
from mnemonic
Agent memory/capture workflow: triggers, silent-capture protocol, minimal memory format, and recall/search commands for saving and retrieving structured agent m
Mnemonic Core defines a structured memory system for agents: trigger phrases for capture and recall, silent-capture rules, file formats, namespaces, and operational commands (capture, search, status). It provides concrete workflows and enforcement rules to ensure reliable memory creation and deduplication.
Use when agents must persist decisions, learnings, patterns, or blockers from conversations; when you need a predictable, auditable memory format and retrieval hooks; or when building agent features that require long-term state. Do NOT use for purely hypothetical or assistant-generated suggestions.
Inferred compatibility: agents able to run shell/CLI commands and write files (Claude, Copilot-style tooling, custom agent runtimes).
Mnemonic Core is a memory/capture skill for agents that defines trigger phrases, namespace taxonomy, and a silent-capture protocol. It provides structured YAML-based memory files with UUID, provenance, and decay metadata. No scripts were bundled, so audit is static-only. The SKILL.md is well-organized with clear trigger phrases and deduplication workflow, though it depends on the mnemonic CLI tool being installed.
Clean skill with no security concerns beyond minor unquoted env var usage. Good namespace design and silent-capture protocol. References/ directory mentioned but not bundled — skill relies on external tooling (mnemonic CLI) to function fully.
Mnemonic Search
Search and synthesize across mnemonic memory files (namespaces, tags, types, full-text) to answer questions and surface sources.
Mnemonic Setup
Installs and configures a filesystem-backed mnemonic memory system for Claude-style agents, creating directory structure, CLAUDE.md hooks, and an initial contex