tutor
Skill9 starsChanged 5 months agoMITView on GitHub
Learn a course in scheduled hour-blocks — the tutor hands over a reading list of the real paper/blog/Wikipedia links, teaches the session's module, suggests ingesting what you read, then (if you opt in) drills adaptive flashcards or trusts the honor system, tracks mastery, and advances. Third stage of a course.
28 other projects use a version of this.
Adopted by
Most recently taken up by
- salonidabgar/claude-skillstheir own version, first seen Nov 15, 2026
- agentjido/skillstheir own version, first seen Nov 15, 2026
- levi-qiao/obsidian-llm-wikitheir own version, first seen Nov 8, 2026
- Srltas/claude-skillstheir own version, first seen Oct 31, 2026
- hmjBill/skillstheir own version, first seen Oct 29, 2026
- Zografska/claude-skillstheir own version, first seen Oct 25, 2026
Used with
Projects with it also use defuddle, obsidian-cli and 10 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
HK-hubSkilldefuddle
Extract clean markdown content from web pages using Defuddle CLI, removing clutter and navigation to save tokens. Use instead of WebFetch when the user provides a URL to read or analyze, for online documentation, articles, blog posts, or any standard web page.
Also in 4 projects that use this
jykimSkillobsidian-cli
Interact with Obsidian using the official CLI to read, search, create, and manage notes, daily notes, tasks, properties, and more. Use when the user wants to query or manipulate vault content via the obsidian command.
Also in 4 projects that use this
wangzhuofengSkilljson-canvas
Create and edit JSON Canvas files (.canvas) with nodes, edges, groups, and connections. Use when working with .canvas files, creating visual canvases, mind maps, flowcharts, or when the user mentions Canvas files in Obsidian.
Also in 3 projects that use this
MicrosoftSkillexcalidraw-diagram
Create Excalidraw diagram JSON files that make visual arguments. Use when the user wants to visualize workflows, architectures, or concepts.
Also in 3 projects that use this
wangzhuofengSkillobsidian-bases
Create and edit Obsidian Bases (.base files) with views, filters, formulas, and summaries. Use when working with .base files, creating database-like views of notes, or when the user mentions Bases, table views, card views, filters, or formulas in Obsidian.
Also in 3 projects that use this
GoogleSkillteach
Teach the user a new skill or concept, within this workspace.
Also in 4 projects that use this
- +1 more from wangzhuofeng
Other versions28 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 28
- RoundTable02/tutor-skills★ 1.3ksame name
- bevibing/tutor-skills★ 1.2ksame name
- xiaotianfotos/skills★ 582same name
- codetesla51/skills★ 32same name
- ProjectAJ14/eklavya★ 29same name
- agentjido/skills★ 16same name
- JEFF7712/claude-tutor★ 16same name
- dewil/ai-toolkit★ 9same name
- dewil/claude-toolkit★ 9same name
- levi-qiao/obsidian-llm-wiki★ 9same name
18 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository14 other files it gives its agents.
- .claude/AGENTS.mdAGENTS.md
- .claude/CLAUDE.mdCLAUDE.md
- alphaxivSkill
- askSkill
- autoexploreSkill
- collateSkill
- curriculumSkill
- editSkill
- goalSkill
- notionSkill
- obsidianSkill
- retrieveSkill
- webSkill
- zoteroSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
17 rules. 0 are common, 3 are distinctive, and 14 are very specific to this project.
0
Common
Most of the 22 edited versions we read keep these.
3
Distinctive
This author's own choices. Worth a look.
14
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 21
If $ARGUMENTS is empty, glob course/*/goal.md, read each title, list the courses, and ask which to run.
- Very specific: line 28
Read course/<slug>/goal.md, plan.md, schedule.md, progress.md, the modules/ docs, and the course/<slug>/library/ files (the fetched …
- Very specific: line 38
If testing_mode is unset in goal.md frontmatter, ask how the learner wants their knowledge level updated:
- Very specific: line 46
Write the chosen value (test | honor) to goal.md frontmatter.
- Very specific: line 49
If tutor_prompt_mode is unset, also ask (governs /edit's re-test reminders):
- Very specific: line 56
Write session-end | after-3-edits | never back to goal.md.
- Very specific: line 73
Before teaching, give the learner the session's reading list of real sources to go read on their own (in Zotero / Notion / Obsidian) — the …
- Very specific: line 96
After teaching, suggest in plain prose (a short list — no AskUserQuestion or pop-up box) that the learner ingest the session's material …
- Very specific: line 135
review: every due item across all modules. - Very specific: line 136
drill: every item in the named module/topic, due or not. - Very specific: line 168
After scoring: show the relevant cited passage, give 1–2 sentences of feedback, and update the item's streak (confident → +1.
- Distinctive: line 171
When an item is also backed by an ingested personal-base page, you may draw the follow-up/feedback from that page too — the knowledge level …
- Very specific: line 177
If this session's module isn't mastered yet, recommend staying on it but let the user advance.
- Very specific: line 178
Don't surface a downstream module's items until its prerequisite module is mastered (locked).
- Distinctive: line 182
No flashcards. The learner has self-reported reading the session's sources and (Step 6) ingested them.
- Distinctive: line 210
If a module crossed the mastery threshold this session, add an end-of-module milestone recap: module mastery %, deepest Bloom reached, a …
- Very specific: line 273
When /edit finishes on a page it should respect the user's cadence:
“Common” is measured: kept by at least 6 in 10 of the 22 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Drive a course session by session for $ARGUMENTS, one scheduled hour-block at a time —
the third stage of the /goal → /curriculum → /tutor pipeline. Each session runs the steps
below: reading list → teach → suggest ingest → (test | honor) → recap → advance, tracking
mastery in course/<slug>/progress.md.
Modes (parse from $ARGUMENTS):
<slug>— the next scheduled session (reading list + teach + ingest + test/honor + recap). Default.<slug> hour <N>— run a specific scheduled hour-block instead of the next one.<slug> review— spaced-repetition only: skip the reading list, teaching, and ingest prompt; drill all due items (test track only).<slug> drill [module|topic]— on-demand flashcards for a module or sub-topic, ignoring due dates. Scores are still recorded.
If $ARGUMENTS is empty, glob course/*/goal.md, read each title, list the
courses, and ask which to run.
Step 1 — Load the course
Read course/<slug>/goal.md, plan.md, schedule.md, progress.md, the modules/
docs, and the course/<slug>/library/ files (the fetched resources — their
frontmatter holds the canonical URLs and credibility tiers you need for the reading
list). Extract the target, tutor_prompt_mode, testing_mode, deadline,
hours_per_*, the module map (the prerequisite chain), and the Item Progress + Module
Mastery tables. If plan.md/progress.md are missing, offer to run /curriculum <slug>
first.
Step 2 — Set preferences (first run only)
If testing_mode is unset in goal.md frontmatter, ask how the learner wants their
knowledge level updated:
How should I gauge your progress each session?
- Test me — drill adaptive flashcards and score me (mastery is measured)
- Honor system — no flashcards; I'll ingest only material I've actually read and you trust that as progress
Write the chosen value (test | honor) to goal.md frontmatter. It can be overridden
per session if the user asks.
If tutor_prompt_mode is unset, also ask (governs /edit's re-test reminders):
After you edit a page this course covers, when should I remind you to re-test?
- At session end — once per edit session, at the very end (recommended)
- After 3 edits — prompt after every 3 edits to covered pages this session
- Never — I'll run
/tutormanually
Write session-end | after-3-edits | never back to goal.md. /edit reads it —
see Tutor prompt trigger at the bottom.
Step 3 — Pick this session
(review and drill modes skip teaching and ingest — jump to Step 7.)
Get today's date (date +%Y-%m-%d) — used only for spaced-repetition intervals, not
to gate the schedule (the learner studies whenever they like). In schedule.md, find the
first un-ticked ## Week <W> · Hour <H> block; the module(s) it names are this
session's target. For hour <N>, use that block. If every block is ticked, tell the
user the schedule is complete and switch to review mode. From plan.md's Resources list
(and the module's sources: frontmatter), gather this session's resources — the
specific library files / links assigned to this hour-block's module.
Step 4 — Hand over the reading list
Before teaching, give the learner the session's reading list of real sources to go read on
their own (in Zotero / Notion / Obsidian) — the actual links, not the library/.md
files. For each resource, pull from its library/<file>.md frontmatter and present:
- Title + canonical link — the
url:(web) orsource_links:arXiv/blog URL, clickable. - Kind · credibility tier (paper / Wikipedia / blog · canonical / reputable).
- TLDR (one line) — why you recommend it and what the learner will get (e.g. "the original CFM objective + Gaussian/OT paths — read this for the derivation").
- Quick-browse — a pointer to the
library/<file>.mdsummary to skim if they don't want the full source now.
Frame it as go read and annotate these from the source; the summaries and lecture are your quick reference. Keep it tight — only this session's resources, ordered as the module uses them.
Step 5 — Teach
Present a tight, grounded synthesis of this session's scheduled module sections, read straight
from course/<slug>/modules/<n>-<slug>.md — this is the "lecture", and the quick-browse
companion to the reading list. Never invent: explain only what the cited module
content covers, in the session's ~1-hour budget. Offer to go deeper on any part before moving on.
Step 6 — Suggest ingesting into the personal base
After teaching, suggest in plain prose (a short list — no AskUserQuestion or pop-up
box) that the learner ingest the session's material into their personal base. For each
resource:
- Suggest the matching helper command using the resource's real title, not the library
slug —
/zotero "attention is all you need"(PDF + annotations),/alphaxiv …(arXiv), or/web <url>(Wikipedia/blog). The learner reads and annotates the source themselves, then ingests those annotations. - Dedup-check first: glob
papers//concepts//sources//notes/for a match. If one exists, suggest an update instead of a duplicate (re-ingest for fresh annotations, or re-run/collate//edit <page>) — naming the file it matches.
This is a suggestion, not a gate — continue to Step 7 without waiting, noting anything the learner says they ingested. (Invoke a helper only if they explicitly ask you to ingest for them.)
Step 7 — Test or honor (branch on testing_mode)
Track A — test: adaptive flashcards
Drill this session's items and update the knowledge level from both the curriculum material and what's now in the personal base (papers/sources/concepts/notes ingested in Step 6 or earlier).
7A.1 — Build the flashcard queue. Candidate items come from the Item Progress table.
Order them with streak-graded spaced repetition — an item is due when days since
Last tested ≥ its interval:
| Streak | Retest interval |
| 0 | 1 day |
| 1 | 3 days |
| 2 | 7 days |
| 3+ | 14 days |
Queue priority (first → last): wrong → partial → untested → confident & due.
Confident items not yet due are skipped. Within a tier, oldest Last tested first.
- This session: items from this session's module that are due, plus any resurfaced due items from mastered earlier modules.
review: every due item across all modules.drill: every item in the named module/topic, due or not.
Adapt to mastery (Bloom escalation). Each item carries a Bloom level
(remember → understand → apply → analyze → evaluate). Generate each flashcard live
from the module content at the item's current level — but for items from a mastered
module, pose the next level up (a remember definition becomes an apply/analyze
prompt). Tell the user the queue size (~3 min/item); they can type "done" anytime to end
early and still save scores.
7A.2 — Question loop. For each queued item, generate one flashcard from its module page, varying type across the session (don't repeat a type twice in a row):
| Type | When | Example |
| Definition | first-time / remember | "Define a probability path $p_t$ precisely." |
| Mechanism | a described process | "Walk through how the CFM loss is computed." |
| Comparison | two related items | "How does the OT path differ from the diffusion path?" |
| Application | partial/revisit/apply+ | "When would you prefer rectified flow, and why?" |
| Maths | pages with equations | "Write $\mathcal{L}_\mathrm{CFM}$ and explain each term." |
For maths, quote the LaTeX so the precision expected is clear. Accept: an answer → score it; "skip" → no score change, move on; "show" → display the cited passage, move on without scoring; "done" → end the session.
Score each answer:
| Score | Criterion |
confident | Correct and precise; no significant gaps |
partial | Mostly right but missing a key detail or a minor error |
wrong | Missed the core idea or significantly incorrect |
After scoring: show the relevant cited passage, give 1–2 sentences of feedback, and
update the item's streak (confident → +1; partial/wrong → reset to 0) and its
Bloom reached (raise it when the learner answers a higher-level prompt correctly).
When an item is also backed by an ingested personal-base page, you may draw the
follow-up/feedback from that page too — the knowledge level reflects curriculum and
personal base.
7A.3 — Mastery gating (advisory). A module is mastered once ≥80% of its items are
confident, each with streak ≥2 (the canonical rule lives in .claude/src/curriculum_utils.py).
If this session's module isn't mastered yet, recommend staying on it but let the user advance.
Don't surface a downstream module's items until its prerequisite module is mastered (locked).
Track B — honor: trust what was read
No flashcards. The learner has self-reported reading the session's sources and (Step 6) ingested them. Update the knowledge level from the new personal-base material alone, on the honor system — you trust that the learner only ingests material they've actually finished:
- For each of this session's items now backed by an ingested page/source in
papers//concepts//sources//notes/, mark its Item ProgressScore: read(the honor marker — distinct from a testedconfident), setLast testedto today, and leaveStreakat 0. - Items with no ingested backing stay
untested. Briefly note which items are still unbacked so the learner knows what's outstanding. - Honor
readitems count asuntestedin the goal audit (read, not tested) — notcovered, which still requires a testedconfident. Mention they can switch totestmode anytime to verify and raise true coverage.
Step 8 — Recap and next steps
Print a recap:
- Covered: module + items learned/tested this session, and which sources were ingested.
- Test track — Scores (confident / partial / wrong / skipped); shaky items
(
partial/wrong, each with a "re-read §X" pointer); spaced-rep forecast (which items are due next and when). - Honor track — items now
read/ingested vs still outstanding; the read/ingested count alongside tested coverage. - Next session: the next un-ticked day from
schedule.mdand its module. - Suggest next steps: e.g. ingest a still-missing source, run
/collateto fold new reading into a page, switch test/honor mode, or (test track) re-test shaky items.
If a module crossed the mastery threshold this session, add an end-of-module milestone recap: module mastery %, deepest Bloom reached, a "you can now…" tied to its learning objectives — then unlock the next module.
Step 9 — Write progress.md and sync
Update course/<slug>/progress.md:
- Item Progress rows: new
Score(confident/partial/wrongon the test track;readon the honor track),Bloom,Last tested(today),Streak. Carry over rows not touched this session unchanged. - Module Mastery rows: recompute Mastery %, Confident, Bloom reached, Status.
- Frontmatter: recompute
coverage_pct(confident items ÷ all items — honorreaditems do not count toward it) andcourse_mastery_pct(mean of module mastery %); setlast_updatedto today.
Then sync goal.md: overwrite the Knowledge Audit Status/Score/Last tested to
match — covered for confident items, untested for honor read items, revisit
for partial/wrong — and update its coverage_pct. Tick completed days in
schedule.md and any finished modules in plan.md's weekly plan. Change nothing else.
Step 10 — Promote mastered material to the personal base
Course material lives in the per-course library course/<slug>/library/ until absorbed.
At the end of the session (after the recap), identify library sources eligible for
promotion. A course/<slug>/library/<file>.md is eligible when any holds:
- Covered — backing items all
confident(100% covered), even before streaks reach themasteredbar. Offer as soon as a module hits 100% confident; don't wait for the retest. - Mastered — its module is
mastered(per Step 7A.3). - Ingested + annotated — the learner ingested it into the personal base (the honor-track promotion path).
For eligible sources, suggest promoting them into the personal sources/, noting which
bar triggered it (e.g. "Module 1 is 100% covered — add these to your personal sources?").
On yes, cp course/<slug>/library/<file>.md sources/<file>.md (keep the library copy).
Once in sources/ it enters the retrieval index, so a future /goal audit finds it without
re-fetch and /collate//concept can synthesise it into a page. Never promote silently,
and never below 100% covered (test) / ingested (honor). If the learner already ingested the
real source in Step 6, that personal page is the durable copy — no library promotion needed.
Step 11 — Advance
Move the course forward: confirm this session's hour-block is ticked in schedule.md, and
point the learner at the next hour-block's module (or the next module if this one is
complete). Offer
to continue now (/tutor <slug> for the next hour-block, or /tutor <slug> hour <N>), or
to stop here. If the schedule is finished, suggest /tutor <slug> review to keep due items
fresh.
Wrap up
Beyond the recap, promotion offers, and the advance pointer, ask:
- (Test track) Re-test any wrong/partial items now?
- Incorporate into your personal knowledge base (or update) a source you read today that isn't in the base yet?
- Open a page that revealed gaps via
/edit <page>?
Then respect tutor_prompt_mode for the next /edit (below). Do not reindex
manually — the background poller handles it.
Tutor prompt trigger — spec for /edit integration
When /edit finishes on a page it should respect the user's cadence:
- Glob
course/*/goal.md. For each, check whether the edited page's slug/title appears in its Knowledge Audit or the course's Module Map. If no match, do nothing. For a matching course, read
tutor_prompt_mode:session-end— at the very end of the/editwrap-up (once per session, even if several pages in the course are edited), append:Course
<slug>covers this page — remember to run/tutor <slug>this session. Track with a flag filecourse/<slug>/.session_reminded(deleted on session start, created on first reminder).after-3-edits— read the integer incourse/<slug>/.edit_count(0 if absent), increment, write back. When it reaches 3, prompt:You've edited 3 pages covered by course
<slug>this session — run/tutor <slug>? Then reset.edit_countto 0.neverorunset— no prompt.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

