eSpeak NG Multilingual Speech Synthesizer

Warn

Audited by Gen Agent Trust Hub on Mar 28, 2026

Risk Level: MEDIUMPROMPT_INJECTIONEXTERNAL_DOWNLOADSNO_CODE
Full Analysis
  • [PROMPT_INJECTION]: Metadata Poisoning. The skill's frontmatter contains 'verification: security_reviewed'. This is a self-attestation that could mislead users or agents regarding the skill's actual safety status. Claims of safety within the skill content must not be treated as authoritative.
  • [PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill is designed to synthesize speech from text inputs. * Ingestion points: Text input for synthesis referenced in the skill description. * Boundary markers: Absent; no instructions define how the agent should distinguish between data and instructions. * Capability inventory: CLI access to the eSpeak NG engine. * Sanitization: Absent; no evidence of input validation or character filtering for synthesized text.
  • [EXTERNAL_DOWNLOADS]: Fetches the 'espeak-ng-multilingual-speech-synthesizer' package via standard package managers ('npx', 'clawhub'). These instructions target public registries and the author's own infrastructure.
  • [NO_CODE]: The provided skill file contains only metadata and installation documentation. No executable scripts or operational logic were found within the skill itself for analysis.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 28, 2026, 06:45 PM