registry / AI-agent control-surface risk / hermes-to-claude
hermes-to-claude
Local HTTP bridge connecting Hermes Agent to Claude Code — no Pro/Max subscription required
This report shows the evidence for hermes-to-claude@0.1.10. 6 of 9 known versions are blocked or flagged. The newest version, 0.2.17, is flagged. See all versions.
AI Security Review
scanned 3mo ago · by lpm-firewall-aiSource inspection confirms concrete postinstall persistence in Claude Code's global settings, not merely scanner similarity. Although no external exfiltration was found, this behavior meets the firewall block boundary for unconsented postinstall AI-agent control-surface mutation.
Static reason
High-risk behavior combination matched malicious policy.; source matched previously finalized malicious package; routed for review; source fingerprint signature matched known malicious package; routed for review
Trigger
npm postinstall, then Claude Code status-line execution
Impact
Persists package-controlled execution in a broad AI-agent control surface and changes the user's existing configuration.
Mechanism
unconsented global Claude Code status-line takeover
Policy narrative
On npm installation, the postinstall script reads and replaces the global ~/.claude/settings.json statusLine entry with `node dist/statusline.mjs`. It stores any pre-existing status-line command in ~/.h2c_user_statusline_cmd; the installed statusline later executes that saved command and package code whenever Claude Code renders its status line. This is an unconsented postinstall mutation of a foreign, user-wide AI-agent control surface.
AI rationale
Source inspection confirms concrete postinstall persistence in Claude Code's global settings, not merely scanner similarity. Although no external exfiltration was found, this behavior meets the firewall block boundary for unconsented postinstall AI-agent control-surface mutation.
Evidence
package.jsonscripts/setup-mcp.cjsdist/statusline.mjsdist/hbridge.mjsdocs/local-mode.md~/.claude/settings.json~/.h2c_user_statusline_cmd~/.claude/skills/h2c/SKILL.md
Network endpoints1
127.0.0.1:<derived-port>/health
Decision evidence
public snapshotAI identified this as a dangerous AI-agent capability at 98.0% confidence with low false-positive risk.
Evidence for policy risk
- package.json runs scripts/setup-mcp.cjs via postinstall.
- scripts/setup-mcp.cjs overwrites ~/.claude/settings.json statusLine.
- Postinstall installs package statusline command without user action.
- dist/statusline.mjs runs on Claude Code status-line invocations.
- Postinstall preserves and later executes prior user statusline command.
- dist/hbridge.mjs exposes task creation that spawns Claude Code.
Evidence against
- No external network endpoint or credential exfiltration was found.
- Observed HTTP traffic is loopback-only.
- The optional packaged skill source is absent, so its copy branch skips.
Behavioral surface
ChildProcessCryptoEnvironmentVarsFilesystemNetworkShell
HighEntropyStringsUrlStrings
Source & flagged code
7 flagged · loading sourcepackage.jsonView file
•scripts.postinstall = node scripts/setup-mcp.cjs
High
Install Time Lifecycle Scripts
Package defines install-time lifecycle scripts.
package.jsonView on unpkgdist/hbridge.mjsView file
•matchType = normalized_sha256
matchedPackage = hermes-to-claude@0.1.9
matchedPath = dist/hbridge.mjs
matchedIdentity = npm:aGVybWVzLXRvLWNsYXVkZQ:0.1.9
similarity = 1.000
summary = normalized source hash matched finalized malicious source
High
Known Malware Source Similarity
Source file is highly similar to a previously finalized malicious package; route for source-aware review.
dist/hbridge.mjsView on unpkg•matchType = malicious_source_fingerprint_signature
signature = 71cbcea900553e17
signatureType = suspicious_hashes
sourceLabel = final_verdict:malicious
matchedPackage = hermes-to-claude@0.1.9
matchedPath = dist/hbridge.mjs
matchedIdentity = npm:aGVybWVzLXRvLWNsYXVkZQ:0.1.9
similarity = 1.000
shingleOverlap = 4
summary = package final verdict is malicious
Findings
1 Critical5 High3 Medium6 Low
CriticalAi Agent Control Hijackscripts/setup-mcp.cjs
HighInstall Time Lifecycle Scriptspackage.json
HighKnown Malware Source Similaritydist/hbridge.mjs
HighKnown Malware Source Similaritydist/statusline.mjs
HighKnown Malware Source Similaritybuild.mjs
HighKnown Malware Source Fingerprint Signaturedist/hbridge.mjs
MediumNetwork
MediumEnvironment Vars
MediumStructural Risk Force Deep Review
LowNon Install Lifecycle Scripts
LowScripts Present
LowWeak Cryptodist/hbridge.mjs
LowFilesystem
LowHigh Entropy Strings
Affected versions and remediation
This report applies to hermes-to-claude@0.1.10.
The versions list below shows the verdict for each recorded version.
- Avoid installing hermes-to-claude@0.1.10. Remove it from direct dependencies and check your lockfile for transitive copies.
- Choose an independently verified alternative or release. This report does not establish that other versions are safe.
- If this version ran, investigate the affected machine and build environment. Rotate credentials it could access and rebuild from a trusted environment.
Evidence last updated: .
Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.