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 capstonegit 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/capstone)<a href="https://agentmods.dev/skills/outlinedriven/outline-driven-development/capstone"><img src="https://agentmods.dev/badge/skills/outlinedriven/outline-driven-development/capstone.svg" alt="Measured on agentmods" 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 | $0.00027 | $0.01060 |
| Opus 5 | $0.00014 | $0.00530 |
| Sonnet 5 | $0.00005 | $0.00212 |
| Haiku 4.5 | $0.00003 | $0.00106 |
Grade A, and why
capstone 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 — 42 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Capstone
Contract
| Field | Bound contract |
|---|---|
| Trigger | The learner is ready to apply cleared concepts in a real project and invokes this skill. |
| Authority | Reversible local: writes only capstone-<slug>.md at the workspace root; rollback is deleting the brief file. No remote mutation. |
| Side effect | Creates or updates capstone-<slug>.md at the workspace root and nothing else; never writes progress lines or edits any progress record. |
| Done | The brief exists at capstone-<slug>.md and follows the format; after the learner returns, the work is judged per item against the brief; concepts the work does not demonstrate are named for practice. |
Inputs
PROGRESS.mdat the workspace root listing cleared concepts. Required for scoping; a concept counts as cleared only when its independent rung passed with no hints.- The domain or goal the project serves. Required; take it from the invocation argument when the argument carries it, otherwise ask once.
- The learner's finished work. Required only for the review stage, which happens when the learner returns with it.
Procedure
- Scope one project. Read
PROGRESS.mdand collect the cleared concepts. IfPROGRESS.mddoes not exist, ask once what the learner can already do without help, treat the answer as the cleared set, and do not ask again. Done when: the cleared concept set is collected from PROGRESS.md or the learner's one answer. - Bound the project before writing anything. It may use only cleared concepts plus at most one unfamiliar concept; with two or more unfamiliar concepts it is a tutorial, so cut scope until at most one remains. The deliverable must be something runnable, readable, or usable; anything smaller is an exercise, so say so and stop rather than write a brief. Done when: the project is bounded to cleared concepts plus at most one unfamiliar, and the deliverable is runnable, readable, or usable.
- Order the milestones so each one leaves something that works. Done when: milestones are ordered with each leaving a working intermediate.
- Write
capstone-<slug>.mdat the workspace root,<slug>derived from the project name, carrying these things: the concepts exercised (the cleared set plus the one unfamiliar concept, if any), the ordered milestones, and the done test. The done test is observable by someone else; "Understand X" is not a done test; state a result another person can check. Done when: the brief file exists with concepts, milestones, and an observable done test. - Yield to the learner. After the brief is written, stop and hand the learner the brief file path and the done test. The learner leaves to do the project; no further work happens in this session. The review stage begins only when the learner returns in a later turn or session with finished work. Done when: the brief path and done test are handed to the learner and the turn ends.
- Review on return. When the learner brings the finished work back, judge it against the brief's done test and concept list item by item, one verdict per concept and per milestone, no overall impression in place of items. Done when: every concept and milestone has a per-item verdict.
- Name every concept the work does not demonstrate as not cleared and give that list to the learner as practice items. Record nothing: this skill writes no progress lines, and ownership of any progress record stays outside it. Done when: the not-cleared concepts are named and handed to the learner as practice items.
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 · -40 tokens per session 1609a5bf91f3
- 4d ago First seen · 42 lines · 67 tokens per session scan A 9970acf11e52
capstone is a skill published in the GitHub repository OutlineDriven/outline-driven-development (52 stars, last pushed 2d ago), licensed Apache-2.0. It adds 27 tokens to every session and 1,060 once invoked, about $0.0001 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
book-to-skill
Use when the user names one book, course, paper, or source document and asks to distill it into a reusable skill. Classifies the source as procedure or reference, writes an attributed SKILL.md, and proves its routing. Not for a folder of sources — use map-corpus.
drill
Use when a concept needs practising rather than explaining: run a scaffolded exercise from worked example to independent problem, quiz the learner, run spaced recall over what they cleared, or probe for the gaps blocking what they want next. For explanation, use explain-concept; for an end-to-end build, use capstone.
writing-lean-proofs
Use when asked to design, write, review, refactor, lint, or performance-diagnose Lean 4 proofs, libraries, or tactics. Not for non-Lean code or changes outside source, API, proof, or linter scope.
scaffold-exercises
Use when a course needs numbered problem, solution, and explainer scaffolds. Creates type-specific files with non-colliding names, populates them, and validates with the course linter. Not for a CLI or Next.js project scaffold — use scaffold-cli or scaffold-nextjs.
explain-concept
Use when a concept needs making clear rather than practising: build the intuition, explain why it exists and what it replaced, trace where it came from, draw it, or contrast it against what it gets confused with. For exercises and grading, use drill.
capstone
Use when the user is ready to apply what they learned to something real: scope a project sized to what they have cleared, write the brief with milestones and a done test, then judge the finished work against it. For single-concept exercises and quizzes, use drill.