| name | remember |
| description | Store in ferrex memory (no args = session summary, with args = store that info) |
| argument-hint | Optional info to store |
Store information in ferrex memory using the ferrex__store MCP tool.
Auto-set namespace from the git repository name (basename of the repo root, not the worktree). If not in a git repo, use the basename of the working directory.
If arguments provided, detect memory type from content:
-
Semantic triple — if the content is a fact, decision, or relationship:
- Extract
subject, predicate, object
- Example: "nix-darwin uses lze for plugin loading" -> subject: "nix-darwin", predicate: "uses", object: "lze for plugin loading"
- Example: "We decided to use ferrex instead of qdrant" -> subject: "project", predicate: "decided", object: "use ferrex instead of qdrant"
- Set
memory_type: "semantic"
-
Procedural — if the content describes a workflow, process, or how-to:
- Store as
content with memory_type: "procedural"
-
Episodic — events, errors, observations, anything else:
- Store as
content with memory_type: "episodic"
Always set:
namespace: git repository name (basename of the repo root, not the worktree); basename of the working directory if not in a git repo
entities: at minimum the project name. Add relevant concept names (tools, modules, patterns mentioned).
context: branch name, relevant file paths
confidence: 1.0 unless the user expresses uncertainty
If no arguments, summarize the conversation and store as episodic with entities for the project, branch, and key topics discussed.
After storing: Confirm what was stored — type, entities, namespace, and memory ID.