Small open-weight models as coding sub-agents: reading the Mellum2.1 release
JetBrains released Mellum2.1, a 12B open-weight model built for coding sub-agents. What it claims, what to verify, and where small models fit.
Blog
The SpanDock blog
News, how-tos and researched articles on AI coding tools, agents, MCP, cost and governance, from the SpanDock team.
Get told when a server needs you: notifications on your account dashboard
Connect a SpanDock server's notifications to your spandock.com dashboard, choose which kinds to see, and send a test. In every plan.
3 min read
Prompt caching: the AI cost control most teams don't measure
Cached input costs a tenth of the normal price on major APIs. How prompt caching works, why coding agents depend on it, and how to see whether yours hits.
4 min read
Observing AI coding tools with OpenTelemetry: what you can measure today
Claude Code, Codex and Copilot now emit OpenTelemetry. What the GenAI conventions cover, what the tools actually send, and how to start measuring.
MCP servers keep trusting the wrong URLs: what recent advisories show
Advisories from Splunk, AWS, Google and the Rust MCP SDK share one pattern: a tool builds a request or command from input it shouldn't trust.
Write guardrails that keep secrets and personal data out of prompts
Add two or three SpanDock guardrails for secrets, personal data and your own patterns, enforce them on every computer, and check that they work.
When model output becomes code: three agent framework advisories from October
Flowise, Langflow and PraisonAI fixed flaws where prompt injection led to code execution. The shared mistake, and what to check in your own agents.
4 min read
Set up the MCP gateway and choose your first connectors
Give every computer's AI tools one governed MCP server: add two or three connectors read-only, set permissions and approvals, and check it works.
5 min read
The stateless MCP specification: what changes for teams running MCP servers
MCP 2026-07-28 drops sessions, adds routing headers, hardens authorisation and deprecates features. What platform and security teams need to plan for.
5 min read
Install a server and pair your team's computers: server, satellite and client
The three SpanDock roles explained, then a first install: choose the mode, activate the server, configure a tool and pair your team's computers.
5 min read
Browse by tag
- #Security (64)
- #AI coding tools (46)
- #Agents (43)
- #LLM costs (24)
- #MCP (20)
- #Evaluation (14)
- #Enterprise (13)
- #Observability (13)
- #Platform engineering (12)
- #Compliance (11)
- #DevOps (10)
- #Data privacy (9)
- #Model routing (7)
- #Open-weight models (7)
- #Research (7)
- #Tutorials (7)
- #Developer productivity (5)
- #Self-hosting (5)
- #Code review (4)
- #Guardrails (4)
- #Networking (4)
- #OpenTelemetry (4)
- #Budgets (3)
- #Prompt engineering (3)
- #Skills (3)
- #MCP gateway (2)
- #Release notes (2)
- #Testing (2)