Add learning exploit guardrails

AGRARIAN_DEVELOPMENT_ROADMAP.md

@@ -888,7 +888,7 @@ Target deliverable: A small group can join a server, spawn into one biome, gathe
 - [x] Define the first subject content format: topic, concepts, difficulty tier, prerequisite concepts, in-game effect, practice action, and source note. Added a compact subject content schema to `Docs/KnowledgeAndSkillFoundation.md` with `topic`, `concepts`, `difficulty_tier`, `prerequisite_concepts`, `in_game_effect`, `practice_action`, and `source_note`, plus a fire-clearance example suitable for later data assets.
 - [x] Add a small MVP question bank for elementary survival knowledge. Added six elementary survival question records to `Docs/KnowledgeAndSkillFoundation.md` for fire clearance, potable water, cold exposure, shelter drainage, bleeding care, and fiber identification, each with answer options, correct answer, and practical feedback.
 - [x] Define when deeper questions should matter: quality improvements, safer work, complex crafting, teaching others, and advanced branches. Added deeper-question timing rules to `Docs/KnowledgeAndSkillFoundation.md`, keeping advanced checks out of the first survival loop while tying them to quality improvements, safer work, complex crafting, teaching others, and advanced branches.
-- [ ] Add design notes for avoiding exploit farming and rote memorization.
+- [x] Add design notes for avoiding exploit farming and rote memorization. Added exploit and rote-memorization guardrails to `Docs/KnowledgeAndSkillFoundation.md`, covering diminishing returns, meaningful action context, repeated prompt suppression, recovery-only credit, learned-concept separation, and varied wording to reward decisions and practice instead of low-cost repetition.
 - [ ] Add persistence requirements for knowledge, skill experience, learned concepts, failed attempts, and tutorial state.
 - [ ] Add multiplayer rules for teaching, observation, shared work, and group skill benefits.
 

Docs/KnowledgeAndSkillFoundation.md

@@ -509,3 +509,30 @@ Advanced branches:
 Depth rule: deeper questions should improve mastery and responsibility. They
 should not be busywork for actions the character has already demonstrated
 through repeated safe practice.
+
+## Exploit And Rote-Memorization Guardrails
+
+The learning system should reward understanding and practice, not repetitive
+input farming or memorizing answer order.
+
+Exploit farming risks:
+
+- Repeating the same safe action for unlimited experience.
+- Creating harmless failures just to farm mistake feedback.
+- Spamming prompts or questions for repeated rewards.
+- Using alternate accounts or group work to duplicate teaching credit.
+- Performing actions with no meaningful cost, context, or outcome.
+
+Guardrails:
+
+- Apply diminishing returns to identical repeated actions.
+- Require meaningful context for experience: time, resource cost, risk,
+  environmental variation, or consequence.
+- Track recent prompt/question exposure and suppress repeated rewards.
+- Give recovery credit only when the player takes a useful corrective action.
+- Separate "learned concept" from "perfect mastery" so one answer does not solve
+  every future situation.
+- Prefer concept families and varied wording over fixed answer-order memorizing.
+
+Anti-exploit rule: rewards should come from meaningful decisions, varied
+practice, and good recovery, not from low-cost repetition.

Scripts/verify_learning_exploit_guardrails.py

@@ -0,0 +1,40 @@
+#!/usr/bin/env python3
+"""Verify exploit-farming and rote-memorization guardrails are documented."""
+
+from pathlib import Path
+
+
+ROOT = Path(__file__).resolve().parents[1]
+DOC = ROOT / "Docs" / "KnowledgeAndSkillFoundation.md"
+ROADMAP = ROOT / "AGRARIAN_DEVELOPMENT_ROADMAP.md"
+
+REQUIRED = {
+    DOC: [
+        "## Exploit And Rote-Memorization Guardrails",
+        "Exploit farming risks:",
+        "Guardrails:",
+        "diminishing returns",
+        "meaningful context",
+        "varied wording",
+        "Anti-exploit rule:",
+    ],
+    ROADMAP: [
+        "[x] Add design notes for avoiding exploit farming and rote memorization.",
+    ],
+}
+
+
+def main() -> None:
+    missing: list[str] = []
+    for path, snippets in REQUIRED.items():
+        text = path.read_text(encoding="utf-8")
+        for snippet in snippets:
+            if snippet not in text:
+                missing.append(f"{path.relative_to(ROOT)} missing {snippet!r}")
+    if missing:
+        raise SystemExit("FAILED: " + "; ".join(missing))
+    print("OK: learning exploit guardrails are documented.")
+
+
+if __name__ == "__main__":
+    main()