bearing files: seven hand-edited (
pyproject.toml,src/citevahti/__init__.py,vscode-extension/package.json,desktop-extension/manifest.json,desktop-extension/manifest.binary.json,.claude-plugin/plugin.json,server.json— two slots) plus two regenerated lockfiles (vscode-extension/package-lock.json,uv.lock),CHANGELOG.mdsection +docs/STATUS.mdheader, PR off a branch with explicit staged paths, the five required checks (pytest py3.10 / py3.12, VS Code compile, ruff, mypy — read the live list from branch protection, not from memory), squash-merge,gh release create vX.Y.Z→ Trusted Publishing + desktop builds (the only two release-triggered workflows — MCP registry, Marketplace, and Open VSX are separate manual publishes), then confirm the published version actually reports itself.
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.
- 11d ago First seen · 110 lines · 103 tokens per session scan A 2d62bd670347
citevahti-release is a skill published in the GitHub repository heidihelena/citevahti (1 stars, last pushed 6d ago), licensed Apache-2.0. It adds 103 tokens to every session and 1,607 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
release-jetbrains
Use when releasing the Kilo JetBrains plugin -- resolve a version ("next rc" or explicit), run the prepare workflow, edit and commit a filtered human-readable changelog on the release PR, then watch publish to completion.
release-nyx
Ship a new nyx-local-ai release end to end — bump versions consistently, run the quality gates (typecheck, smoke tests, package), install locally, tag and push so CI publishes the installer artifacts. Use when the user asks to release, ship, publish, bump the version, or cut a new build of the extension.
refactor-research-code
Clean research code for public release while preserving reported results. Uses releaseaudit.py to separate safe mechanical cleanup from ask-first behavior risks and identity leaks. Use for release-ready repos, dead experiment branches, config extraction, documented entrypoints, and double-blind code cleanup.
finishing-a-development-branch
Presents options for merge, PR, or cleanup. Use when work is complete, tests pass, and you must decide how to integrate.
release-checklist
Walk the pre-release steps in order and stop at the first one that fails.
bump
Use when bumping Wheeler version after shipping features. Trigger when user says 'bump version', 'release', 'cut a release', 'version bump', 'update version', or after a set of features have been committed and the user wants to formalize a new version.