Pinned Loading
-
leash
leash PublicA safety-first framework for AI agents that act in a real web browser: a from-scratch browser-driving agent core with an inline policy guardrail and a tamper-evident audit log.
Python
-
agentevals
agentevals PublicForked from langchain-ai/agentevals
tool_permission evaluator contributed to LangChain's agentevals (PR #108): a deterministic, LLM-free least-privilege check that scores whether an agent called only authorized tools via an allowlist…
Python
-
deepeval
deepeval PublicForked from confident-ai/deepeval
ToolPermissionMetric contributed to DeepEval (PR #2826): a deterministic, LLM-free least-privilege check that scores whether an agent called only authorized tools via an allowlist and denylist — a …
Python
-
covenant
covenant PublicA CI gate for AI-agent behavior that grades actions, not answers: Covenant replays an agent's full trajectory against declarative least-privilege and safety policies — catching permission escalatio…
Python
-
fastmcp
fastmcp PublicForked from PrefectHQ/fastmcp
OAuth discovery hardening for FastMCP (fixes #4390): path-scopes authorization-server metadata to the issuer subpath per RFC 8414, so multiple MCP OAuth servers sharing one host under different bas…
Python
If the problem persists, check the GitHub status page or contact support.