fix: add request body size limit to proxy - #7
Open
hobostay wants to merge 1 commit into
Open
Conversation
The proxy had a 1KB size limit on the control endpoint (/_proxy/mode) but no limit on model API request bodies. A client could send an arbitrarily large request, causing memory exhaustion in the proxy process. Add a 50MB limit with a 413 response on overflow. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
g-roliveira
added a commit
to g-roliveira/deepclaude
that referenced
this pull request
Aug 6, 2026
…parity Summary of fixes and features applied from aattaran/deepclaude issues and PRs, plus original improvements: Model coverage (Issue aattaran#39): - Add claude-fable-5, claude-opus-5, claude-sonnet-5 to MODEL_REMAP - Tier-based _default fallback (fable/opus/sonnet/haiku) so new Claude models degrade predictably instead of silently routing to the wrong backend - Warning log when model is forwarded unmapped Proxy resilience (PR aattaran#18): - proxyRes error handler — upstream TCP reset mid-response no longer crashes the entire proxy process - proxyReq error after headers-sent: destroy response instead of injecting JSON into the SSE stream (which corrupted the parser) Thinking-block continuity (PR aattaran#24): - Drop top-level thinking/context_management on non-Anthropic routes instead of stripping all thinking blocks from history - Fixes DeepSeek 400: "content[].thinking must be passed back" Proxy in normal mode (PR aattaran#9): - Default launch now starts the proxy; cost tracking and live /switch work in all sessions, not just --remote - Use ANTHROPIC_API_KEY instead of ANTHROPIC_AUTH_TOKEN so subscription OAuth tokens are handled correctly Body size limit (PR aattaran#7): - 50 MB cap on /v1/messages request body, returns 413 Forward unknown args to claude (Issue aattaran#25): - Bash: -- separator support for explicit passthrough - PowerShell: ValueFromRemainingArguments captures unknown flags PowerShell parity (PR aattaran#5, aattaran#8): - --switch/-s parameter with backend name normalization - try/finally ensures proxy cleanup on Ctrl+C/crash Model [1m] suffix: - All model names include [1m] token window hint so Claude Code doesn't assume 200k context and auto-compact prematurely Project documentation: - CLAUDE.md with architecture overview, proxy routing, SSE normalization, model remapping, and development conventions
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
1 participant
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.
Summary
proxy/model-proxy.jsProblem
The
/_proxy/modecontrol endpoint already had a 1KB body size limit, but the main model API request path (/v1/messages) collected request body chunks into memory without any bound. A client (or a bug in the calling code) could send an arbitrarily large request body, causing the proxy Node.js process to consume unbounded memory and potentially crash.Test plan
/v1/messagesrequest through the proxy🤖 Generated with Claude Code