Alex Mashkovtsev f68f6da406 fix(agent-runner): derive MCP allowedTools from registered mcpServers ...

Claude Code 2.1.116+ treats SDK `allowedTools` as a hard whitelist:
servers whose namespace isnt listed are filtered out before the agent
ever sees them, regardless of `permissionMode: bypassPermissions` or
any `permissions.allow` in settings. The static TOOL_ALLOWLIST only
contained `mcp__nanoclaw__*`, so any MCP wired via add_mcp_server (or
directly in container.json) was silently dropped.

Derive `mcp__<sanitized-name>__*` entries at the SDK call site from
the already-aggregated `this.mcpServers` map, mirroring the SDKs own
sanitization rule (chars outside [A-Za-z0-9_-] become _).

Prior diagnosis by @jsboige in #2028 (withdrawn, not upstreamed).

2026-05-04 16:49:53 +08:00

..

scripts

docs: add v1→v2 action-items analysis + SDK signal probe tool

2026-04-20 01:00:04 +03:00

src

fix(agent-runner): derive MCP allowedTools from registered mcpServers

2026-05-04 16:49:53 +08:00

bun.lock

fix(container): bump Claude Code to 2.1.116 and Agent SDK to ^0.2.116

2026-04-20 23:28:35 -04:00

package.json

fix(container): bump Claude Code to 2.1.116 and Agent SDK to ^0.2.116

2026-04-20 23:28:35 -04:00

tsconfig.json

chore: delete v1 reference code

2026-04-18 14:23:47 +03:00