Conversation
The input-shape guard (#455) makes every read that treats an argument as a list go through StringList, so a string is never dropped. It cannot see the reverse: a handler that declares an argument as an array and reads it with in.String(), which returns "" for an array and drops it without an error. These tests find every array-typed argument from the live tool schemas on both servers, with writes enabled, and require each to have a valid call in a table. Each call sends a genuine JSON array to a recording server, and every value must reach the request. A new array argument fails until it has an entry; a stale entry fails too. Ten arguments today -- four on Gateway (metrics measures and dimensions, request connection_ids, connection rules) and six on Outpost -- and all ten deliver correctly, so this is a guard, not a fix. The expected strings are the encodings captured from the recording server, not assumed. Verified against the failure: reading connection_ids with in.String() fails the Gateway test -- and shows the retry going out as {}, i.e. retry every connection instead of the two chosen, a silent widening. Reading include with in.String() fails the Outpost test. Removing a table entry fails with a message naming the argument. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_012XtSQ2kfpcRXXqweskgXkH
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
The follow-up #455 names in its Not covered section.
#455 makes every read that treats an MCP argument as a list go through
StringList, so a bare or comma-separated string is never dropped. It cannot see the reverse: a handler that declares an argument as an array and reads it within.String(), which returns""for an array and drops it without an error.What this adds
A test on each server that:
A new array argument fails until it has an entry; a stale entry fails too. The expected strings are the encodings captured from the recording server, not assumed.
Today
Ten array arguments, and all ten deliver correctly — a guard, not a fix.
metrics_read.measures,metrics_read.dimensions,request_write.connection_ids,connections_write.rulesattempts_read.include,config_write.unset,destinations_read.topics,destinations_write.topics,metrics_read.measures,metrics_read.dimensionsProven against the failure
connection_idswithin.String()fails the Gateway test — and the recorded request shows why it matters: the retry goes out as{}, meaning retry every connection, instead of the two chosen. A silent widening of a write.includewithin.String()fails the Outpost test.CI's unit-test steps (
go test -short ./pkg/...,go vet ./...) pass locally by exit code.Intended for v3.0.3.
🤖 Generated with Claude Code
https://claude.ai/code/session_012XtSQ2kfpcRXXqweskgXkH