Back to skill

Security audit

Google Health

Security checks for vulnerabilities and agentic risk

Overview

ClawScan rates this skill suspicious with medium confidence, mainly because scanners report that authentication can overwrite or delete some unrelated local files and that large responses may be silently truncated.

Consider the reported local-file overwrite or deletion risk and possible incomplete results before installing. The Skillspector analysis was partial because several local references could not be resolved, so the evidence is incomplete; the verdict is suspicious, not a claim that the author acted maliciously.

Vulnerability Patterns
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (9)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The unconditional promise not to modify unrelated existing files is undermined by accepting empty files as owned and recognizing legacy caches through a directory-name suffix rather than an exact approved location. A misconfigured or attacker-influenced cache path can consequently cause unrelated files to be replaced or deleted during legitimate authentication operations; the read-only remote API does not protect local files.

Content

No source excerpt is available for this finding.

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · internal/auth/tokencache.go (reported line 218)May include surrounding context.

go
// persistingSource wraps a refreshing TokenSource and re-saves the token to disk
// whenever .Token() yields a newer one (an access-token rotation after a
// refresh). x/oauth2 handles refresh-token rotation internally; we only need to
// persist the result so the on-disk access token stays fresh between runs
// (GOAL.md §8 — ghealth doesn't do this; we should).
type persistingSource struct {
	base               oauth2.TokenSource

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · internal/auth/tokencache.go (reported line 243)May include surrounding context.

go
}

// changed reports whether tok differs from the last persisted token in a way
// worth writing (a new access token or a new expiry).
func (p *persistingSource) changed(tok *oauth2.Token) bool {
	if p.last == nil {
		return true

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · internal/cli/auth.go (reported line 120)May include surrounding context.

go
report := authStatusReport{TokenPath: cfg.TokenCache}
			if tok != nil {
				report.Present = true
				report.Valid = tok.Valid() // non-empty access token AND not expired.
				if !tok.Expiry.IsZero() {
					report.Expiry = tok.Expiry.UTC().Format(time.RFC3339)
				}

RawGet silently truncates large responses despite the manifest's complete-output guarantee.

Medium
Category
Excessive Agency
Confidence
97% confidence
Finding

LimitReader makes an oversized body appear to end normally, so RawGet can return success with only the first 32 MiB rather than the complete response. This is a data-integrity issue for health-data consumers, although downstream JSON validation may reject the truncated result rather than silently accept it, and exploitation requires an oversized response from the permitted service.

Content

No source excerpt is available for this finding.

Existing empty or whitespace-only files can be overwritten or deleted despite the permission limiting modifications to the tool's own token cache.

Medium
Category
Excessive Agency
Confidence
99% confidence
Finding

The ownership guard explicitly accepts any existing empty or whitespace-only regular file. SaveToken can then replace it with credentials and DeleteToken can remove it, violating the local-file boundary when the configured cache path points to an unrelated file; atomic replacement does not correct this authorization error.

Content

No source excerpt is available for this finding.

Permission 'env: the five GOOGLE_HEALTH_* variables listed under envVars' is declared but no corresponding code capability (env) was detected.

Low
Category
MCP Least Privilege
Confidence
65% confidence
Finding

Declared permissions with no matching code capability may indicate removed functionality or pre-staging for future abuse.

Content

No source excerpt is available for this finding.

Permission 'file_read: config.json, the OAuth token cache, and a legacy token cache at the older <user config dir> path when migrating it' is declared but no corresponding code capability (file_read) was detected.

Low
Category
MCP Least Privilege
Confidence
65% confidence
Finding

Declared permissions with no matching code capability may indicate removed functionality or pre-staging for future abuse.

Content

No source excerpt is available for this finding.

The sessions documentation promises no filtering, but the parsed view drops undateable exercise points.

Low
Category
Excessive Agency
Confidence
99% confidence
Finding

The comment at L022–L025 says sessions applies 'NO filtering' and returns every exercise type logged. However, L135–L137 skips points without a parseable start time, so the parsed output is not necessarily complete. This behavior is accurately disclosed in the manifest and runtime warning, but contradicts the code documentation's unconditional completeness claim.

Content

No source excerpt is available for this finding.