Security context
What an agent needs to avoid regressing past fixes and find the next vuln in this repo.
kayba-ai/agentic-context-engine
main @ 96f7c9c
8
Fixes
0
CVEs
HIGH
Peak severity
Highlights
Path Traversal: 4 prior fixes. Scrutinize any change in this area.
ace_next/integrations/mcp/handlers.py: most-fixed (4 issues). Treat as high-risk during review.
7 high-severity fixes in this history; regressions here are high-impact.
Recurring patterns
The bug types that recur here, drawn from past fixes, not open vulnerabilities.
Sandbox Escape: Attackers executing sandboxed Python code can access underlying system operations if bare builtins or standard library modules are not thoroughly restricted. Multiple iterations of bypasses highlight the difficulty of fully isolating the execution namespace.
Path Traversal: Improperly validated skillbook paths allow malicious agents to read or write arbitrary files on the host filesystem. This is particularly critical because tool executions automatically process directory structures which can be manipulated via symlinks or traversal sequences.