メニュー
Searches past sessions, memory archives, and session history for keywords, decisions, or context. Use when looking up previous session content, finding historical decisions, or querying memory by topic. Invoke with /search-memory "keyword".
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Runs Obsidian lint checks on .crabshell/ documents and reports violations. Use when you want to verify document health: missing frontmatter, broken wikilinks, INDEX.md sync issues, orphaned files. Invoke with /lint to run checks, or /lint fix to auto-apply safe fixes.
BM25 full-text search across all D/P/T/I/H/W/K documents in .crabshell/. Returns ranked results by relevance. Use when looking for specific discussions, plans, tickets, or investigations by topic keyword. Invoke with /search-docs <query>.
Creates project-specific verification tools when they don't exist, or runs existing ones against Intent Anchor items. Use when verification is needed and no project verification tool exists. Invoke with /verifying to create verification manifest, or /verifying run to execute existing tools.
Create a Crabshell K knowledge page from Codex. Use for verified facts and operational tips discovered through project work.
Create and run a Crabshell I investigation document from Codex. Use for evidence-heavy research that needs multiple source types, cross-review, and documented conclusions; not for quick questions.
Convert CLAUDE.md instructions into Codex-oriented AGENTS.md guidance. Use when the user wants Claude project rules made available to Codex.
| name | search-memory |
| description | Searches past sessions, memory archives, and session history for keywords, decisions, or context. Use when looking up previous session content, finding historical decisions, or querying memory by topic. Invoke with /search-memory "keyword". |
Use the Node.js absolute path from your context's "Node.js Path" section (injected by the plugin on every prompt).
If not available in context, fall back to node.
IMPORTANT: The scripts/ folder is in the plugin directory, NOT the current project.
From "Base directory for this skill:" above, derive the scripts path:
/skills/search-memory from the end/scripts/ to get the scripts directoryExample: if Base directory is ~/.claude/plugins/cache/.../skills/search-memory, then Scripts path is derived by removing /skills/search-memory and appending /scripts/.
Use this full path when running node commands below.
IMPORTANT: Get the project root from your context's "Project Root Anchor" section.
Look for: Your ACTUAL project root is: <path>
Use this value as {PROJECT_DIR} in all commands below.
If not available in context, use your current working directory.
Search through session history and memory archives.
/crabshell:search-memory [query]
Use full path from above:
# Search across all memory layers
"{NODE_PATH}" "{SCRIPTS_PATH}/counter.js" search-memory "query" --project-dir="{PROJECT_DIR}"
# Include L1 raw sessions (slower but thorough)
"{NODE_PATH}" "{SCRIPTS_PATH}/counter.js" search-memory "query" --deep --project-dir="{PROJECT_DIR}"
# Filter by type
"{NODE_PATH}" "{SCRIPTS_PATH}/counter.js" search-memory --type=decision --project-dir="{PROJECT_DIR}"
"{NODE_PATH}" "{SCRIPTS_PATH}/counter.js" search-memory --type=theme --project-dir="{PROJECT_DIR}"
"{NODE_PATH}" "{SCRIPTS_PATH}/counter.js" search-memory --type=issue --project-dir="{PROJECT_DIR}"
# Search all memory layers for "auth"
"{NODE_PATH}" "{SCRIPTS_PATH}/counter.js" search-memory "auth" --project-dir="{PROJECT_DIR}"
# Deep search including L1 sessions
"{NODE_PATH}" "{SCRIPTS_PATH}/counter.js" search-memory "auth" --deep --project-dir="{PROJECT_DIR}"
--deep flag for thorough L1 session search