vitron-ai/alethia
Agent-native E2E runtime with verifiable safety. 16 MCP tools including alethia_propose_tests (agent generates tests from a URL), alethia_assert_safety (proves destructive actions are blocked), and the expect block: NLP primitive unique to Alethia. Zero-IPC; 2-5x faster than Playwright MCP per flow; signed evidence packs. Works with Claude Code, Cursor, Cline.
95
94%
Does it follow best practices?
Impact
98%
2.80xAverage score across 5 eval scenarios
Low
Low-risk findings worth noting
W011: Third-party content exposure detected (indirect prompt injection risk).
What this means
The skill exposes the agent to untrusted, user-generated content from public third-party sources, creating a risk of indirect prompt injection. This includes browsing arbitrary URLs, reading social media posts or forum comments, and analyzing content from unknown websites.
Why it was flagged
The SKILL.md explicitly instructs the agent to call alethia_propose_tests / alethia_assert_safety with an arbitrary target URL and to consume returned pageContext (title, headings, buttons, inputs), PlanRun step results, and screenshots—i.e., it fetches and interprets untrusted third‑party web pages as part of its workflow, which can materially influence subsequent actions.
Where we found it
arbitrary target URL
url · 7 sites
The SKILL.md explicitly instructs the agent to call alethia_propose_tests and alethia_assert_safety with an arbitrary user-provided target URL, then consume returned pageContext (title, headings, buttons, inputs), PlanRun step results, and screenshots — fetching and interpreting untrusted third-party web content that can influence subsequent agent actions.
- Audited
- Security analysis