agentleFS
Agents4Academia-AIAgents4Academia-AI/UReKA/.claude/skills/curriculum/SKILL.md

curriculum

Skill9 starsChanged 5 months agoMITView on GitHub

Build a curriculum for a goal — pull vetted resources (papers, Wikipedia, blogs, tutorials), sequence them into modules, schedule against your hours, and compile a cited course doc per module. Second stage of a course; consumes /goal, hands off to /tutor.

  • Reads credentials

3 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 0 are distinctive, and 6 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 6

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    If $ARGUMENTS starts with edit , extract the slug and jump to Edit Mode.

    line 39
  2. Very specific:

    If course/<slug>/plan.md already exists, ask: update it (→ Edit Mode) or overwrite?

    line 57
  3. Very specific:

    Reuse, don't refetch: if the file already exists in this library, another course's course/*/library/, or the personal sources/, cite/copy …

    line 74
  4. Very specific:

    If the paper pull fails (no .env/network), say so and continue with web + base only.

    line 77
  5. Very specific:

    Sync goal.md: update the Knowledge Audit Status to pulled for items you pulled resources for this run.

    line 134
  6. Very specific:

    Add/remove a module or a resource (pull new ones via Step 2 for just that topic).

    line 159

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Build the curriculum for a course on $ARGUMENTS — like a university syllabus: pull factual, verifiable, widely-used resources (papers + Wikipedia + blogs + tutorials/docs), sequence them into modules (foundational → core → advanced), schedule them against the deadline and hours, and compile one cited course doc per module.

This is the second stage of the learning pipeline — it consumes the goal and hands off to the tutor:

/goal <topic>   →  course/<slug>/goal.md   (target + scope + knowledge audit)
/curriculum <slug>  →  plan.md + schedule.md + modules/ + initialized progress.md
/tutor <slug>       →  each session: reading list → teach → (test|honor) → advance

Output lives under course/<slug>/:

  • plan.md — module map, resources, weekly plan, feasibility
  • schedule.md — hour-by-hour breakdown driven by your weekly hours
  • modules/<n>-<slug>.md — compiled, cited learning material per module
  • progress.md — initialized so /tutor has a populated item list from the first session

Fetched material — a full concept web (built via /autoexplore) — goes into this course's per-course library course/<slug>/library/, laid out in vault-style type subfolders sources/, papers/, concepts/ — not the personal sources/, and kept out of the retrieval index (resolved by filesystem; graduates into personal sources/ only when /tutor promotes a mastered item). Module docs cite it at ../library/<kind>/<file>.md, plan.md at library/<kind>/<file>.md. Every library source records the resource's canonical URL (url:/source_links:) so /tutor can build a reading list of real paper/blog/Wikipedia links.

Orchestrates /autoexplore (resource acquisition), /web, /alphaxiv, /collate, /retrieve. Read .claude/AGENTS.md "Page conventions" before writing any module doc.

If $ARGUMENTS starts with edit , extract the slug and jump to Edit Mode.


Step 1 — Resolve the course and load the goal

Slugify $ARGUMENTS to kebab-case and look for course/<slug>/goal.md.

  • Goal exists → read it. Take the target, scope (in/out), deadline, hours_per_week, and Knowledge Audit (covered / pulled / untested / missing) straight from it — that scope is what you pull resources against. This is the normal path.
  • No goal → the goal stage hasn't run. Offer to run /goal <topic> first (recommended). If the user would rather not, scope inline: parse topic + deadline (date +%Y-%m-%d, "open-ended" if absent) + hours per week (the only time budget; suggest 5 h/week — do not assume a per-day rate), draft an in/out scope, and write a minimal course/<slug>/goal.md from .claude/Templates/goal.md before continuing.

If course/<slug>/plan.md already exists, ask: update it (→ Edit Mode) or overwrite?

Step 2 — Build the library concept web (delegated to /autoexplore)

Fill the goal's missing/pulled gaps by running the /autoexplore concept-web build against this course, so the curriculum is designed over a real interlinked web, not just flat abstracts. Invoke /autoexplore with destination course/<slug>/library/, --hops 0 (search-only — citation snowball is the slow part, and the queries already surface the canonical papers; bump to --hops 1 only if the user wants citation-depth), topic/scope = the goal's target + in/out scope, and skip the standalone synthesis page (the module docs are the synthesis). It writes the type-subfolders papers/<slug>.md, concepts/<slug>.md, and sources/alphaxiv_*/sources/web_* (Wikipedia + vetted blogs, credibility-scored). Keep volume modest — this is the most expensive step.

Constraints (enforced by /autoexplore, restate when invoking):

  • Never write the personal sources//papers//concepts/ — only the course library.
  • Reuse, don't refetch: if the file already exists in this library, another course's course/*/library/, or the personal sources/, cite/copy it instead of re-fetching.
  • Canonical URL on every kept file (url: web, source_links: papers) — /tutor's reading list.
  • If the paper pull fails (no .env/network), say so and continue with web + base only.

Track every kept source with its kind and credibility tier for the Resources list.

Step 3 — Design the module sequence

Cluster the library concept web (the concept_* pages + their paper_*/web_* sources, plus anything audited from the personal base) into modules (thematic units), ordered foundational → core → advanced. Use the concept web's cross-links to spot natural module boundaries and the prerequisite chain:

  • earlier publication year / survey (is_review) / encyclopedic overview → earlier;
  • a concept many others link to (a foundational hub) → earlier;
  • more overlap with what's already in the base → more accessible → earlier;
  • a resource/concept introducing many not-yet-covered concepts → more advanced → later.

Give each module a name, level label, assigned resources, and learning objectives. Aim for 3–6 modules unless the timeline is long. The module order is also the prerequisite chain the tutor uses for mastery gating.

Step 4 — Schedule against the timeline

  • Estimates: ingest+read a new paper ≈ 4 h; read a compiled module doc ≈ 1–2 h; Wikipedia/overview ≈ 0.5–1 h; tutorial/blog ≈ 1 h; exercises ≈ as noted.
  • Weekly plan (plan.md): group modules into weeks given hours_per_week.
  • Hour-block schedule (schedule.md): break each week into ~1-hour blocks sized by hours_per_week (e.g. 3 h/week → Hour 1, Hour 2, Hour 3 that week), anchoring each block with a ## Week <W> · Hour <H> heading and naming the module it covers — /tutor reads these to pick the next session's teaching target. Do not schedule by day or assume a per-day rate — the week's hours are the only budget, and the learner spends them whenever they like.
  • Feasibility: compare total estimated hours to available hours. If it overflows, flag clearly and ask: extend deadline, cut scope, or keep ambitious.

Step 5 — Compile module docs

For each module, write course/<slug>/modules/<n>-<slug>.md from .claude/Templates/module.md, synthesising only that module's cited sources, per AGENTS.md "Page conventions" and /collate semantics. Cite every claim inline with a relative link to its library file (../library/concepts/<slug>.md, ../library/papers/<slug>.md, ../library/sources/{web,alphaxiv}_<slug>.md) or a base page (../../../{papers,concepts}/…). Include learning objectives, key equations, further reading, and a rich ## Self-check — the seed pool /tutor draws live flashcards from, so cover every learning objective. You may delegate synthesis to /collate where a clean page maps to a module.

Step 6 — Write plan.md, schedule.md, and initialize progress.md

  • plan.md from .claude/Templates/curriculum.md: timeline + feasibility, module map table, weekly plan, resources list with tiers, notes. (Target/scope/audit stay in goal.md.) In the resources list, link each item both to its library file (library/<kind>/<file>.md, e.g. library/papers/<slug>.md) and to its canonical URL — the plan doubles as the reading list /tutor will surface.
  • schedule.md: the hour-by-hour breakdown with ## Week <W> · Hour <H> anchors.
  • progress.md from .claude/Templates/progress.md: initialize it so /tutor starts with a full item list — every module's learning items in the Item Progress table at Score: untested, Bloom: —, Streak: 0, and a Module Mastery row per module at 0% · available (locked for modules whose prerequisite isn't first). Set coverage_pct: 0, course_mastery_pct: 0.
  • Sync goal.md: update the Knowledge Audit Status to pulled for items you pulled resources for this run.

Wrap up

Per AGENTS.md "Review": report course/<slug>/ written (plan + schedule + N module docs + initialized progress); counts of resources pulled by kind/tier and any dropped (low-credibility) ones; anything left out; whether the schedule fits the hours. Then ask whether the user is happy or wants to co-edit — loop via /edit course/<slug>/modules/<n>-<slug>.md (or plan.md) until they're happy. Do not reindex manually — the background poller rebuilds the index after edits settle. Finally, point them to /tutor <slug> to start learning: each day it hands over a reading list of the real paper/blog/Wikipedia links, teaches the module, prompts you to ingest what you read into your personal base, then (if you opt in) tests you — or trusts the honor system — before advancing.


Edit Mode — /curriculum edit <slug>

Adjust an existing curriculum without rebuilding from scratch.

  1. Load course/<slug>/goal.md + plan.md (+ schedule.md); show current state.
  2. Ask what to change (several allowed):
    • Change deadline or hours (→ recompute the weekly plan + hour-block schedule).
    • Add/remove a module or a resource (pull new ones via Step 2 for just that topic). Adding/removing a module also adds/removes its rows in progress.md.
    • Mark modules/items done.
    • Add free-form notes.
  3. Apply the edits; if timing changed, redo Step 4's assignments.
  4. Write the updated plan.md/schedule.md (+ any new module docs, + synced progress.md rows) and report what changed. Do not clobber learner scores already in progress.md — only add/remove rows.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.