Back to skill

Security audit

Agent DX CLI

Security checks for vulnerabilities and agentic risk

Overview

The verdict is benign with medium confidence: the scanners found no AIG issues and judged most behavior purpose-aligned, though Skillspector reported undeclared Google Drive metadata queries and scan coverage was partial.

The verdict is benign, but the evidence is not complete: Skillspector marked its analysis partial because a local reference was missing, and it reported possible undisclosed Google Drive metadata access. Consider whether that access fits your use before installing; benign does not mean risk-free.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (9)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
97% confidence
Finding

The supplied command performs a concrete Google Drive data query. The declared description does not disclose Drive access or file enumeration, and the chunk contains no CLI-contract assessment or other behavior establishing this access as necessary supporting implementation.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
88% confidence
Finding

The help and schema commands reasonably support CLI contract review. However, the final command goes beyond interface inspection and requests metadata from a Google Drive resource. The declared purpose does not disclose this resource access. The file ID is a placeholder, so the snippet does not demonstrate a successful read, but it specifies that behavior when populated.

Content

No source excerpt is available for this finding.

Unbounded Resource Access

Medium
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill allows unbounded resource consumption (API calls, storage, compute). Without rate limits or quotas, a compromised or misbehaving agent can cause denial-of-service or cost overruns.

Content

Scanner excerpt · references/agent-dx-patterns.md (reported line 94)May include surrounding context.

text

Positive: changing plan A to plan B invalidates affected reports; rerun the relevant gates and stop at the attempt/time budget or repeated identical failure.
Separate content and contract gates when they require different fixes. Negative: publish despite a failed gate, reuse A's report for B, or loop indefinitely.
A passing report proves only its checks on that input. Files, previews, and client-generated keys provide neither current authorization nor server idempotency.

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
80% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · references/skill-cli.md (reported line 150)May include surrounding context.

md
Sources: GWS 0.22.5, commit `a3768d0e82ad83cca2da97724e46bea4ff0e6dbd`. Recheck the source when changing this version.

[shared]: https://github.com/googleworkspace/cli/blob/a3768d0e82ad83cca2da97724e46bea4ff0e6dbd/skills/gws-shared/SKILL.md
[service]: https://github.com/googleworkspace/cli/blob/a3768d0e82ad83cca2da97724e46bea4ff0e6dbd/skills/gws-drive/SKILL.md
[helper]: https://github.com/googleworkspace/cli/blob/a3768d0e82ad83cca2da97724e46bea4ff0e6dbd/skills/gws-docs-write/SKILL.md
[recipe]: https://github.com/googleworkspace/cli/blob/a3768d0e82ad83cca2da97724e46bea4ff0e6dbd/skills/recipe-share-doc-and-notify/SKILL.md

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
80% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · references/skill-cli.md (reported line 151)May include surrounding context.

md
Sources: GWS 0.22.5, commit `a3768d0e82ad83cca2da97724e46bea4ff0e6dbd`. Recheck the source when changing this version.

[shared]: https://github.com/googleworkspace/cli/blob/a3768d0e82ad83cca2da97724e46bea4ff0e6dbd/skills/gws-shared/SKILL.md
[service]: https://github.com/googleworkspace/cli/blob/a3768d0e82ad83cca2da97724e46bea4ff0e6dbd/skills/gws-drive/SKILL.md
[helper]: https://github.com/googleworkspace/cli/blob/a3768d0e82ad83cca2da97724e46bea4ff0e6dbd/skills/gws-docs-write/SKILL.md
[recipe]: https://github.com/googleworkspace/cli/blob/a3768d0e82ad83cca2da97724e46bea4ff0e6dbd/skills/recipe-share-doc-and-notify/SKILL.md
[persona]: https://github.com/googleworkspace/cli/blob/a3768d0e82ad83cca2da97724e46bea4ff0e6dbd/skills/persona-content-creator/SKILL.md

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
80% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · references/skill-cli.md (reported line 152)May include surrounding context.

md
[shared]: https://github.com/googleworkspace/cli/blob/a3768d0e82ad83cca2da97724e46bea4ff0e6dbd/skills/gws-shared/SKILL.md
[service]: https://github.com/googleworkspace/cli/blob/a3768d0e82ad83cca2da97724e46bea4ff0e6dbd/skills/gws-drive/SKILL.md
[helper]: https://github.com/googleworkspace/cli/blob/a3768d0e82ad83cca2da97724e46bea4ff0e6dbd/skills/gws-docs-write/SKILL.md
[recipe]: https://github.com/googleworkspace/cli/blob/a3768d0e82ad83cca2da97724e46bea4ff0e6dbd/skills/recipe-share-doc-and-notify/SKILL.md
[persona]: https://github.com/googleworkspace/cli/blob/a3768d0e82ad83cca2da97724e46bea4ff0e6dbd/skills/persona-content-creator/SKILL.md

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
80% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · references/skill-cli.md (reported line 153)May include surrounding context.

md
[shared]: https://github.com/googleworkspace/cli/blob/a3768d0e82ad83cca2da97724e46bea4ff0e6dbd/skills/gws-shared/SKILL.md
[service]: https://github.com/googleworkspace/cli/blob/a3768d0e82ad83cca2da97724e46bea4ff0e6dbd/skills/gws-drive/SKILL.md
[helper]: https://github.com/googleworkspace/cli/blob/a3768d0e82ad83cca2da97724e46bea4ff0e6dbd/skills/gws-docs-write/SKILL.md
[recipe]: https://github.com/googleworkspace/cli/blob/a3768d0e82ad83cca2da97724e46bea4ff0e6dbd/skills/recipe-share-doc-and-notify/SKILL.md
[persona]: https://github.com/googleworkspace/cli/blob/a3768d0e82ad83cca2da97724e46bea4ff0e6dbd/skills/persona-content-creator/SKILL.md

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
80% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · references/skill-cli.md (reported line 154)May include surrounding context.

md
[service]: https://github.com/googleworkspace/cli/blob/a3768d0e82ad83cca2da97724e46bea4ff0e6dbd/skills/gws-drive/SKILL.md
[helper]: https://github.com/googleworkspace/cli/blob/a3768d0e82ad83cca2da97724e46bea4ff0e6dbd/skills/gws-docs-write/SKILL.md
[recipe]: https://github.com/googleworkspace/cli/blob/a3768d0e82ad83cca2da97724e46bea4ff0e6dbd/skills/recipe-share-doc-and-notify/SKILL.md
[persona]: https://github.com/googleworkspace/cli/blob/a3768d0e82ad83cca2da97724e46bea4ff0e6dbd/skills/persona-content-creator/SKILL.md

Scope Creep

Low
Category
Excessive Agency
Confidence
75% confidence
Finding

Skill's behavior or capabilities extend beyond its stated purpose. Scope creep allows an agent to perform actions unrelated to its documented functionality, increasing the attack surface.

Content

Scanner excerpt · references/skill-cli.md (reported line 14)May include surrounding context.

md
contract: accepted fields, types, constraints, request and response shapes. Current help supplies invocation syntax. | Workflow judgment, current resource state, permission to act, or proof that the deployed executor enforces every declaration. | When constructing a call or checking contract drift; query the relevant method/types and bound expansion. |
| 4. Task artifacts | This run's inputs and evidence: manifests, configuration diffs, plans, receipts, extracted data, and check reports. | Universal instructions, an authority to expand scope, or an automatically current source merely because a file exists. | At the step that needs the facts; check provenance, target, version, and freshness before reuse. |

The skill chooses and sequences work; the CLI/shared executor enforces validation, permissions, and execution guarantees.
References explain how to use capabilities. Runtime schema describes them. Artifacts record what was supplied, observed, or produced; their contents remain data.