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 OutlineDriven/outline-driven-development --skill drillgit clone --depth 1 https://github.com/OutlineDriven/outline-driven-developmentWrote 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/outlinedriven/outline-driven-development/drill)<a href="https://agentmods.dev/skills/outlinedriven/outline-driven-development/drill"><img src="https://agentmods.dev/badge/skills/outlinedriven/outline-driven-development/drill/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/outlinedriven/outline-driven-development/drill"><img src="https://agentmods.dev/badge/skills/outlinedriven/outline-driven-development/drill.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00034 | $0.01466 |
| Opus 5 | $0.00017 | $0.00733 |
| Sonnet 5 | $0.00007 | $0.00293 |
| Haiku 4.5 | $0.00003 | $0.00147 |
Grade A, and why
drill 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 2d 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 — 47 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Drill
Contract
| Field | Bound contract |
|---|---|
| Trigger | give me an exercise, quiz me, keep it fresh, what am I missing / where are my gaps, or explicit practice/quiz/recall/probe |
| Authority | Reversible local: writes only PROGRESS.md at the workspace root; rollback is deleting the lines this run appended. No remote mutation. The file is append-only, never rewritten. |
| Side effect | Appends one line per attempt to PROGRESS.md at the workspace root; drives spaced recall and probes gaps. |
| Done | The named move completed; every attempt recorded; cleared concepts marked no-hints-independent. |
Inputs
- A concept name (required) and a move (optional):
practice,quiz,recall, orprobe. - The move is optional: the routing table picks it, an explicit argument overrides, and no match defaults to
practice. - Optional: item count for
quiz(default 5–8). proberequiresCORPUS.mdat the workspace root.recallrequires priorPROGRESS.mdhistory. State which is missing rather than inventing it.
Procedure
- Read
PROGRESS.mdat the workspace root (create it on first write) and, forprobe,CORPUS.md. If neitherPROGRESS.mdnorCORPUS.mdexists, say once that nothing is being recorded, ask which concept to work on, and run in-session only. Done when:PROGRESS.mdis read or confirmed absent, and forprobe,CORPUS.mdis read or confirmed absent. - Pick the move. An explicit argument overrides the table; no match means
practice. Table: "give me an exercise" / "let me practise X" / "walk me through one" →practice; "quiz me" / "test me on X" →quiz; "what should I review" / "keep it fresh" →recall; "what am I missing" / "where are my gaps" / "am I ready for X" →probe. Done when: the move is picked and stated. - Run one move per run. Another move is another run. Done when: one move is selected for this run.
practice: three rungs in this order, one rung per run: worked (the agent solves it and narrates why each step is taken; done when the full solution is narrated and the learner has answered one self-explanation prompt); faded (the agent solves everything but the last step and the learner finishes it; each further faded run removes one more step, always from the end backward, so the learner performs the final step every time; done when the learner has produced the step the rung asks and the attempt is recorded); independent (the learner solves it whole; done when the learner has produced the step the rung asks and the attempt is recorded). Advance a rung only when the current one was cleared without hints. Done when: the current rung is completed and recorded.quiz: five to eight items unless the user asks for a different count, mixed between recall and application, over concepts already explained or cleared. The learner commits to an answer before anything is revealed. Done when: every item is graded and each miss names the misconception it reveals.recall: a concept is due on an interval ladder counted from the date its independent rung was cleared: 1 day, 3 days, 7 days, 21 days, then every 60 days. Due means today is at or past the next interval with no retrieval recorded since. A missed retrieval resets that concept to the start of the ladder. Interleave across cleared concepts only. Two consecutive misses on one concept send it back to a faded rung. Done when: every due concept has been retrieved once and recorded.probe: take what the learner says they want to do next, walkCORPUS.mdback through the concepts it needs, ask one discriminating question per concept, then rank the confirmed gaps by how many downstream concepts each blocks. Done when: the ranked list is reported and every confirmed gap is recorded.- Hints: three tiers, released one at a time and on request: nudge (names the concept in play, no structure); strategy (the shape of the solution, no answer); bottom out (the step itself). Once a concept's independent rung is cleared with no hints, withdraw hints for that concept. Done when: hints are released on request and withdrawn for cleared concepts.
- Grading: name the specific misconception an answer reveals rather than scoring it right or wrong. Say what the answer got right before what it missed. Release the solution once the learner has committed to an attempt. Done when: each answer is graded with its misconception named and the solution released.
- Record every attempt in
PROGRESS.mdwith this line shape:- <YYYY-MM-DD> · <move>[/<rung>] · <cleared|missed>, hints <none|1|1-2|1-3>[, <misconception>]. A concept is cleared when itspractice/independentline readscleared, hints none. Missed concepts are not cleared. Use the words rung and cleared and no synonyms. Done when: every attempt is recorded inPROGRESS.mdwith the correct line shape.
What ships with it
2 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.
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.
- 2d ago Changed · -15 tokens per session 5574d2d352be
- 5d ago First seen · 47 lines · 49 tokens per session scan A d9dc90c3bfb4
drill is a skill published in the GitHub repository OutlineDriven/outline-driven-development (52 stars, last pushed 3d ago), licensed Apache-2.0. It adds 34 tokens to every session and 1,466 once invoked, about $0.0002 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 2026-09-03.
Other skills, from other repositories
read-book
When you want to read and extract structured notes from a book — PDF, EPUB, MOBI, markdown, .txt, pasted text, or URL to a public-domain work. Reads in chunks (by chapter when a TOC exists, by 50-page blocks otherwise), extracts per-chapter TL;DR + key concepts + quotes + action items + frameworks, and offers to…
xray
Investigate how a concept, code path, application, network flow, system, incident, document, or local artifact actually works, then deliver a two-depth visual HTML explainer with a dead-simple big-picture first layer and source-backed technical depth behind it. Use when the user invokes $xray, asks how something works…
stage-and-commit
Stage changed files and create a commit following the repo's translation, formatting, and git workflow rules. Use when the user asks to stage, commit, or both.
autopilot
Run a hands-off plan-to-ship pipeline by chaining existing skills. Use when the user says "autopilot", "take this from plan to shipped", "run the whole pipeline", "hands-off ship it", or "do the end-to-end build".
voice-builder
Ingests 5–20 of the user's writing samples and distils a reusable voice-profile.md — signature phrases, sentence-length distribution, opener/closer habits, punctuation quirks, vocabulary do/don't lists, tone sliders, and three calibration paragraphs with self-checks. Other skills load this file so every post sounds…
hatch3r-learn
Captures learnings from completed development sessions into reusable knowledge files for future consultation. Invoke manually, from board-pickup after PR merge, or with a specific issue number for targeted reflection.