Rul1an/assay
by Various
Policy-as-code for MCP agents: deny risky tool calls before they run, prove what ran with verifiable evidence, and enforce egress in the kernel (eBPF/LSM, Linux). Deterministic, of
MCP
Rul1an/assay
Added 15 June 2026
Overview
Rul1an/assay is a Rust-based policy engine that enforces tool call policies for MCP agents at the kernel level using eBPF or LSM on Linux. It blocks risky tool calls before execution, provides verifiable evidence of agent actions, and enforces egress controls. The system is deterministic and offline-first with bounded claims.
Best for
Best for
Developers building secure MCP agents with deterministic policy enforcement
Use cases
- Block unauthorized tool calls in MCP agents
- Generate verifiable audit logs of agent actions
- Enforce network egress policies at the kernel level
Notes
Rul1an/assay is a Rust-based policy engine that enforces tool call policies for MCP agents at the kernel level using eBPF or LSM on Linux. It blocks risky tool calls before execution, provides verifiable evidence of agent actions, and enforces egress controls. The system is deterministic and offline-first with bounded claims.
7 stars on GitHub. Last updated 2026-06-15. Licensed MIT.
Use cases
- Block unauthorized tool calls in MCP agents
- Generate verifiable audit logs of agent actions
- Enforce network egress policies at the kernel level
Pros
- Deterministic enforcement without runtime dependencies
- Kernel-level egress control via eBPF or LSM
- Bounded claims provide verifiable evidence of agent behavior
Cons
- Linux-only due to eBPF or LSM dependency
- Very early stage with only 7 GitHub stars
- Narrow scope limited to MCP agent tool call enforcement
Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.
Pros
- Deterministic enforcement without runtime dependencies
- Kernel-level egress control via eBPF or LSM
- Bounded claims provide verifiable evidence of agent behavior
Cons
- Linux-only due to eBPF or LSM dependency
- Very early stage with only 7 GitHub stars
- Narrow scope limited to MCP agent tool call enforcement
Pairs with
Other entries in the index that connect to this one. Click through to see the chain.
Claude Computer Use
Anthropic
Claude takes the mouse and keyboard. A vision-based agent that controls a real desktop, not just a browser.
Claude Code
Anthropic
Anthropic's terminal-native coding agent. Reads your repo, edits files, runs tests, ships PRs.
Cline
Cline
Open-source autonomous coding agent that lives inside VS Code. BYO model key, watch it work.