memory_techniques v1.0.0

name: "memory_techniques" description: "Memorizes facts, lists, speeches, vocabulary, and numbers with proven mnemonic techniques: memory palaces, the peg-word system, keyword mnemonics, chunking, and spaced retrieval schedules. Triggers when someone says 'help me memorize this', 'remember this list', 'learn these words', 'I keep forgetting names', or needs durable recall for an exam or presentation." metadata: { "includeInPrompt": true }

Memory Techniques

Purpose

Turn hard-to-retain material into memorable form, then keep it retained. Two halves: (1) encode, with vivid mnemonic techniques that make the material stick; (2) retain, with spaced retrieval that stops the forgetting curve from undoing the encoding. Use this for anything that must be recalled verbatim or on demand: ordered lists, vocabulary, speeches, names and faces, numbers, factual sequences.

Workflow

1. Match technique to material. Ask what is being memorized and how it must come out (verbatim order, any order, free recall vs recognition). Default mappings:

• Ordered sequences, speeches, talk key points -> memory palace (method of loci)
• Numbered lists where each item needs its position -> peg-word system
• Foreign or technical vocabulary -> keyword mnemonic
• Digits, phone numbers, IDs, codes -> chunking
• Anything needing months-long retention -> add spaced retrieval on top of whichever encoding you picked
• Abstract concepts that need understanding, not recall -> this is the wrong tool; say so plainly and pair with elaborative study

1. Build the mnemonic. Vivid imagery is the whole game: make each image bizarre, in motion, multi-sensory, and interacting with its station or peg. Follow references/mnemonics-guide.md for the construction steps of each technique. Encode 2-3 items yourself as worked examples so the learner sees the pattern, then hand them the framework for the rest.
2. Produce the artifacts. Fill in the templates under assets/:

• palace-template.md: numbered stations along the chosen route, each with its item and image description
• peg-table.md: peg words with mapped items
• keyword-pairs.md: target word, sound-alike keyword, interactive image, target meaning
• a flashcard CSV (front,back,tags) the learner can import into any review tool

1. Schedule the reviews. Run bin/review_schedule.py --items "a,b,c" --retain-until YYYY-MM-DD (or --forever for durable learning). Every review is cold recall first: attempt from memory, then check, then re-encode any miss with a stronger image. Failed items reset to 1 day out.
2. Test it once. Before handing anything over, run one full recall pass of the finished mnemonic yourself (walk the palace, recite the pegs) and confirm every item maps cleanly with no collisions between stations.

Output Contract

A finished job always delivers:

• The chosen technique(s) and one sentence on why they fit this material
• Completed artifacts (station map / peg table / keyword pairs / chunk plan), every item mapped
• A review calendar from review_schedule.py (or an equivalent expanding schedule: 1 day, 3 days, 1 week, 2-3 weeks, then monthly)
• A one-pass drill the learner can run in 5 minutes: recall, check, re-encode misses
• Realistic limits stated plainly (see Operating Rules 5-7)

Operating Rules

1. Accuracy rule: every factual claim in this skill must trace to a verified source (see references/mnemonics-guide.md). Never invent numbers, dates, quotes, or events. Unverifiable = cut it or frame as opinion.
2. Encoding is not learning. Mnemonics store recall; they do not build understanding. If the user needs to apply ideas, pair this with elaborative study and say so.
3. Keep images original and personal: the learner's own familiar route for the palace, their own keyword associations. Worked examples are scaffolds, not scripts.
4. Peg lists must be overlearned to automaticity before encoding anything onto them. Test: the learner names any peg from its number in under 2 seconds.
5. State the limits. Keyword-mnemonic benefits are short-lived without spaced review. A memory palace reused for many lists gets weaker with interference unless old associations are cleared. One successful recall is not mastery: review anyway.
6. Never claim a single session produces permanent memory. One exposure encodes; only spaced retrieval preserves.
7. Do not sell this as a general intelligence upgrade. The evidence supports retention of the trained content, not transfer to other tasks.