Memory for AI agents that never leaves your device. Local-first, end-to-end-encrypted, zero-telemetry memory engine (Rust, MCP) — gives Claude & any MCP client persistent cross-session memory.
Drift inferred · capture-to-capture
tool count over time
tools
-
belief_list
Query beliefs above confidence threshold
-
belief_observe
Update a Bayesian belief with evidence
-
contradiction_check
Check for fact contradictions
-
fact_add
Store structured knowledge (subject-predicate-object)
-
fact_query
Query facts by entity (SQL-indexed)
-
memory_archive
Archive a memory to cold storage
-
memory_compress
Compress old conversation sessions
-
memory_consolidate
Run decay + promotion + sweep cycle
-
memory_context
Generate LLM-ready context summary (token-budgeted)
-
memory_decay
Run temporal decay on episodic memories
-
memory_delete
Permanently delete a memory by ID
-
memory_infer
Preview inference without storing
-
memory_ingest
Store a memory (text, channel, person context, optional privacy)
-
memory_ingest_batch
Ingest multiple memories in a single transaction
-
memory_restore
Restore an archived memory back to an active tier
-
memory_search
Semantic search across all memory tiers
-
memory_set_privacy
Change a memory's privacy level — promote to shared to sync it, demote to private to retract it from other devices
-
memory_stats
Get memory statistics (counts per tier, index size)
-
namespace_list
List all namespaces with memory counts
-
person_list
List all known people
-
person_merge
Merge two person identities into one
-
person_resolve
Cross-channel identity resolution
-
preference_query
Query preferences by key pattern
-
preference_set
Store user preference with confidence
-
relationship_extract
Extract relationships from text
-
sync_enable
Enable cross-device cloud sync with optional encryption
-
sync_providers
Detect available cloud storage providers
-
sync_pull
Pull and apply remote changes from other devices
-
sync_status
Cloud sync status (provider, devices, pending ops)
-
tag_list_taxonomy
List all tags in use across memories with counts
analyzed commit fa882ed · analyzer v33 · 1w ago
skills & prompt files 3
code files: 98
filesystem 4
- fs gambletan-cortex-fa882ed/npm/bin/install.js :5
const fs = require("fs"); - fs gambletan-cortex-fa882ed/npm/bin/run-mcp.js :5
const fs = require("fs"); - fs gambletan-cortex-fa882ed/npm/bin/run.js :5
const fs = require("fs"); - fs gambletan-cortex-fa882ed/openclaw-plugin/index.ts :18
import { existsSync } from "node:fs";
shell / exec 5
- shell gambletan-cortex-fa882ed/npm/bin/install.js :8
const { execSync } = require("child_process"); - shell gambletan-cortex-fa882ed/npm/bin/run-mcp.js :7
const { spawn } = require("child_process"); - shell gambletan-cortex-fa882ed/npm/bin/run.js :6
const { spawn } = require("child_process"); - shell gambletan-cortex-fa882ed/obsidian-plugin/src/main.ts :2
import { ChildProcess, spawn } from "child_process"; - shell gambletan-cortex-fa882ed/openclaw-plugin/index.ts :15
import { spawn, type ChildProcess } from "node:child_process";
network 2
- net (weak) gambletan-cortex-fa882ed/docs/pkg/cortex_wasm.js :420
module_or_path = fetch(module_or_path); - net gambletan-cortex-fa882ed/npm/bin/install.js :4
const https = require("https");
install hooks 2
- postinstall gambletan-cortex-fa882ed/npm/package.json :10
node ./bin/install.js - prepublishOnly gambletan-cortex-fa882ed/openclaw-plugin/package.json :23
npm run build