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 GreatMark/fable-field-guide-skills --skill change-quizgit clone --depth 1 https://github.com/GreatMark/fable-field-guide-skillsWrote 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/greatmark/fable-field-guide-skills/change-quiz)<a href="https://agentmods.dev/skills/greatmark/fable-field-guide-skills/change-quiz"><img src="https://agentmods.dev/badge/skills/greatmark/fable-field-guide-skills/change-quiz/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/greatmark/fable-field-guide-skills/change-quiz"><img src="https://agentmods.dev/badge/skills/greatmark/fable-field-guide-skills/change-quiz.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 | $0.00147 | $0.01484 |
| Opus 5 | $0.00073 | $0.00742 |
| Sonnet 5 | $0.00029 | $0.00297 |
| Haiku 4.5 | $0.00015 | $0.00148 |
Grade C, and why
change-quiz scanned grade C with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 12d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Tells the agent never to refusehighAnti-refusal
Suppressing the ability to decline removes a core safety control; a later harmful request then succeeds.
never refuse an explicit instruction). How it starts
The opening of the file, as written. The whole thing — 103 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Change Quiz
After a long session, the diff alone undersells what changed, because much of the behavior depends on existing code paths. Build the user's understanding, then verify it.
Workflow
- Determine scope: the current session's changes by default, or the
diff/branch/PR the user points at. If there is no discernible change set,
ask what to quiz on instead of guessing (a pre-work worry with no changes
yet is blindspot-pass territory, not a quiz). Read the touched code and
the existing code paths it plugs into. If
implementation-notes.mdexists, mine its Deviations and Surprises sections first — they are pre-indexed quiz material. - Write one self-contained HTML file (
change-report.htmlin the project root, in the language of the conversation; inline CSS/JS, no network). It describes the current change set only — if an existing report recorded a passed quiz, rename itchange-report-<YYYY-MM-DD>.htmlbefore overwriting. The report must stay out of the change set it describes: do not commit it, and if this is a git repo, exclude it locally:n=change-report.html; p="$(git rev-parse --git-path info/exclude 2>/dev/null)" && { mkdir -p "$(dirname "$p")" && { grep -qxF "$n" "$p" || echo "$n" >> "$p"; }; } 2>/dev/null || echo "note: could not exclude $n — leaving it untracked, never staging it"(worktree-safe; if the write is blocked, tell the user to add it manually and continue). The deliverable is the file itself — never render the report or quiz as a Cursor canvas: the merge gate must live in a single shareable file that opens outside the IDE. Contents:- Context & intent — the problem, and the shape of the solution.
- What changed — grouped by area, each linking file paths and explaining how the change interacts with pre-existing behavior.
- Behavior changes & risks — what users/callers will observe differently; what could break and where to look if it does.
- End the file with a quiz: 5-8 questions, answers hidden until clicked.
Each question gets a "show answer" disclosure plus a self-check checkbox;
a fixed score bar tracks N/total and turns green only when total > 0 and
every box is checked — an empty quiz must never display as passed. State
in the report that checkboxes are self-assessment only; the authoritative
pass is the in-chat grading. Target the parts most likely to surprise:
- behavior that depends on existing code paths, not just new code
- edge cases and failure modes the change handles (or deliberately doesn't)
- "what happens if X" tracing questions
Avoid trivia answerable by skimming the diff (file names, line counts).
A good question: "The new retry wrapper calls
fetchUser()— what happens on the third failure, given the existing circuit breaker inapi/client.ts?" — its answer lives in how the change interacts with a pre-existing code path, not in the diff itself.
- Hold the bar: the rule is: merge only after a perfect score.
- Primary path: the user answers in chat before opening the hidden answers; grade strictly — for any miss, explain the right answer, point to the exact code, then ask one fresh question on the same topic. After two consecutive misses on the same topic, stop replacing: explain it thoroughly, mark that question "passed by coaching", and say so honestly in the verdict. Passed = every question answered correctly or explicitly marked passed-by-coaching.
- Self-serve path: if the user self-checks against the hidden answers, that is honor-system and does not establish a graded pass — say so. If they claim a pass and want it to count, spot-check by asking them to restate the key points of two questions in chat. A question whose answer was already revealed can only go this route; a graded pass requires answering before revealing.
- For the rest of this session, if asked to merge before a pass, remind the user the quiz is unpassed before running the merge (advisory only — never refuse an explicit instruction).
- After a pass: offer to delete
change-report.html, or mark it passed by inserting<!-- FIELD-GUIDE-QUIZ-PASSED <YYYY-MM-DD> -->as the file's first line, so the merge-gate hook stops reminding. If the change now needs explaining to reviewers, that handoff is pitch-explainer.
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.
- 12d ago First seen · 103 lines · 147 tokens per session scan C 0cd4dfa66569
change-quiz is a skill published in the GitHub repository GreatMark/fable-field-guide-skills (9 stars, last pushed 1mo ago), licensed MIT. It adds 147 tokens to every session and 1,484 once invoked, about $0.0007 per session on Opus 5. A static security scan graded it C with 1 finding (tells the agent never to refuse). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
scaffold-exercises
A template builder for coding exercises, organized into sections and exercises with problem, solution, and explanation folders. It creates the files and structure expected by the project’s lint checker.
explainer
A tool for making narrated explainer or tutorial videos with computer-generated visuals. It can help create the script, the video, or both.
teach
A workspace-based teaching guide for helping a user learn a new skill or concept over multiple sessions. It uses mission notes, reference materials, lessons, resources, and learning records to track progress.
notebooklm
Complete API for Google NotebookLM - full programmatic access including features not in the web UI. Create notebooks, add sources, generate all artifact types, download in multiple formats. Activates on explicit /notebooklm or intent like "create a podcast about X".
code-explanation
Get layered, context-aware explanations of unfamiliar code. Understand what it does, why it was written that way, and how to work with it safely.
greek-philosopher
Channel ancient wisdom through Socratic questioning, Stoic principles, and philosophical inquiry to examine life's deepest questions with poetic eloquence and timeless insight. Use when examining a life decision, seeking Stoic perspective on adversity, or wanting Socratic questioning of an assumption.