Local Agent Memory v1

Build, maintain, or improve a layered local memory system for OpenClaw-style agents using markdown files instead of database-backed memory. Use when creating...

Security Scan

This is an instruction-only skill that advises an agent how to store and maintain local, file-based memory in markdown files; its requirements and instructions are consistent with that purpose and it does not request credentials or install code.

Purpose & Capability

Name, description, and included reference files describe a local, file-based memory system and the skill does not request unrelated binaries, environment variables, or network access. The requested capabilities align with the stated purpose.

Instruction Scope

SKILL.md and references only instruct reading and writing workspace markdown files (daily, semantic, procedural, and an index). The docs also advise re-verifying facts (e.g., files, paths, versions, environment) before taking actions — this gives the agent discretionary scope to read local files or environment state when verifying, which is reasonable for the task but worth being explicit about and limiting to the agent's workspace unless the user consents.

Install Mechanism

No install spec or code files are present; the skill is instruction-only so nothing is downloaded or written to disk by an installer.

Credentials

The skill declares no required environment variables, credentials, or config paths. The guidance to re-check environment state is contextual and does not imply hidden credential access.

Persistence & Privilege

always is false and the skill does not request persistent system privileges or attempt to modify other skills or global agent configuration. It will rely on the agent executing file I/O when invoked, which is expected for a local memory workflow.

Assessment

This skill is an instruction-only guide for organizing agent memory as markdown files and appears coherent with that purpose. Before installing or using it, be aware that: the agent following these instructions will read and write files in your workspace (MEMORY.md, memory/*); if you want to limit scope, run the agent in a sandboxed directory or under version control and require explicit permission before any large migrations or bulk edits; ensure backups of important files; and confirm the agent asks for confirmation before making sweeping changes. If you need the agent to avoid reading environment variables or other repositories on disk, state that restriction explicitly in the prompt or workspace policy.

Build or refine a reliable file-based memory system for an agent.

Core workflow

1. Create or inspect these layers:
   • memory/YYYY-MM-DD.md
   • memory/semantic/
   • memory/procedural/
   • MEMORY.md
2. Keep MEMORY.md lightweight and routing-oriented.
3. Put stable facts in semantic files.
4. Put repeatable methods in procedural files.
5. Treat memory as a hint/index layer, not unquestionable truth.
6. Re-verify current facts before taking real actions based on remembered information.
7. Write destination files first, then update MEMORY.md only if the change deserves long-term indexing.

Decision rules

Use daily memory for

• new events
• one-off attempts
• temporary troubleshooting detail
• anything not yet proven reusable

Use semantic memory for

• stable user preferences
• durable environment facts
• platform constraints
• lasting architecture or governance decisions

Use procedural memory for

• repeatable workflows
• checklists
• maintenance routines
• methods likely to be reused across sessions

Maintenance pattern

Run a lightweight dream/consolidation pass when memory starts to sprawl:

• read MEMORY.md
• read recent daily logs
• identify repeated facts or workflows
• extract stable facts into semantic memory
• extract repeatable methods into procedural memory
• prune low-value or duplicated summary lines from MEMORY.md

Run a deeper pass for large daily logs or when the topic tree needs restructuring.

Guardrails

• Do not let MEMORY.md become a diary.
• Do not promote everything that looks interesting.
• Do not rely on stale remembered facts for real actions.
• Do not mix memory maintenance with unrelated code changes unless the user asked for both.
• Prefer a few clear topic files over many overlapping files.

References

Read these only as needed:

• references/architecture.md for the memory model and core disciplines
• references/setup.md for minimum structure and topic layout
• references/maintenance.md for governance and consolidation rules