Add Gremlin Q&A Part 2 blog post skeleton

Author: Gremlin

apps/blog/content/blog/gremlin-qa-your-questions-answered/index.mdx

@@ -0,0 +1,50 @@
+---
+title: "Gremlin Q&A: Your Questions Answered"
+slug: "gremlin-qa-your-questions-answered"
+date: "2026-07-24"
+authors:
+  - "Tyler Hogarth"
+metaTitle: "Gremlin Q&A: Your Questions Answered"
+metaDescription: "A follow-up Q&A on Gremlin, Prisma's autonomous engineering agent. Answers to the most common questions about multi-repo support, open sourcing, Mastra channels, sandboxes, logging, model providers, architecture, and Bun."
+heroImagePath: "/gremlin-qa-your-questions-answered/imgs/hero.png"
+heroImageAlt: "Gremlin Q&A illustration with question and answer icons, agent workflow diagrams, and technology stack symbols"
+tags: ["ai"]
+---
+
+In [Gremlin: turning open tasks into pull requests](/gremlin-turning-open-tasks-into-pull-requests), I wrote about how Gremlin uses [Mastra](https://mastra.ai/docs) for orchestration and [OpenCode](https://opencode.ai/) in a sandbox to turn Sentry issues, Linear tasks, and Slack instructions into reviewable pull requests. The response brought a wave of practical questions — this post answers the most common ones.
+
+## Does Gremlin have access to multiple repos with multiple stacks?
+
+*Explain that Gremlin can work across any stack because the coding agent operates in a sandbox, which isolates it from the host environment and lets it adapt to whatever repository and runtime it is dropped into.*
+
+## Is Gremlin going to be open sourced?
+
+*Describe how Gremlin started as an internal proof of concept that proved valuable enough to expand, but was designed for specific runtimes and integrations that make it less useful as a general-purpose open-source release. Note that 80%+ of the stack is already open source thanks to [Mastra](https://mastra.ai/docs) and [OpenCode](https://opencode.ai/).*
+
+## Are you using Mastra's Channels feature to handle Slack webhook events?
+
+*Confirm that Gremlin uses [Mastra's channels](https://mastra.ai/docs) for Slack event handling. Explain that relying on these primitives rather than building custom integrations is what makes using Mastra so effective — it handles the plumbing so Gremlin can focus on the orchestration logic that matters.*
+
+## What is used for the sandboxes?
+
+*Explain that [E2B](https://e2b.dev/) is the current sandbox provider, but this layer is flexible and could use almost any sandbox technology. That flexibility is precisely why sandboxes are so handy — the orchestration layer does not need to care about what runs inside them.*
+
+## What kind of logs does Gremlin persist?
+
+*Describe the two categories of logs Gremlin stores: agent-level logs (tool calls, agent delegation, decision traces) and runtime logs from the [Mastra](https://mastra.ai/docs) application itself. Explain why both are needed for debugging and observability.*
+
+## Which providers are you using for open models?
+
+*Explain that Gremlin uses any provider that meets compliance, security, and reliability standards rather than being tied to a single provider. Reference the point from the original post about OpenCode supporting many LLM providers and the model layer being exchangeable.*
+
+## Is Gremlin just one Mastra agent?
+
+*Clarify that Gremlin is a single Mastra codebase that orchestrates multiple other agents and sub-agents. The top-level agent handles routing and coordination, while delegation sends work to specialised sub-agents for implementation, validation, and other steps.*
+
+## Is Bun only used for the Mastra server runtime, or also in sandbox runtimes?
+
+*Explain that [Bun](https://bun.sh/) is used only for the Mastra agent runtime. The sandboxes must be able to run any kind of runtime based on the target codebase — they are not constrained to Bun or any single runtime.*
+
+---
+
+These were the questions that came up most often after the original post. If there are more, they may become a third installment. In the meantime, the best way to understand Gremlin is still to read the full architecture in [Gremlin: turning open tasks into pull requests](/gremlin-turning-open-tasks-into-pull-requests).