Specialize streamable HTTP headers by request type

[daleselaji-dev]

Problem

StreamableHTTPTransport._prepare_headers() used the POST request shape for every outbound request. The standalone SSE GET, resumption GET, reconnection GET, and bodyless DELETE therefore advertised or sent headers that did not match their request semantics.

Root Cause

The transport shared Accept: application/json, text/event-stream and Content-Type: application/json across POST, SSE GET, and DELETE call sites. SSE responses are consumed as text/event-stream, while bodyless GET/DELETE requests have no JSON body.

Solution

Parameterize header preparation while preserving the POST defaults. SSE GET paths now advertise only text/event-stream and omit Content-Type; DELETE omits Content-Type while retaining the existing session/protocol headers.

Changes

• Add optional accept and content_type parameters to _prepare_headers().
• Apply SSE-specific headers to the standalone, resumption, and reconnection GET paths.
• Remove Content-Type from session DELETE requests.
• Add deterministic regression coverage for POST, SSE GET, and DELETE header shapes.

Testing

• Baseline uv run --frozen pytest tests/client/test_streamable_http.py -q — 36 passed.
• Final focused suite: same command — 37 passed.
• uv run --frozen ruff format --check on changed files — passed.
• uv run --frozen ruff check on changed files — passed.
• uv run --frozen pyright src/mcp/client/streamable_http.py — 0 errors.
• Targeted coverage command was attempted but this Windows environment produced no coverage data; no coverage pass is claimed.
• Repository pre-commit executable was not available in the environment; not verified.
• git diff --check — passed.

Compatibility/Risk

POST requests retain the existing Accept and Content-Type defaults. Only transport-internal SSE GET and DELETE headers change; session and negotiated protocol headers remain included.

Notes for Reviewer

The new test locks the wire-level contract without requiring a live server. Per-message metadata headers remain layered after the transport defaults, as before.

Linked Issue

Fixes #3503

[github-actions]

This PR has been closed automatically. This repo only keeps pull requests open when they come from a maintainer, or from a contributor a maintainer has assigned to the linked issue, and you aren't currently assigned to #3503.

If a maintainer assigns you to #3503, this PR reopens on its own and there's nothing more you need to do here. Assignment is a maintainer call based on capacity; comments that only ask to be assigned don't factor in. What does help is engaging on the issue itself by confirming the repro, explaining why it matters for your use case, or describing the approach you'd take.

You're welcome to keep pushing commits here (just avoid force-pushing, since GitHub can't reopen a rewritten branch), but that on its own won't get the PR reviewed or the issue assigned, and realistically most auto-closed PRs stay closed. There's no need to open a new PR either way.

CONTRIBUTING.md has the full reasoning, but in short:

• We're a small team with very little capacity to review community PRs right now.
• Many recent PRs are AI-generated with little human review, and reviewing one carefully still costs a maintainer as much time as it ever did. A well-described issue is usually more useful to us than the code.

Maintainers: reopen, remove missing-issue-link, or add bypass-issue-check to override.