القائمة
Store, search, and prune persistent agent memory. Supports semantic search, namespace isolation, and selective pruning to keep memory lean and relevant across sessions.
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
| id | memory-manager |
| name | Memory Manager |
| description | Store, search, and prune persistent agent memory. Supports semantic search, namespace isolation, and selective pruning to keep memory lean and relevant across sessions. |
| tags | ["memory","persistence","search"] |
| apiKeyRequired | false |
| commands | [{"name":"search","description":"Semantic search over stored memories","args":["query","limit"]},{"name":"store","description":"Persist a new memory entry","args":["content","namespace"]},{"name":"prune","description":"Remove outdated or low-relevance memories","args":["namespace","before_date"]}] |
Persistent cross-session memory for an agent. Entries survive restarts and session clears.
| Tool | Purpose |
|---|---|
memory_store | Save a new memory entry with optional namespace tag |
memory_search | Full-text + semantic search across stored memories |
memory_list | List memories in a namespace |
memory_delete | Remove a specific memory by id |
Use memory_store to save:
content: "User prefers reports in Markdown with H2 section headers."
namespace: "preferences"
Before starting a long task, call memory_search with the task topic
to surface relevant prior context.
After completing a project, use memory_delete or memory_list + memory_delete
to remove stale entries that are no longer relevant.
preferences — user preferences and style rulesfacts — domain facts learned during interactionstasks — task state that should survive session boundaries<project-id> — project-scoped memories~/.agentflyer/ — no external service required.memory_search uses BM25 full-text search; no vector embedding is needed.