Changelog
Release history for the @socialneuron/mcp-server package and platform updates. For full release notes and downloads, visit the GitHub Releases page.
v1.6.1 -- March 22, 2026
Security hardening and stability improvements.
- Added request body size limits for safer input handling
- Improved error sanitization to prevent information leaks in API responses
- Removed sensitive identifiers from the authentication chain
- Added generation rate limiting (20 requests per minute)
- Resolved 0 npm audit vulnerabilities
What this means for you
Your API integrations are now more secure and more reliable. Generation requests are rate-limited to prevent accidental overuse, and error responses no longer expose internal details. If you use the MCP server or REST API, you benefit from cleaner error messages and safer input handling with no changes needed on your side.
v1.6.0 -- March 21, 2026
REST API layer and developer experience improvements.
- REST API: New HTTP endpoints alongside MCP, with OpenAPI 3.1 specification
- Universal tool proxy: Call any MCP tool via
POST /v1/tools/:namewith standard HTTP - 15 convenience endpoints for common operations (publish, analytics, credits)
- 18 usage examples covering CLI, MCP, and REST workflows
- SDK package scaffolding for typed client libraries
- CLI improvements: Better error messages and help text
What this means for you
You can now integrate Social Neuron into any application using standard HTTP requests -- no MCP client required. Build custom dashboards, connect to Zapier or Make, or automate your workflow from any language. The 18 ready-to-use examples get you started in minutes.
v1.3.0 -- March 13, 2026
Agent-native CLI v2 with tool discovery and content plan management.
- Tool discovery:
sn toolslists all MCP tools, filterable by--moduleand--scope - Introspection:
sn infoshows version, tool count, auth status, and credit balance - Content plan CLI:
sn plan list|view|approvefor managing content plans from the terminal - Presets:
sn preset list|show|save|deletewith 6 built-in platform presets - Interactive REPL:
socialneuron-mcp replwith tab completion and persistent auth - Progressive disclosure: New
search_toolstool reduces token usage from 55K to ~500 tokens - Unified JSON envelope: All CLI output includes
schema_version,ok,command, and typed error fields - Fixed flag validation to run before auth -- missing flags now show the correct error
- Fixed duplicate platform normalization in the publish handler
What this means for you
Manage your entire content pipeline from the terminal. Plan your week, approve content ideas, and publish posts without opening a browser. The new presets let you save your favorite platform configurations and reuse them instantly.
v1.2.1 -- March 11, 2026
Documentation and security cleanup.
- Corrected pricing information in README to match actual tiers
- Rewrote scopes section with tool-to-scope mapping
- Added telemetry opt-out instructions (
DO_NOT_TRACK=1) - Removed unused PKCE from device code flow
- Improved logout messaging with link to server-side revocation
What this means for you
The documentation now accurately reflects current pricing and scopes. If you prefer not to send usage telemetry, you can disable it with a single environment variable. Logging out now clearly confirms that your session has been revoked server-side.
v1.2.0 -- March 10, 2026
Gateway security and developer experience.
- Zero-config for npm users: Public anon key embedded -- no
SUPABASE_ANON_KEYneeded - Gateway token system: HMAC-SHA256 tokens for secure communication between services
- Standardized error responses: All error paths return structured JSON with
error,message, and actionable hints - Tightened API key database policies from public to authenticated role
- Daily credit reset job for API key usage tracking
- Updated
.npmignoreto exclude test files and sensitive patterns
What this means for you
Getting started is now simpler -- install the npm package and authenticate. No manual configuration of environment variables required. Error messages now tell you exactly what went wrong and how to fix it, instead of returning cryptic codes.
v1.1.0 -- February 27, 2026
New CLI commands for automation and monitoring.
- Added
sn autopilotfor viewing automation configuration - Added
sn usagefor account usage statistics - Added
sn loopfor closed-loop performance summaries - Added
sn creditsfor credit balance checks - Refactored CLI architecture for better maintainability
What this means for you
Check your credit balance, review account usage, and monitor your content performance loop without leaving the terminal. The sn loop command shows you how your published content is feeding back into better ideation -- the learning loop in action.
v1.0.0 -- February 17, 2026
Initial public release.
- 60+ MCP tools across 20 modules covering the full content lifecycle
- OAuth 2.1 authentication with browser flow, device code flow, and API key paste
- Scope-based access control: 6 scopes from read-only to full access
- Cloud mode via gateway proxy -- your service key never leaves the server
- Secure credential storage using system keychain (macOS, Linux) with file fallback
- Auto-configuration for Claude Desktop and Claude Code
- CLI tools:
sn publish,sn preflight,sn quality-check,sn status,sn posts,sn refresh-analytics - Credit tracking with balance checks and budget monitoring
- Per-user rate limiting (100 requests per minute default)
- Agent safety features: loop detection, session caps, and kill switch
What this means for you
This is the foundation. 60+ tools covering every step of the content lifecycle -- from generating ideas and creating content to publishing, tracking analytics, and feeding performance data back into your next batch. Connect Claude Desktop, Cursor, or any MCP-compatible client and control Social Neuron entirely through AI conversation.
Default scope change in v1.0.0: API keys without explicit scopes default to mcp:read (read-only). If you need write access, request mcp:full when generating your API key.
Ready to try this? Sign up free and get 100 credits to start.