القائمة
LogosDB — incremental index of a file or directory (changed/new files only; fast refresh, concise output).
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
LogosDB — store a memory, fact, or note in the global cross-session namespace (or any namespace). Concise output.
LogosDB semantic memory plugin: global cross-session memory, thinking-trace storage, incremental /index, /remember, /search, /forget. Triggers: semantic memory, LogosDB, logosdb MCP, semantic-memory plugin, global memory, thinking traces.
LogosDB — delete a memory row by id or by semantic query match (concise output).
LogosDB — concise semantic search over indexed memory (optional time window).
| name | index |
| description | LogosDB — incremental index of a file or directory (changed/new files only; fast refresh, concise output). |
| argument-hint | <path> [--namespace=name] [-n name] |
| model | sonnet |
User invoked with: $ARGUMENTS
Index a file or directory into LogosDB for semantic search using incremental mode (requires logosdb-mcp-server ≥ 0.7.11).
Parse the arguments:
--namespace=<name> or -n <name> sets the collection name (default: code).Call the logosdb_index_file MCP tool with the resolved path, namespace, and incremental: true.
Incremental: new or changed files only; changed files have old chunks removed first; unchanged files are skipped; directory runs also drop chunks for files removed from disk (see LogosDB mcp/README.md). Cost: proportional to changed chunks, not the whole tree — suitable for session-start refresh or after edits when the user wants memory aligned with disk.
Session load (semantic-memory plugin): the model-invoked semantic-memory skill requires /index . at every session start when the plugin is active.
Respond with one line only. Do not echo the raw tool JSON. Use the tool result fields indexed, indexed_files, skipped_files, pruned_files, and namespace:
Indexed {indexed} chunks · {indexed_files} updated · {skipped_files} skipped · {pruned_files} pruned → '{namespace}'
If indexed_files == 0 && skipped_files > 0, append in the same line: (no changes since last index — touch the file or delete the namespace to force re-index).
Setup, namespaces, and CLAUDE.md template: skills/semantic-memory/SKILL.md and skills/semantic-memory/references/claude-md-template.md.