The Spaced Repetition Systems for Agents
ReviewAudited by ClawScan on May 10, 2026.
Overview
This is a coherent local spaced-repetition review skill, but it stores review data and is meant to influence the agent’s persistent memory, so users should supervise what gets remembered.
This skill appears benign and purpose-aligned for local spaced-repetition memory review. Before installing, be comfortable running its local Python CLI, avoid putting secrets into cards, and manually check any memory add/delete/update actions so inaccurate cards do not shape future agent behavior.
Findings (3)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Using the skill may execute bundled local Python code and depend on local Python packages even though the registry does not declare them.
The skill expects the agent/user to run bundled Python scripts via `uv`, while the registry metadata declares no install spec or required binaries. This is central to the stated purpose, but setup requirements are not fully declared.
uv run python scripts/main.py status
Run it only from a trusted install location and make sure the expected Python tooling and dependencies are understood before use.
If inaccurate, private, or unwanted information is reviewed into memory, the agent may reuse it in later conversations.
The skill is intentionally designed to change persistent agent memory after review. That is disclosed and purpose-aligned, but wrong or sensitive cards could affect future agent behavior.
Agent memory system: must be updated explicitly by the agent (add/delete/update), based on reflection.
Review each proposed memory add/delete/update action before allowing it to become persistent, and avoid storing secrets or sensitive personal details as cards.
Information placed into cards can remain on disk as local review history.
The local storage layer persists card questions, answers, FSRS card state, and review logs in SQLite. This is expected for spaced repetition, but card content may contain user preferences or corrections.
CREATE TABLE IF NOT EXISTS cards ( ... question TEXT NOT NULL, answer TEXT NOT NULL, card_json TEXT NOT NULL ... );
Treat cards as persistent local memory: do not add secrets, and use the provided override/remove commands to correct or delete unwanted cards.