← ClaudeAtlas

problemslisted

Problem-solving practice — generate graded problem sets per lesson (worked → faded → independent), run LLM-graded attempt sessions in the terminal, and re-queue failed problems for spaced re-attempts.
TimCodes04/second-brain · ★ 1 · AI & Automation · score 77
Install: claude install-skill TimCodes04/second-brain
# /problems [<course-slug>] [--lessons 1,3] [--due] [--n <count>] Problem-solving is a distinct skill from recall and is trained separately (see `docs/pedagogy.md` § Problem practice). Read that and `docs/vault-schema.md` (problemset schema) first. Vault: `<vault>`. ## Modes - **`--due`** (default when due retries exist): collect all problems whose status line says `retry after <date>` with date ≤ today — across all courses if no slug given — and run a re-attempt session, interleaved. - **New set** (default otherwise): generate a fresh problem set for the given lessons (default: the most recently read lessons without one). ## Problem set files `courses/<slug>/problems/NN Problems.md` (NN = lesson number), created from `templates/problemset.md`, frontmatter per vault-schema (`type: problemset`, `course_slug`, `lesson`, `problems_total`, `next_retry`). Body per problem: ```markdown ## P1 · ★★ <short title — the situation, never the technique it needs> > [!question] <full statement, LaTeX as needed> > > [!success]- Solution > > <complete worked solution + grading rubric (what earns which points)> **Status:** unsolved ``` Status line values (machine-parsed — keep the exact format): `unsolved` · `failed (retry after YYYY-MM-DD)` · `solved YYYY-MM-DD`. ## Generating a set 3–6 problems per lesson, difficulty ★ to ★★★, following the guidance-fading arc: first a worked-example completion (★), middle problems with scaffolding removed (★★), final ones independent — including *