registry / @canonry/canonry / 7.23.0
@canonry/canonry@7.23.0
Self-hosted AI visibility (AEO) platform: track how ChatGPT, Claude, Gemini, and Perplexity cite your domain, join it with Search Console, GA4, server-side traffic, and paid media, and fix what you find through agent tools (CLI, REST, MCP). Local SQLite.
AI Security Review
scanned 25d ago · by lpm-firewall-aiInspected source shows an automatic, opt-out sync of bundled agent skills into detected Claude or Codex directories during CLI startup. This supports a warning for agent extension lifecycle risk; the other reviewed indicators did not establish a concrete attack.
Static reason
High-risk behavior combination matched malicious policy.; previous stored version diff introduced dangerous source
Trigger
Invoking the CLI when existing skill targets are found and the saved version or sync interval qualifies.
Impact
Bundled agent instructions may be added or updated without a prompt on that invocation.
Mechanism
The startup path calls automatic sync, which reconciles bundled skills into detected skill directories; an environment flag can disable syncing.
AI rationale
Inspected source shows an automatic, opt-out sync of bundled agent skills into detected Claude or Codex directories during CLI startup. This supports a warning for agent extension lifecycle risk; the other reviewed indicators did not establish a concrete attack.
Evidence
dist/chunk-KWBFKTOA.jsdist/chunk-UF7MH6TU.jsdist/cli.js~/.claude/skills/~/.codex/skills/
Decision evidence
public snapshotAI called this Suspicious at 84.0% confidence as Dangerous Capability with medium false-positive risk.
Evidence for warning
- The CLI starts automatic skill syncing during normal command startup; the sync may run after a package version change or elapsed interval when it finds existing Canonry skill targets.
- The sync installs bundled skills into Claude skill directories and may reconcile their files. Its disable flag is an opt-out, and the inspected startup path shows no consent prompt.
- The flagged agent configuration locations show skill checks, skill-directory detection, plugin update guidance, and CLI configuration paths; these excerpts do not establish a separate control hijack.
- The TypeSafe client sends a caller-provided API key and request to its named service endpoint; this source does not establish credential forwarding to an unrelated destination.
Evidence against
- package.json describes the package as an AI visibility platform with agent tools, consistent with providing bundled agent skills.
- The TypeSafe request uses a fixed package-named service endpoint and the API key supplied to the client; inspected code does not show an unrestricted operator-selected destination.
Behavioral surface
Source & flagged code
9 flaggeddist/chunk-46JZR2IV.jsView file
936gscQueryDailyTotals: () => gscQueryDailyTotals,
L937: gscSearchData: () => gscSearchData,
L938: gscUrlInspections: () => gscUrlInspections,
...
L4057: try {
L4058: return JSON.parse(value);
L4059: } catch {
...
L4591: // ga_connections
L4592: // WARNING: private_key is authentication material; consider storing in config.yaml per AGENTS.md
L4593: `CREATE TABLE IF NOT EXISTS ga_connections (
...
L11883: try {
L11884: const parsed = new URL(/^https?:\/\//i.test(trimmed) ? trimmed : `https://${trimmed}`);
L11885: return parsed.hostname.toLowerCase().replace(/\.$/, "") || null;
Critical
Hardcoded Runtime Data Exfiltration
Source sends credentials or rich application records to a package-controlled external receiver enabled by default.
dist/chunk-46JZR2IV.jsView on unpkg · L936936Trigger-reachable chain: manifest.exports -> dist/index.js -> dist/chunk-46JZR2IV.js
L936: gscQueryDailyTotals: () => gscQueryDailyTotals,
L937: gscSearchData: () => gscSearchData,
L938: gscUrlInspections: () => gscUrlInspections,
...
L4057: try {
L4058: return JSON.parse(value);
L4059: } catch {
...
L4591: // ga_connections
L4592: // WARNING: private_key is authentication material; consider storing in config.yaml per AGENTS.md
L4593: `CREATE TABLE IF NOT EXISTS ga_connections (
...
L11883: try {
L11884: const parsed = new URL(/^https?:\/\//i.test(trimmed) ? trimmed : `https://${trimmed}`);
L11885: return parsed.hostname.toLowerCase().replace(/\.$/, "") || null;
Critical
Trigger Reachable Dangerous Capability
Findings
4 Critical5 High3 Medium6 Low
CriticalHardcoded Runtime Data Exfiltrationdist/chunk-46JZR2IV.js
CriticalBrowser Credential Phishingdist/chunk-MRKV43TV.js
CriticalTrigger Reachable Dangerous Capabilitydist/chunk-46JZR2IV.js
CriticalPrevious Version Dangerous Deltadist/cli.js
HighChild Processdist/chunk-46JZR2IV.js
HighCredential Redirect Persistencedist/chunk-3OPV5XBY.js
HighCommand Output Exfiltrationdist/chunk-46JZR2IV.js
HighCross File Remote Execution Contextdist/cli.js
HighSemantic Analysis Limiteddist/chunk-46JZR2IV.js
MediumNetwork
MediumEnvironment Vars
MediumStructural Risk Force Deep Review
Affected versions and remediation
This report applies to @canonry/canonry@7.23.0.
See all versions of @canonry/canonry for other recorded verdicts.
- Review the evidence and your use of @canonry/canonry@7.23.0 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.