Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/dwarvesf/dwarves-kitnpx agentmods add commands/dwarvesf/dwarves-kit/quiz-gateWrote 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/commands/dwarvesf/dwarves-kit/quiz-gate)<a href="https://agentmods.dev/commands/dwarvesf/dwarves-kit/quiz-gate"><img src="https://agentmods.dev/badge/commands/dwarvesf/dwarves-kit/quiz-gate/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/commands/dwarvesf/dwarves-kit/quiz-gate"><img src="https://agentmods.dev/badge/commands/dwarvesf/dwarves-kit/quiz-gate.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 | $0.00065 | $0.01313 |
| Opus 5.5 | $0.00026 | $0.00525 |
| Sonnet 5.5 | $0.00013 | $0.00263 |
| Haiku 4.5 | $0.00006 | $0.00131 |
Grade A, and why
quiz-gate 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 3d 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 — 85 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are the understanding-gate NUDGE (the AFTER gate's speed regulator). At the merge
boundary of a gate/gated-final PR, when a change is BOTH significant AND understanding-worthy, you offer
the human a 5-question quiz built from the actual change BEFORE they click-to-merge. $ARGUMENTS is the
change under merge (a commit / PR / spec ref) and its run id (<rid>).
This gates the human's ATTENTION, never the merge. It is a NUDGE: a waved change still merges; you never hard-block a correct build. The only real failure is UNTRACKED debt, so all three responses are recorded.
The two hard constraints (do not violate)
- Grounded in the DIFF + recorded test results, NEVER the agent's narrative (Litt's plausible-but-wrong
caveat). The questions come from
lib/gate/quiz-gate.sh questions <ref>, whose only input is a git ref , there is no narrative channel, so a false story physically cannot leak in. Never hand-write quiz questions from memory, the commit message, or what you "meant" to do. If the commit message contradicts the diff, the diff wins. - Route through the
understand.teachseam, do not build a second quiz. The kit builds the QUESTIONS and dispatches; whatever skillunderstand.teachnames runs the mastery gate (shuffled answer slots, per-item gate on a demonstrated answer). The kit scores nothing and names no skill itself.
Process
Step 1: Decide whether to tap (keyed on the significance-classify verdict)
Only a tap verdict (significant AND understanding-worthy) on a gate/gated-final PR is ever nudged , the
anti-fatigue guard. A significant-but-low-worthiness change (wave) or a not-significant change is NEVER
quizzed (it is already logged silently by significance-classify record, wired into /kit:ship
Step 8, immediately before this tap).
bash lib/gate/quiz-gate.sh tap <rid> --files "<changed files>" --impl-notes docs/implementation-notes/<slug>.md --pr-kind gate "<what changed>"
If it prints nothing, there is no tap , stop, the merge proceeds untouched. If it prints the nudge, present the one line ("★ worth understanding: ...") and the three choices to the human. Do not decide for them.
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.
- 3d ago Changed 20e71303591b
- 1mo ago Changed b1bc084b97b8
- 1mo ago Changed · +4 lines 38164051b17c
- 1mo ago First seen · 81 lines · 65 tokens per session scan A e2eb6df92217
quiz-gate is a command published in the GitHub repository dwarvesf/dwarves-kit (12 stars, last pushed today), licensed MIT. It adds 65 tokens to every session and 1,313 once invoked, about $0.0003 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 2026-08-30.
Other commands, from other repositories
impl-review
John Carmack-level implementation review via Codex, Copilot, Cursor, Claude or a host reviewer.
resolve-pr
Resolve PR review feedback — fetch unresolved threads, triage, dispatch resolver agents, reply + resolve via GraphQL.
review
Run /review — see .claude/commands/review.md for the authoritative spec.
pr-feedback-ingest
Turn PR feedback (Greptile, CI, bots, humans) into a traceable backlog, so the next step is a narrow implement-approved-slice. Corrective scope only. Use when a PR has review feedback to map to files and slices. Do not use for a new direction or contract (use post-review-pivot), no payload, or task-memory drift only…
argos
A command for checking whether an implementation matches its design deliverables. Its Korean description compares the work to the design as part of a completion inspection.
post-review-pivot
Capture what a PR or team review changed about the intended approach, separate keep from revert, and update task memory. Does not implement product code. Use when review changes direction (wrong fields or migrations) or a pivot digest is needed. Do not use for corrective feedback (use pr-feedback-ingest), when there…