Agent Memory Journal

Durable episodic memory for agents — file-based, inspectable, and easy to review. Use when an agent needs to keep a working journal of what happened, what ma...

Security Scan

The skill is internally coherent: it implements a local, file-based agent memory system and only reads/writes files under a configurable memory root; it does not request credentials or network access.

Purpose & Capability

Name/description (durable, file-based episodic memory) matches the code and runtime instructions. The code implements journaling, recall, ingestion, promotion, and CLI convenience commands and only requires filesystem access under the configured memory root.

Instruction Scope

SKILL.md and CLI simply describe adding notes, remembering facts, searching, ingesting, and maintenance tasks. The runtime code reads and writes markdown files in the memory directories, updates 'last_seen' metadata, and rebuilds a 'hot' file. There are no instructions or code paths that send data to external endpoints, call system shells, or access unrelated system state beyond the configured filesystem paths.

Install Mechanism

No install spec or external downloads are declared. The package is provided as Python source files; nothing in the manifest attempts to fetch or execute code from remote URLs.

Credentials

The skill declares no required environment variables or credentials and does not attempt to read secrets. All configuration is file-based (config.json under the memory root).

Persistence & Privilege

The skill writes persistent files under the configured memory root (default is '.', or a '.memory' subdir). validate_hot_path permits the configured hot file to be in the memory root or its parent directory, so the hot-file may be placed in the repository/workspace root; this is likely intentional but is a point to check before deploying. always:false and normal model invocation mean it does not force inclusion or elevated platform privileges.

Assessment

This skill appears to do what it says: it stores and searches agent memories on the local filesystem and exposes a Python API and CLI. Before installing, check where the memory root will be (default is the current directory or a '.memory' folder) and inspect .memory/config.json (hot_path) to ensure the 'hot' file won't be written to an unintended location (validate_hot_path allows the memory root or its parent). Because the skill writes persistent files, consider: (1) running it in a controlled directory or container, (2) setting appropriate filesystem permissions, (3) reviewing the code if you need stricter input sanitization or to prevent any chance of storing sensitive secrets, and (4) being aware that an autonomous agent invoking this skill can create or modify files within the allowed paths (this is expected for a memory skill).

Skill: Agent Memory Journal

Manage and retrieve long-term memory using a 3-tier structure (Hot, Warm, Cold).

Primary integration surface: the Python Journal API. The CLI is a convenience layer for shell and cron workflows.

Tools

• agent-memory-journal note <text>: Add episodic note.
• agent-memory-journal remember <text> --category <cat>: Add core memory.
• agent-memory-journal search --query <q>: Search memory.
• agent-memory-journal forget <id>: Supersede a memory.
• agent-memory-journal ingest: Run promotion and rebuild cycle.

Core Categories

• decision: Strategic choices.
• constraint: Hard rules or limitations.
• gotcha: Lessons learned or bugs.
• preference: User or agent preferences.
• capability: New skills or tools.

Guidelines

1. Episodic first: Use note for observations.
2. Explicit curation: Use remember for facts that must persist.
3. Atomic pinning: Use --pinned for items that MUST be in the configured hot file.
4. Verification: Run doctor periodically to ensure integrity.

Agent installation

See docs/agent_install.md for:

• workspace installation
• CLI symlink setup
• startup contract for agents
• hot promotion target override via .memory/config.json