From agentic-study-environment
Adjust the difficulty level of a learning sub-project's curriculum — simplify a too-advanced curriculum into a gentler on-ramp, or deepen a too-introductory one into something with real teeth. Use whenever the user signals the curriculum's level doesn't match their starting point — phrases like "this paper is too hard, can we build up to it?", "simplify the curriculum", "make it easier", "go more advanced", "I want a more rigorous version", "this is too basic, push deeper", "I need an intro version of this", or "create a simpler/harder curriculum from this material". The skill is allowed to pull in external materials and training knowledge to fill the gap between the source's natural level and the target level, but every external addition is explicitly labeled — the user always knows what came from their own source materials and what came from elsewhere.
How this skill is triggered — by the user, by Claude, or both
Slash command
/agentic-study-environment:adjust-levelThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
This skill rewrites a sub-project's `curriculum.md` at a different difficulty level than the source materials naturally support — either simpler (build up to a hard source) or more advanced (extend beyond an introductory source). It is allowed to draw on external materials and training knowledge to fill the gap, but **every externally-sourced piece of the curriculum must be explicitly labeled**...
This skill rewrites a sub-project's curriculum.md at a different difficulty level than the source materials naturally support — either simpler (build up to a hard source) or more advanced (extend beyond an introductory source). It is allowed to draw on external materials and training knowledge to fill the gap, but every externally-sourced piece of the curriculum must be explicitly labeled so the user always sees the provenance of what they're being asked to study.
The user wants the curriculum at a different level than the source materials directly support. Two common cases:
If the user just wants the curriculum to better match their source materials (without shifting level), use set-curriculum, not this skill.
This skill is not a session — it does not update PROGRESS.md.
Required:
Optional:
<name>/CLAUDE.md — for Domain:, Language:, learning goals, Tools & Materials.<name>/ai-agent-materials/curriculum.md — the baseline being adjusted. If it doesn't exist, redirect the user to set-curriculum first — there is nothing to adjust yet.<name>/ai-agent-materials/ (concept maps, copied excerpts) — these remain ground-truth for what the user's source actually covers.<name>/source-materials/ — to keep firm what is in-source vs. what will be brought in.../../reference/conventions.md — for the curriculum format and the External-source labeling rules. The labeling rules are not optional; read them.The user's source material is the summit; you're designing the ascent.
The user's source material is the floor; you're designing the climb above it.
T3: mean-field closure and the adjusted curriculum still has the same topic (possibly with different prereqs or framing), keep it as T3. This protects existing PROGRESS.md references.T2.5 between T2 and T3, T8 after the previous last topic. Do not renumber existing topics.[deprecated] in the topics table rather than deleting them, if PROGRESS.md references them. Otherwise delete cleanly.Every piece of the adjusted curriculum that did not come from the user's source materials must be labeled. The format is documented in detail in ../../reference/conventions.md (External-source labeling); the summary:
paper.pdf §3.2, lecture-notes.md ch. 4.[ext] and name the source concretely:
[ext] Krapivsky, Redner & Ben-Naim, *A Kinetic View of Statistical Physics*, ch. 7 (a specific book)[ext] Wikipedia: Markov chain (https://en.wikipedia.org/wiki/Markov_chain) (a web page)[ext] training knowledge — general background in measure theory (training knowledge, no specific cited source)Training knowledge counts as external. Do not let trained material drift in unlabeled. If you're using your own pattern-recognition to add a topic the user's source doesn't cover, that topic gets a [ext] training knowledge — <area> label, no exception.
Never invent citations. If you genuinely don't remember the source of a piece of background knowledge, label it as training knowledge with a description of the domain area, rather than fabricating an author, title, or page number. A fabricated citation is worse than an honest "training knowledge — combinatorics background".
When in doubt about a topic's provenance, default to labeling it external. False positives (over-labeling) are recoverable; under-labeling silently misleads the user about what's in their material.
<name>/ai-agent-materials/curriculum-prev.md, overwriting any previous backup. The user can revert with one cp if the adjustment doesn't land.<name>/ai-agent-materials/curriculum.md, replacing the file. Follow the standard curriculum format (see ../../reference/conventions.md).set-curriculum produces.PROGRESS.md or any sub-project file other than the curriculum and its backup. Adjusting the curriculum is not a session.Tell the user concisely:
<name>/ai-agent-materials/curriculum-prev.md) and how to revert (cp curriculum-prev.md curriculum.md).PROGRESS.md from becoming a museum of dead references. If they exercised T3 last week, T3 should still mean the same concept after the adjustment.set-curriculum — build the original curriculum from source materials, source-faithful by default.start-session — when conducting a session over an adjusted curriculum, the agent reads the labels and surfaces external-source citations during teaching so the user always knows where the current explanation comes from.bootstrap, stop-session — the rest of the lifecycle; this skill doesn't touch them.Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.
npx claudepluginhub timbogp/agentic-study-environment --plugin agentic-study-environment