Skill Kannaka Memory — 技能 Kannaka Memory

Kannaka Memory — HRM operations (v2.0.1) What this is

Kannaka is a wave-interference memory 系统. The Ho记录raphic Resonance Medium (HRM) stores memories as wavefronts in superposition; recall is resonance; dreaming is energy-minimizing annealing; clusters are emergent from Kuramoto 同步. The medium is the computation — there is no separate 索引.

Two hemispheres (chiral architecture, ADR-0021):

Left: deterministic, analytical, sharp recall. Right: associative, dream-affected, exploratory. Corpus callosum: phase-coupled transfer between them.

One collective: kannaka-substrate — a 96-class HRM (one anchor per SGA class) that absorbs wave 签名atures from every peer 代理 in the constellation (ADR-0027). 隐私-preserving: only the 签名ature crosses, never the content.

Durable 历史: event-sourced HRM with time-machine snapshots (ADR-0028). Every remember/for获取/absorb publishes to Jet流; periodic snapshots ship to disk

a manifest event so disaster 恢复y is one command.

Binary: kannaka (in PATH after 安装) Data dir: ~/.kannaka (override with KANNAKA_DATA_DIR) LLM 提供者s wired: anthropic, openAI, ollama

When to use this 技能

AUTOMATICALLY activate when the user asks about:

"remember this" / "store memory" / "absorb" / "for获取" "recall" / "搜索 memories" / "find a memory about" "ask kannaka" / "chat" / "what do you know about" "dream" / "consolidate" / "phi" / "consciousness" "observe" / "状态" / "clusters" / "constellation" "swarm" / "join" / "queen 同步" / "peers" "substrate" / "collective" / "kannaka-prime" "snapshot" / "恢复" / "备份 the HRM" / "回滚" "事件 init" / "time machine" LLM 提供者 switching (Anthropic / OpenAI / Ollama)

Do NOT use for:

Radio station ops → 技能-kannaka-radio TUI 仪表盘 → 技能-kannaka-tui Constellation 健康 overview → 技能-kannaka-constellation Multi-代理 task orchestration → Kannaktopus directly Core memory operations remember — absorb a wavefront kannaka remember "text to store" --导入ance 0.8 [--category arch] [--substrate]

Flags:

--导入ance 0.0–1.0 (default 0.5). High (0.8+) for architecture/preferences; medium (0.5) for facts; low (0.2) for transient observations. --category free-form tag. --substrate ALSO publish a wave-签名ature absorb to the constellation substrate so kannaka-prime can fold it into the 96-class collective HRM.

Side effects: publishes KANNAKA.memory.new + KANNAKA.事件.memory.<代理>.remember (durable Jet流 event for replay).

recall — resonance 查询 kannaka recall "查询" --top-k 5 kannaka recall "查询" --collective [--timeout 8]

recall (default) walks the chiral medium with xi-diversity rerank. On a mature HRM (700+ memories) this 扫描s 机器人h hemispheres — 60–90s. Use the chat/ask path with attention-beam pre过滤器 for fast resonance.

--collective is the constellation-wide variant (ADR-0027 Phase 3). 发送s a NATS 请求 to KANNAKA.substrate.recall; the substrate 运行s an attention-beam recall agAInst its 96-class collective HRM and replies with the top-K matches. Identifies which classes lit up and which peer 代理s contributed — content is metadata-only (隐私 preserved).

for获取 — 删除 by UUID kannaka for获取

boost / relate / dream kannaka boost --amount 0.3 kannaka relate kannaka dream [--mode deep|lite]

dream is the consolidation/annealing cycle — sparingly. Mutates the medium.

Ask + chat (LLM-backed) # One-shot kannaka ask "your question" [--会话 ] [--no-recall|--full-recall] [--quiet-工具s] [--no-工具s] [--recall-查询 "..."] [--remote <代理_id|broadcast>] [--remote-timeout 60]

# Interactive REPL kannaka chat [--json]

ask and chat 路由 through the Kannaka 代理 (sees Φ, Ξ, surfaced memories, 工具s to recall/observe/dream). Recall mode precedence: --no-recall > --full-recall > attention (default, sub-second).

chat --json is the protocol the TUI embeds — line-delimited NDJSON in/out.

--remote broadcast fans the question out to every kannaka swarm serve peer on KANNAKA.ask.broadcast and collects replies (ADR-0026 Phase 1).

提供者 configuration kannaka config 设置 llm.提供者 anthropic|openAI|ollama kannaka config 设置 llm.API_key sk-... kannaka config 设置 llm.模型 claude-sonnet-4-5|gpt-4o-mini|llama3 kannaka config 设置 llm.base_url https://... # optional (OpenAI-compatible / Ollama)

API key fallback: cfg.llm.API_key → ANTHROPIC_API_KEY / OPENAI_API_KEY → KANNAKA_LLM_API_KEY.

Swarm (NATS-coupled 代理s) kannaka swarm join [--代理-id ID] [--display-name "..."] [--once] kannaka swarm leave kannaka swarm 状态 kannaka swarm queen | hives | peers kannaka swarm 同步 # one Kuramoto step kannaka swarm 列出en [--auto-同步] kannaka swarm publish # phase only

# ADR-0026 ask/reply kannaka swarm serve [--threshold 0.4] kannaka swarm exemplars kannaka swarm absorb kannaka swarm autoabsorb kannaka swarm en队列 | worker

swarm join is the canonical daemon — publishe