loci

Structured memory system using domains, decay, and links to store, recall, and maintain important AI context as an organized memory palace.

Security Scan

The skill's files and runtime instructions match its stated purpose (a local, file-backed memory palace); it reads/writes a local JSON file, has no network calls, and requests no credentials or unusual permissions.

Purpose & Capability

Name/description describe a structured memory system and the bundled script (scripts/loci.mjs) implements storing, recalling, walking, pruning, linking, exporting, and domain management. Required resources (Node.js, local palace file) align with the stated purpose; there are no unrelated credentials or external services requested.

Instruction Scope

SKILL.md instructs the agent to run the included Node CLI which reads/writes a palace file at ~/.openclaw/workspace/loci_palace.json by default and can load a user-supplied template file. This is consistent with the purpose. Note: the tool will read any template path the user supplies and will write the palace JSON to disk, so stored content will persist locally.

Install Mechanism

No install spec; the skill is instruction-only with a bundled script. There are no downloads, package installs, or external installers declared. Risk is low because only local code included and no remote fetches are performed.

Credentials

The skill requires no environment variables, no credentials, and no config paths beyond the single palace JSON it manages. That scope is proportional to a local memory-store utility.

Persistence & Privilege

always is false and disable-model-invocation is false (normal). The skill only writes to its own palace file (by default under the user's home) and does not modify other skills or global agent settings. Autonomous invocation is allowed by platform default but not combined with other red flags.

Assessment

This skill appears to be what it claims: a local, file-backed memory palace implemented as a Node CLI. Before installing or running: (1) review any memories you store — they are persisted to ~/.openclaw/workspace/loci_palace.json (or to any path you pass) and could contain sensitive data, so avoid storing secrets; (2) ensure the file's filesystem permissions and backups meet your privacy needs; (3) the tool can load a template file you supply—only provide trusted templates to avoid reading unexpected local files; (4) running the skill executes the included Node script, so ensure Node is available and you trust the bundled code; (5) because the skill is allowed to be invoked autonomously by the agent (platform default), consider whether you want the agent to call it without prompting. Overall there are no network calls or credential requests, but treat stored content on disk as persistent data and manage accordingly.

loci — Structured Memory for AI Agents

GitHub: https://github.com/bayhax/loci | ClawHub: clawhub install loci

Quick Start

LOCI="node <skill_dir>/scripts/loci.mjs"

# Initialize palace (first time only)
$LOCI init

# Store a memory
$LOCI store work "Switched to Claude Opus model per user preference" --tag model --tag preference

# Recall memories
$LOCI recall "what model does the user prefer"

# Walk through palace (do this during heartbeats)
$LOCI walk

# See overview
$LOCI status

Core Concepts

Domains — Categories that organize memories (like rooms). Each has a capacity limit. Default domains: work, knowledge, people, tools, preferences, archive.

Memories — Individual pieces of context stored in a domain. Each has:

• Unique ID (e.g. e3a7f2c1)
• Content text
• Tags for categorization
• Links to related memories
• Weight that decays over time

Decay — Memories lose weight exponentially based on time since last access. Formula: weight = base_weight × e^(-decay_rate × days_since_access) Default decay rate: 0.05 (half-life ≈ 14 days).

Links — Bidirectional connections between related memories across any domain.

Commands

Command	Purpose
init	Create new palace (once)
store <domain> <content>	Add memory. Options: --tag, --link
recall <query>	Search. Options: --domain, --top N
walk	Traverse all memories, report health
prune	Remove decayed memories. Options: --threshold, --dry-run
status	Overview of all domains
inspect <id>	View memory details + links
link <id1> <id2>	Connect two memories
domains	List/add/remove domains
export	Export as markdown or JSON. Options: --format md|yaml

Integration with Heartbeats

During heartbeat walks, run:

$LOCI walk --decay 0.05

This reports domain health, identifies fading memories, and updates the walk timestamp. Periodically follow with $LOCI prune --dry-run to review candidates for removal.

When to Store vs. When to Skip

Store: Decisions, user preferences, environment quirks, lessons learned, important people/relationships, recurring patterns.

Skip: Transient task details, one-off commands, things already in daily memory files.

Rule of thumb: If future-you would benefit from knowing this in 2 weeks, store it.

Palace File

Default location: ~/.openclaw/workspace/loci_palace.json Override with --palace PATH on any command. Format: JSON. Zero external dependencies — only requires Node.js (ships with OpenClaw).