菜单
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".
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
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