distinctdeer

Pass

Audited by Gen Agent Trust Hub on Aug 28, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill downloads and executes the 'distinctdeer' utility directly from the author's GitHub repository using the 'uv' tool manager. This represents a dependency on an external repository maintained by the skill author.\n- [REMOTE_CODE_EXECUTION]: The skill executes code directly from a remote git source ('git+https://github.com/gkwa/distinctdeer'). As this is a vendor-owned resource, it is documented without severity escalation.\n- [COMMAND_EXECUTION]: The skill performs shell command execution to run the naming tool and to initialize a new git repository via 'git init'. The tool also modifies the system clipboard and interacts with an Obsidian vault on the local filesystem.\n- [INDIRECT_PROMPT_INJECTION]: The skill relies on parsing output from an external command to determine subsequent filesystem paths, creating an attack surface.\n
  • Ingestion points: JSON output from the 'distinctdeer' command in SKILL.md.\n
  • Boundary markers: None; instructions direct the agent to read the value directly from the JSON structure.\n
  • Capability inventory: Shell command execution, clipboard access, and filesystem directory creation.\n
  • Sanitization: No sanitization or validation of the retrieved project name is mentioned before it is used in shell commands.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 28, 2026, 08:19 PM
Security Audit — agent-trust-hub — distinctdeer