Repository navigation
The Board's MCP tools wait for a busy board and name an unreachable one - #10
Merged
Merged
Conversation
work-board mcp waited 10 seconds per tool call, less than the 25 seconds the Board may spend on its Fleet view, so a busy Fleet read as a stopped board. It now waits 30 seconds, reports a slow board as busy and an unreachable board as not reachable. The README warns against npx -y, whose stale cached install stops the server from starting.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Why?
Claude sessions read the Work Board through the
work-boardMCP server. In Marvin's sessions the server failed with "Connection closed" and its tools stayed gone for the whole session. It looked as if the server needed the Board at startup.That is not the cause. The server starts, lists its tools and answers each call without a Board; a story now proves it. The session is registered as
npx -y @shivaedev/work-board mcp …. npm keeps one cached install for that command. When a release pins another Effect version, npm refuses to update the cached copy (ERESOLVE … Conflicting peer dependency: effect@4.0.7), and the process exits before Claude Code connects. Every session since the Board left Platform's workspace hit this, whether the Board ran or not.A second, smaller fault was in the server itself. Each tool call waited 10 seconds. The Board may spend up to 25 seconds on its Fleet view when Fleet is busy, so a busy Board was reported as "No board answers".
How?
Launch. The README now names
pnpm dlx @shivaedev/work-board mcp <dir>for clients without a global install, and says whynpx -yfails. pnpm resolves into a fresh folder per version, so a new Effect pin does not break it.Wait. A tool call now waits 30 seconds: the Board's 25-second Fleet view budget (from #9) plus 5 seconds for the Board's own index and document reads. The MCP client takes the budget from the same constant as the Fleet view, so the two cannot drift apart.
Messages. A Board that does not answer in time is reported as busy ("did not answer within 30 seconds. It may be busy; try again."). A Board that cannot be reached is reported as "not reachable", with the command that starts it.
Callouts
npx -y @shivaedev/work-board, and npm's stale cached copy (0.30.1) stays until someone deletes that npx cache folder or changes the command topnpm dlx.