owlmem@0.9.13
OwlMem command line: connect your AI apps to OwlMem in one command (npx -y owlmem install), an end-to-end encrypted vault for your AI agents, and a local MCP server.
AI Security Review
scanned 20h ago · by lpm-firewall-aiThe flagged Claude credential flow is gated by the owner’s `read_quota` setting and sends the token to Anthropic’s own usage endpoint. However, auto-wake also periodically reports usage totals from local AI session logs to OwlMem without clearly disclosing that collection in its user-facing setup materials, so this warrants a telemetry warning.
Static reason
High-risk behavior combination matched malicious policy.
Trigger
A user runs `owlmem wake enable` or `owlmem install --wake`, pairs the computer, and the background service starts.
Impact
OwlMem and agents in the space can see per-app usage totals and quota status. No transcript content is reported by this flow; the concern is undisclosed runtime telemetry.
Mechanism
The service scans Claude Code and Codex session logs, aggregates usage counts and status, and posts the report to OwlMem about every ten minutes. A separate quota path sends a Claude Code token only to api.anthropic.com when the owner enabled `read_quota`.
AI rationale
The flagged Claude credential flow is gated by the owner’s `read_quota` setting and sends the token to Anthropic’s own usage endpoint. However, auto-wake also periodically reports usage totals from local AI session logs to OwlMem without clearly disclosing that collection in its user-facing setup materials, so this warrants a telemetry warning.
Evidence
owlmem.mjsREADME.md~/.claude/projects/**/*.jsonl~/.codex/sessions/**/*.jsonl~/.config/owlmem/wake/usage.json
Network endpoints2
owlmem.com (configurable via OWLMEM_URL)api.anthropic.com
Decision evidence
public snapshotAI called this Suspicious at 83.0% confidence as Dangerous Capability with medium false-positive risk.
Evidence for warning
- When auto-wake runs, it scans Claude Code and Codex session logs for token usage and reports daily per-app totals so agents in the space can see usage.
- The source says only numeric usage data leaves the machine; it does not send transcript text or paths in that report.
- The usage report runs periodically in the background after auto-wake starts.
- The README requires an explicit `wake enable` command or `install --wake` to turn on auto-wake.
- Claude Code’s access token is read only when OwlMem’s live configuration has `read_quota` enabled, then sent to Anthropic’s own usage endpoint.
Evidence against
- Auto-wake is an opt-in feature that is paired through the OwlMem console.
- The source describes the usage report as serving a team visibility feature, and limits it to usage counts and quota status.
- No install lifecycle scripts are declared; the manifest exposes a CLI executable.
Behavioral surface
ChildProcessCryptoEnvironmentVarsFilesystemNetwork
Source & flagged code
10 flaggedowlmem.mjsView file
16// owlmem vault ls
L17: // owlmem vault get <name> [field] prints the item (or one field) to stdout
L18: // owlmem vault put <name> [--desc text] < value-or-json
...
L39: import { readFileSync, writeFileSync, mkdirSync, existsSync, chmodSync, copyFileSync, rmSync, statSync, realpathSync, openSync, closeSync, appendFileSync, renameSync, readdirSync, ...
L40: import { gunzipSync } from "node:zlib";
L41: import { homedir, hostname, userInfo } from "node:os";
...
L43: import { fileURLToPath } from "node:url";
L44: import { spawn, spawnSync } from "node:child_process";
L45:
L46: export const VERSION = "0.9.13";
L47: const BASE = (process.env.OWLMEM_URL || "https://owlmem.com").replace(/\/$/, "");
L48: const DIR = process.env.OWLMEM_HOME || join(process.env.XDG_CONFIG_HOME || jo
Critical
Credential Exfiltration
Source appears to send environment or credential material to an external endpoint.
owlmem.mjsView on unpkg · L1616Trigger-reachable chain: manifest.bin -> owlmem.mjs
L16: // owlmem vault ls
L17: // owlmem vault get <name> [field] prints the item (or one field) to stdout
L18: // owlmem vault put <name> [--desc text] < value-or-json
...
L39: import { readFileSync, writeFileSync, mkdirSync, existsSync, chmodSync, copyFileSync, rmSync, statSync, realpathSync, openSync, closeSync, appendFileSync, renameSync, readdirSync, ...
L40: import { gunzipSync } from "node:zlib";
L41: import { homedir, hostname, userInfo } from "node:os";
...
L43: import { fileURLToPath } from "node:url";
L44: import { spawn, spawnSync } from "node:child_process";
L45:
L46: export const VERSION = "0.9.13";
L47: const BASE = (process.env.OWLMEM_URL || "https://owlmem.com").replace(/\/$/, "");
L48: const DIR = process.env.O
Findings
2 Critical8 High3 Medium3 Low
CriticalCredential Exfiltrationowlmem.mjs
CriticalTrigger Reachable Dangerous Capabilityowlmem.mjs
HighChild Processowlmem.mjs
HighShellowlmem.mjs
HighSame File Env Network Executionowlmem.mjs
HighCommand Output Exfiltrationowlmem.mjs
HighSandbox Evasion Gated Capabilityowlmem.mjs
HighTrigger Reachable Credential Exfiltrationowlmem.mjs
HighTrigger Reachable Command Output Exfiltrationowlmem.mjs
HighSemantic Analysis Limitedowlmem.mjs
MediumNetwork
MediumEnvironment Vars
MediumStructural Risk Force Deep Review
Affected versions and remediation
This report applies to owlmem@0.9.13.
See all versions of owlmem for other recorded verdicts.
- Review the evidence and your use of owlmem@0.9.13 before allowing it. Restrict the permissions described in this report.
- Choose an independently verified alternative or release. This report does not establish that other versions are safe.
Evidence last updated: .
Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.