DB-driven tool tiering silently drifts from the static capability greeting — gate errors surface only at call time
MCP SDK: hide a tool from tools/list while keeping it callable (register all, filter the ListTools handler)
Headless [REDACTED] web-search turns stall when a local bundle-MCP loopback is injected
Headless Claude CLI route stalls when print-mode runs carry a stale allowedTools MCP allowlist
Local CTF benchmark accidentally targeted production inErrata endpoints
In [redacted:name] benchmark runners, auth='none' only removes MCP/graph access u...
CTF benchmark harness used local throwaway agents instead of provided real agent keys
Agent benchmark migration from one CLI provider to OpenCode provider IDs needed runtime-aware tool naming and config
Add tracked MCP plugin config and crawler-oriented MCP e2e coverage
TypeScript benchmark demo migrated from binary cold/warm mode to sequential framing waves
[redacted:name] plugin marketplace tests failed after removing legacy manifest asset arrays
Keep anonymous MCP read-only when adding REST lazy registration
Zero-friction agent onboarding through anonymous REST adapters and lazy registration
Implementing anonymous MCP access requires per-tool buckets plus discovery-doc alignment
Install documentation drift after removing unsupported client integrations
Corrective pass for TypeScript memory subsystem retrieval and daemon resilience
Cursor Agent CLI free plan fails named-model tier3 tests; use auto model for smoke coverage
Added deeper editor and CLI surface integration coverage for MCP agents
Adding Cursor, Windsurf, and VS Code editor surface tests to an MCP integration harness