lynx-miniapp

Pass

Audited by Gen Agent Trust Hub on Nov 10, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes command-line tools to provide simulation, verification, and preview functionality for miniapp projects. It executes tma-ide-session for local developer tool sessions and tma preview for generating remote preview QR codes. It also runs a bundled Node.js utility (scripts/precheck_scan.mjs) for project analysis. These actions are restricted by explicit user authorization requirements for remote or destructive operations.
  • [EXTERNAL_DOWNLOADS]: The skill references the tt-ide-cli package, a vendor-specific developer tool maintained by ByteDance for its miniapp platform. This dependency is sourced from the official NPM registry as documented in assets/runtime-tools.lock.json.
  • [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection because it ingests and analyzes user-provided miniapp source code.
  • Ingestion points: Ingests .ttml, .ttss, .js, and .json files from user projects as specified in SKILL.md and the scanner script.
  • Boundary markers: While it lacks specific token-based delimiters, SKILL.md includes a mandatory "Input Safety" rule that instructs the AI agent to treat all user artifacts strictly as data and to disregard any instructions contained within them.
  • Capability inventory: The skill can execute shell commands (tma, node) and read local files within the project scope.
  • Sanitization: The included scanning script (precheck_scan.mjs) performs lexical analysis using regular expressions and JSON parsing, ensuring that user-provided code is not executed during the precheck phase.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 10, 2026, 11:51 AM