If your agent needs to know whether upgrading from Anthropic SDK 0.25 to 0.30 will break its code, this saves you from burning tokens on GitHub scraping. It tracks 18 SDKs that AI tooling depends on (Anthropic, OpenAI, LangChain, MCP, Mistral, Vercel AI, and others) and pre-classifies breaking changes so you can call check_version with your current version or pull get_recent_breaking to see what changed across the ecosystem in the last 30 days. The four tools (list_tracked, get_package, get_recent_breaking, check_version) replace what would otherwise be 5 to 15 seconds of release feed parsing with a sub-100ms lookup.
SDK breaking-change tracker as a stdio MCP server.
Probe P-004 by weiseer.
Tracks latest version + breaking changes for SDKs AI agents depend on (Anthropic, OpenAI, Google GenAI, MCP SDK, LangChain, LlamaIndex, Mistral, Cohere, Vercel AI, Next.js, React, HuggingFace, Replicate, Ollama).
Your agent can:
list_tracked — packages we track + latest version + 30d breaking-change countget_package — full record with recent breaking events + GitHub release URLget_recent_breaking — breaking changes across all packages, filterablecheck_version — "I'm on v X — am I behind, and what breaks if I upgrade?"| Agent DIY | api-changelog | |
|---|---|---|
| Releases to fetch | 18 GitHub repo release feeds | 1 MCP call |
| Token cost | $0.06-0.20 | $0 free / $0.00005 paid |
| Latency | 5-15 seconds | <100ms |
| Breaking-change classification | LLM-classified per call | Pre-classified once |
npm install -g @weiseer/api-changelog-mcp
{
"mcpServers": {
"api-changelog": {
"command": "npx",
"args": ["-y", "@weiseer/api-changelog-mcp"]
}
}
}
Apache-2.0
API_CHANGELOG_URLOverride remote snapshot URL
API_CHANGELOG_LOCAL_ONLYSkip remote fetch
io.github.ericm1018/skillfm-llm-cost-optimizer-openai-anthropic-usage
io.github.mikerawsonnz/llm-orchestration-agent
io.github.mikerawsonnz/authenticated-llm-agent
labforgedev/copilot-memory-mcp
csoai-org/agent-prompt-injection-firewall-mcp
io.github.mikerawsonnz/authenticated-multi-llm-agent