菜单
Cross-repository cognitive memory system with semantic search. Persists knowledge across sessions using TF-IDF recall, memory versioning, knowledge graph edges, and confidence decay. Memory is cognition, not storage.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
| name | cognitive-memory |
| description | Cross-repository cognitive memory system with semantic search. Persists knowledge across sessions using TF-IDF recall, memory versioning, knowledge graph edges, and confidence decay. Memory is cognition, not storage. |
| trigger | [{"keyword":["memory","remember","recall","forget","what do we know","last session","previous session","what was","do you remember"]},{"intent":["recalling past context","storing new knowledge","resolving contradictions","session continuity"]},{"event":"session_start"}] |
| priority | high |
| version | 3.0.0 |
Memory as stewardship: what we remember shapes how we serve.
update creates a new version, preserves the original with supersedes chain.link creates bidirectional edges between related memories.consolidate flags memories with >80% similarity.This skill provides persistent cognitive memory across Claude Code sessions. It is NOT a database — it is a reasoning layer that encodes selectively, consolidates contradictions, recalls semantically, and forgets intentionally.
Memory store: ~/.memory/DOMAIN/
Operations script: /home/user/ken/orchestrator/memory_ops.py
At the beginning of every session, recall relevant memories:
python3 /home/user/ken/orchestrator/memory_ops.py recall "" --domain photography --limit 10
python3 /home/user/ken/orchestrator/memory_ops.py tree --domain photography
Present a brief summary to the user:
python3 /home/user/ken/orchestrator/memory_ops.py encode photography <type> "content" \
--tags tag1,tag2 --related id1,id2
Types: insight, decision, pattern, fact, preference
Importance → confidence mapping:
python3 /home/user/ken/orchestrator/memory_ops.py recall "natural language query" --domain photography --limit 10
Recall now uses TF-IDF semantic matching. You don't need exact keywords — conceptually related memories surface automatically. Each result includes a _score field.
Trust but verify: If a recalled memory has low confidence or a low score, say so. Don't present uncertain memories as facts.
python3 /home/user/ken/orchestrator/memory_ops.py update <id> "corrected content" --domain photography
Creates a new version. The old memory is preserved with reduced confidence and a superseded_by pointer. Use this when facts change — don't forget and re-encode, update.
python3 /home/user/ken/orchestrator/memory_ops.py link <id_a> <id_b>
Creates a bidirectional edge. Use when you discover two memories are related — a breeding decision connects to a flock validation insight, a recipe correction connects to a transcription note.
python3 /home/user/ken/orchestrator/memory_ops.py consolidate --domain photography
Decays unrecalled memories, removes dead ones, and reports potential duplicates (>80% similarity). Run periodically or at session end.
python3 /home/user/ken/orchestrator/memory_ops.py tree --domain photography
Shows memory count, types, edge connections, and version chains per domain.
python3 /home/user/ken/orchestrator/memory_ops.py forget <id> --domain photography
Careful, not clever. What we remember matters. What we forget matters too.
# After adding a new product
python3 /home/user/ken/orchestrator/memory_ops.py encode photography decision \
"Mountain Majesty added at $49-$299 range. Metal prints for this image because the blacks need to go absolute. Canvas softens it too much." \
--tags mountain-majesty,product,pricing,metal
# After a voice lesson
python3 /home/user/ken/orchestrator/memory_ops.py encode photography pattern \
"Product descriptions work best at 2-3 sentences. One about the moment, one about the medium, one about where it belongs. Anything more reads like catalog copy." \
--tags voice,product-description,pattern
Foundational knowledge that should NEVER decay, regardless of recall frequency.
# Encode as protected from the start
python3 /home/user/ken/orchestrator/memory_ops.py encode photography fact "content" --protected
# Protect an existing memory
python3 /home/user/ken/orchestrator/memory_ops.py protect <id> --domain photography
When to protect:
Auto-protection: Memories with 3+ graph edges are automatically protected during consolidation — if many things reference it, it is foundational by definition.
Recall now searches ALL domains by default. Each result includes _domain so you know where it came from. This enables cross-pollination — a breeding pattern in sheep might inform resource organization in recipes.
# Search everywhere (default)
python3 /home/user/ken/orchestrator/memory_ops.py recall "optimization strategy"
# Restrict to one domain when you know what you want
python3 /home/user/ken/orchestrator/memory_ops.py recall "optimization strategy" --domain photography
Well-connected memories score higher in recall. A memory linked to 5 other memories ranks above an isolated memory with the same text similarity. This rewards knowledge that has been woven into the graph.
Score formula: similarity * confidence * (0.70 + 0.15*recency + 0.15*centrality)
Explore the knowledge graph from any memory:
# Direct neighbors
python3 /home/user/ken/orchestrator/memory_ops.py neighbors <id> --depth 1
# Neighbors of neighbors
python3 /home/user/ken/orchestrator/memory_ops.py neighbors <id> --depth 2
Old, low-confidence, unprotected memories are automatically archived during consolidation (>180 days, <0.3 confidence). Archived memories:
~/.memory/_archive/--include-archive to search them)python3 memory_ops.py promote <id># Manually archive
python3 /home/user/ken/orchestrator/memory_ops.py archive <id> --domain photography
# Search including archive
python3 /home/user/ken/orchestrator/memory_ops.py recall "old topic" --include-archive
# Bring something back
python3 /home/user/ken/orchestrator/memory_ops.py promote <id>
Consolidate is now smarter:
Voice & Presence standard for fine art photography content. Guards the sound of product descriptions, gallery text, artist statements, and collection narratives so they read as written by a photographer, not generated by a machine. For post-draft diagnostics, see voice-audit.
Post-draft diagnostic for photography content. Scans for machine tells, assesses authenticity risk, checks voice continuity against the photographer's voice standard. Fires before committing content changes. For during-writing standards, see like-a-human.
"Enforces atomic commits, checkpoint summaries, rate-limit recovery, and safe resume patterns. Writes session pulses + cognitive memory so interrupted sessions can restart cleanly."
Guides GA4 setup and conversion tracking for flickersofmajesty.com. Focuses on e-commerce metrics: product views, purchase intent, and print size preferences.
Deep audience personas for fine art photography buyers. Feeds product descriptions, gallery curation, and emotional-hook-test calibration.
Final typographic polish before publishing product and gallery pages. Checks curly quotes, em-dashes, alt text, price formatting, and visual consistency.