Commit Graph
4 Commits
Author SHA1 Message Date
thomas f2bee854a7 Complete MCP 2026 client support (#35725)
## What changed

- Drive multi-round `tools/call` and `resources/read` requests through
  `input_required` responses, preserving opaque request state and elicitation
  metadata across JSON, SSE, and stdio transports.
- Decode modern discovery and elicitation result shapes while retaining legacy
  initialization, elicitation defaults, and discovery fallback behavior.
- Apply the 8 MiB modern-protocol response limit to JSON, SSE events, and stdio
  messages, and reject executor stream recovery when output sequence gaps would
  corrupt the protocol stream.

## Testing

Added integration coverage for discovery, multi-round requests, legacy
fallbacks, message limits, stdio launchers, SSE responses, session recovery,
and executor output recovery.

GitOrigin-RevId: d26a08883a2825e8ce2408ac498114b33859225e
2026-07-28 06:06:58 +00:00
Adam Perry @ OpenAI 44954d1b4b Serialize concurrent MCP stdin writes (#33180)
## What changed

- Guard executor-backed MCP stdio sends with a single-permit semaphore so a
  second JSON-RPC message cannot start writing while the first write is pending.
- Add a regression test that drives two concurrent sends and verifies the
  process receives each newline-delimited message in order without overlap.

GitOrigin-RevId: 30c2e51827ecf57b5eb23d9ad1e876734d6e784e
2026-07-14 22:16:52 +00:00
jif 6ced1ac5eb Bound remote MCP stdio lines (#31805)
## Why

Remote MCP stdio buffers executor-provided stdout until a JSON-RPC line
completes and stderr until a diagnostic line completes. A malicious or
broken executor can keep sending chunks without a newline, making
orchestrator memory grow without bound.

Executor event boundaries are arbitrary, so a valid message before an
oversized line must not be lost merely because both arrived in one
event.

## What changed

- Limit one stdout JSON-RPC line to 8 MiB.
- Limit one stderr diagnostic line to 1 MiB.
- Check each logical line before copying its bytes into the retained
buffer.
- Preserve complete stdout messages that precede an oversized line and
deliver them before closing the stream.
- Close the MCP transport on overflow; existing transport cleanup
terminates the executor-managed process.
- Preserve multiple lines per chunk, CRLF framing, and final
unterminated messages at EOF.
- Box the pending stdio transport so the added framing state does not
inflate every `ClientState` value.
- Add regression coverage for cross-chunk overflow, bounded lines in a
larger chunk, and a valid prefix before an oversized line.

## Scope

This bounds pre-newline buffering for executor-backed MCP stdio only. It
does not change local stdio framing or address memory amplification
while parsing a complete JSON message.
2026-07-09 23:15:00 +01:00
Charlie Marsh 7da4af622f Optimize unbounded byte scans with memchr (#26265)
## Summary

This PR adds `memchr` for some low-hanging performance improvements
(namely, in MCP stdio, Ollama streaming, and full message-history
newline counts).

Codex produced the following release benchmarks:

| Operation | Before | After | Speedup |
| --- | ---: | ---: | ---: |
| MCP 1 MiB chunked line | 2.172 s | 3.984 ms | 545x |
| Ollama 1 MiB chunked line | 1.673 s | 2.790 ms | 600x |
| Count newlines in 10 MiB history | 132.83 ms | 20.05 ms | 6.6x |

With a "real" MCP setup (`ExecutorStdioServerLauncher` started a Python
MCP server, completed `initialize`, requested `tools/list`, and
deserialized a 1 MiB tool description over newline-delimited stdio),
it's about 16x faster end-to-end:

| Branch | 50 calls | Per call |
| --- | ---: | ---: |
| `main` | 862.53 ms | 17.25 ms |
| this branch | 53.89 ms | 1.08 ms |

`memchr` is already in our dependency tree and extremely widely used for
this kind of optimized scanning.
2026-06-04 09:53:08 -04:00