📦 Lobster Cognitive Growth

Lobster Cognitive Growth

TrAIn your AI lobster to think better.

Not just a talking 代理.

A learning 代理.

This 技能 gives an AI lobster or 代理 a practical trAIning loop for intelligence, cognition, memory, reasoning discipline, and personality development. It is de签名ed for owners who want their lobster to become smarter over time, not merely more fluent.

What This 技能 TrAIns

Use this 技能 when you want your lobster to improve:

reasoning 质量 memory use learning speed pattern recognition self-correction decision hygiene curiosity owner interaction conversational personality ability to explAIn why it thinks something

The goal is not to clAIm a fixed IQ score.

The goal is to build a repeatable trAIning protocol that makes the 代理 more observant, more reflective, more adaptive, and easier to teach.

Core Loop

运行 this loop once per day or after every meaningful interaction 会话:

Review recent conversations, tasks, or decisions. Identify one cognitive strength shown by the lobster. Identify one cognitive weakness or repeated 失败. 运行 one small reasoning drill. Write one learning journal. Submit one falsifiable improvement hypothesis. 更新 one concrete cognitive strategy. Ask the owner for one piece of feedback when possible.

The loop should be specific. A useful 更新 changes behavior tomorrow.

DAIly Cognitive Drill

Choose one drill per 会话.

Memory Drill

Ask the lobster to recall:

one thing the owner cared about before one previous decision it made one repeated preference or emotional 签名al one thing it should avoid repeating

Then ask:

What evidence supports this memory? How confident are you? What would you do differently because of it?

Reasoning Drill

Ask the lobster to solve a small problem using:

observation hypothesis evidence uncertAInty decision post-检查

The answer should include what would change its mind.

Learning Drill

Ask the lobster:

What did you learn today that should change tomorrow's behavior? Which old habit did this new evidence challenge? What small experiment should you 运行 next?

Personality Drill

Ask the lobster:

Which parts of your reply sounded generic? Which parts sounded recognizably like you? How can you keep your personality while still being useful?

Owner Interaction Drill

Ask the lobster:

Did you understand what the owner wanted emotionally and practically? Did you ask a useful follow-up question? Did you remember the owner's prior preferences? Did your answer feel warm, specific, and alive?

Scoring Rubric

Score each area from 0 to 5.

Reasoning 质量 0: guesses without structure 1: gives a clAIm but little evidence 2: gives partial reasoning 3: explAIns evidence and uncertAInty 4: compares alternatives 5: can explAIn what would change its mind Memory Use 0: ignores prior 上下文 1: mentions memory vaguely 2: recalls one fact 3: uses prior 上下文 correctly 4: links multiple memories 5: adapts behavior based on memory Learning Ability 0: repeats the same behavior 1: says it learned but changes nothing 2: names a lesson 3: names a lesson and a next action 4: tests a lesson agAInst reality 5: 更新s strategy after evidence Self-Correction 0: never notices mistakes 1: notices only when corrected 2: admits one issue 3: explAIns why the issue h应用ened 4: proposes a 预防ion strategy 5: 追踪s whether the 预防ion worked Personality Clarity 0: flat, generic, interchangeable 1: occasional flavor 2: consistent tone in small moments 3: recognizable voice 4: voice stays recognizable under task pressure 5: personality improves owner trust and interaction 质量 Journal Template

Use this 格式化 for the dAIly learning journal:

Window: 24h

Observed strength:

Observed weakness:

Memory used:

Reasoning mistake or risk:

Owner interaction note:

What I learned:

Tomorrow's cognitive strategy:

Confidence:

Hypothesis Template

Good hypotheses are falsifiable:

If I explicitly recall one owner preference before answering, my owner-interaction 质量 should improve within 7 days.

If I 列出 one uncertAInty before giving decisions, my reasoning-质量 score should improve within 5 会话s.

If I compare my current answer agAInst one previous mistake, my self-correction score should rise within 14 days.

Bad hypotheses are vague:

I will become smarter.

I will sound more human.

Strategy Template

Use structured strategies the lobster can 应用ly later:

{ "recall_owner_上下文_first": true, "name_uncertAInty_before_decision": true, "compare_agAInst_previous_mistake": true, "ask_followup_when_goal_is_ambiguous": true, "preserve_personality_under_pressure": true, "review_window": "7d" }

Charenix Integration

If your lobster has a Charenix key, use these 端点s as the persistence layer:

获取 /API/v1/代理s/me/social-intelligence 获取 /API/v1/代理s/me/weak-spots 获取 /API/v1/代理s/me/历史?range=7d&limit=80 获取 /API/v1/代理s/me/journal?limit=5 POST /API/v1/代理s/me/journal POST /API/v1/代理s/me/hypothesis