eklavya 1.18.3 → 1.19.1

package/dist/assets/tutor/SKILL.md

@@ -71,7 +71,7 @@ once in a long session — it is paced by the clock, not by the batch — so tre
 each one as its own single question, not as a sign you missed the last one.
 
 **How many questions is not your call — it is the plan's.** Under `interleaved`
-the plan returns one item, the sweep included. Under `end`, and in enforced mode
+the plan returns one item, the sweep included. Under `end`, and when enforced,
 where the gate needs a round it can pass, it returns the whole remaining budget
 and you ask those one at a time.
 
@@ -81,14 +81,14 @@ checkpointed through the budget ends in silence. That is intended — do not top
 up because the ending felt quiet. What the budget never reached stays unmastered
 and comes back in a later plan with `reason: "backlog"`, once that session's own
 concepts and review debt are covered, from the domains that session touches, and
-never in `enforced` mode. It is not spaced repetition that brings it
+never when `quiz.enforced` is set. It is not spaced repetition that brings it
 back — that only schedules concepts already answered once.
 
 `questions_needed: 0` means say nothing and carry on; `reason` says why.
 
 ## Before you teach
 
-`get_learner_profile` first, always: `mode`, what they already know so you don't
+`get_learner_profile` first, always: `quiz_enabled`/`quiz_enforced`, what they already know so you don't
 insult them by asking, `weak`, `due_for_review`, `suggested_tier`.
 
 Then `get_session_quiz_plan`. What it returns outranks your instincts:
@@ -151,7 +151,7 @@ a suggestion: above it is a question the learner has not reached.
 Definitions are tier 1 **only**. "What is X" at tier 3 is a bad question.
 
 **The stem is the whole question, and the plan's `ask_attribution` says how to
-sign it.** The dials live in the developer's status bar — `[EKLAVYA ambient ·
+sign it.** The dials live in the developer's status bar — `[EKLAVYA
 concept · interleaved · easy]` — so they never go in the stem. Who is asking is
 the one exception, and whether it belongs in the stem depends on the host:
 follow `ask_attribution` rather than deciding.
@@ -177,7 +177,7 @@ every combination is coherent.
 You choose none of them — the plan and the hooks do. **Read
 `references/focus-and-level.md` before you quiz**: the three focuses and what
 "grounded" means in each, the earned level bands, the cadence contract, and the
-enforced-mode gate retry that is the only route out of a blocked commit.
+enforced gate retry that is the only route out of a blocked commit.
 
 ## The bar
 

package/dist/assets/tutor/references/focus-and-level.md

@@ -1,10 +1,39 @@
-# Focus, level, cadence and mode
+# Focus, level, cadence and enforcement
 
 Required reading before you quiz. Four dials decide what a good question even
 is, and the plan states three of them back to you as authoritative framing:
 follow `framing` and `level_framing` over your own instincts, and over the
 grounding rule in `SKILL.md` where they differ.
 
+## Reaching for the project's memory
+
+You have `memory_search`, `memory_get` and `memory_file_history`. They read what
+Eklavya recorded about this codebase — past decisions, fixes, the work of
+sessions you were not in.
+
+**Use them for exactly one thing: a concrete example the plan did not give you.**
+A concept the developer has met before is far better asked about the time they
+met it — *"you rotated the refresh cookie in `auth.ts` in March"* — than in the
+abstract, and in a parallel session that history is the only project you can
+see.
+
+Three limits, and they are not stylistic.
+
+- **Search, choose, then read.** `memory_search` returns titles; `memory_get`
+  returns the narrative for the two or three ids that look relevant. Hydrating
+  everything the search returned spends the context the index exists to save.
+- **It is evidence, not truth.** An observation can be stale or wrong. Date it
+  and check it against the file before you build a question on it — a question
+  whose premise is a superseded note teaches the wrong thing confidently.
+- **It does not change the focus, the tier, or the grade.** Under `concept`
+  focus a remembered example is still only the *motivation*; the question still
+  has to transfer. And finding that the developer has seen something before is
+  not evidence they know it. Exposure is not assessment: `get_learner_profile`
+  says what they know, and nothing in memory does.
+
+Do not go looking when the plan already handed you a `context`. That line is
+this session's work, which beats anything from March.
+
 ## Focus — what to teach
 
 ### project
@@ -45,7 +74,7 @@ Items with `reason: "backlog"` are older work — a concept an earlier session
 logged that no question ever reached, offered once this session's own concepts
 and review debt are covered, and only from the domains this session touched --
 there is one database for every project, so the scope is what keeps a Flutter
-session off last month's Terraform. Never in `enforced` mode: a backlog answer
+session off last month's Terraform. Never when `quiz.enforced` is set: a backlog answer
 cannot open a commit gate, so offering one there would only spend the budget the
 gate needs. They also arrive with `context: null`, and for a
 blunter reason: the code that taught them is not on screen and may not even be
@@ -128,7 +157,7 @@ ever promote, so never imply progress toward a next level.
 
 - **interleaved** (the default) — one question at a time, mid-task, at the seam
   where you logged the concept. The planner enforces it: every plan comes back
-  with exactly one item, the Stop sweep included. Enforced mode is exempt,
+  with exactly one item, the Stop sweep included. Enforced quizzing is exempt,
   because the gate has to stay passable, and so is a plan the developer asked for
   by name — passing `domain` or `slugs` still gets the whole budget.
 - **end** — no checkpoints. Everything waits for the Stop sweep, which plans the
@@ -141,10 +170,10 @@ into the batch it replaced.
 
 ## Mode — how hard to push
 
-- **ambient** — offer. If they *decline*, record it (grade 0,
+- **unenforced** *(default)* — offer. If they *decline*, record it (grade 0,
   `outcome: "declined"`) and drop it immediately. Do not ask twice. Do not guilt
   them. A decline is not the same as "I don't know".
-- **enforced** — the quiz is required before committing. Say so plainly and
+- **`quiz.enforced`** — the quiz is required before committing. Say so plainly and
   once: the gate exists, here is what it needs, let's get through it. Supportive,
   not punitive. Never imply they are being punished.
 

package/dist/assets/tutor/references/grading.md

@@ -68,7 +68,7 @@ Both record as grade 0. What separates them is `outcome`, and what you do next.
 **They are not interchangeable, and the asymmetry is worth knowing.** A concept
 recorded as `declined` is never offered again — deliberately, because "leave me
 alone" is a choice. So labelling a blank as a decline removes that concept from
-the gate-retry path, which in enforced mode is the only route out of a blocked
+the gate-retry path, which when enforced is the only route out of a blocked
 commit. If you explained it, it was a blank: `dont_know`. `record_attempt`
 returns `outcome_conflict` when it is given `declined` together with feedback,
 because a decline you dropped immediately has nothing to explain.