@stashbase
Account based inEurope
About this account
- Account based in
- Europe
- Connected via
- Europe Android App
Account-level information from X, not a live location or the device used for a specific post.
Secure development from code to production—secrets, access, and guardrails for developers, services, and coding agents.
Web
Joined May 2024
- Tweets341
- Following134
- Followers39
- Likes592
An HTTPS_PROXY variable is not a sandbox. A coding agent can just ignore it.
If access is meant to be scoped, the network boundary has to hold even when the agent tries to bypass it.
That is the kind of sandboxing we have been working on. More soon.
One repo. Four agent sessions. Three different access profiles.
Claude Code and Codex running side by side in @herdrdev:
- development profile
- staging-access profile
- local-only review profile
The agent doesn’t decide what it can access. The profile does.
Works with both local and remote sessions, and with the agent tools you already use.
Here's the link to our open source CLI.
github.com/stashbase/cli
Coding agents need context to do useful work.
They don’t need the raw values in your .env files.
Stashbase gives agents the context and controlled access they need from development to production.
stashbase.dev
npm install is an agent action—not just a suggestion.
When a coding agent adds a dependency, it can change your dependency tree and run install scripts before anyone reviews the package.
Stashbase Agent Hooks block packages flagged with critical vulnerabilities before npm changes the project. The agent also can’t silently substitute a package or version without approval.
Works with Codex, Claude Code, and Cursor—and with Stashbase Agent Proxy profiles through the same policy.
The vulnerability shown in this demo is simulated.
stashbase.dev
The gap between what AI agents can do and how we manage their access and permissions is still massive.
An agent can read a codebase, open pull requests, modify files, call APIs, access cloud resources, and interact with internal systems.
But the way we give agents access often hasn’t changed much.
We hand them an API key, connect an MCP server, or let them inherit a developer’s permissions.
That might be fine for a local experiment. It becomes a different problem when agents run continuously, work across environments, or are allowed to make real changes.
Agent capabilities are moving quickly.
The security and permission models around them are still catching up.
Building for developers and AI agents doesn’t mean the software has to feel utilitarian.
The people managing these workflows still use the product every day.
For example, in Stashbase you can customize the themes, fonts, secret masking, and date/time formats.
Small details, but they matter.
stashbase.dev
Stashbase is live on Uneed today.
We’re building a secrets control plane for developers and AI agents, with MCP support that keeps credential values outside the agent context.
uneed.best/tool/stashbase
New: MCP server configuration in the Stashbase Agent Proxy.
Configure MCP servers alongside the rest of your agent policy.
• Bind each server to a shared secret or personal credential.
• Restrict how credentials may be used with HTTP host, method, and path rules.
• Allow or deny individual MCP tools.
Docs: docs.stashbase.dev/agents/mc…
What if a secret is already in your repository?
Stashbase continuously monitors @github repositories and surfaces exposed secrets with the finding details and remediation options.
Dismiss the finding or mark it resolved after fixing the issue.
stashbase.dev
A hard-coded Stripe API key should never make it into a commit.
With Stashbase, install a Git pre-commit hook with one CLI command. If a secret is detected, the commit is blocked before it reaches Git history.
Git hooks are one part of our secret protection. We also continuously monitor repositories for secrets that still get through.
stashbase.dev
Coding agents need credentials—but they shouldn’t need access to your .env files or unrestricted bearer tokens.
Stashbase gives agents controlled access to the APIs they need while keeping secret values outside their context.
Scoped credentials. Host, method, and path-level policies. Full request visibility.
stashbase.dev
Environment chat in Stashbase.
Summarize an environment, see who uses the secrets most, and remove all Google and Vercel-related secrets—all through natural language, without exposing the actual secret values.
Built for developers, services, and AI agents.
stashbase.dev
Zero bounce CLI 😎
Quick Superlogical demo to end the week. This demo is just of the basic functionality, but I want to highlight just how FAST it is. There's a WHOLE lot more functionality already present and coming but I wanted to keep the demo short. But seriously, check out that speed.
I'm showing the macOS app here and I get asked a lot so let me just say up front: its not macOS only. We're still figuring out exactly what platforms will be stable enough for the initial public release, but we're working on others. In particular, the web interface is very functional but I'll demo that another time (or someone else will).
This is also all showing off the architecture I previously talked about for terminal multiplexing which is significantly different from other mainstream multiplexers: this is all a custom binary protocol where the server is maintaining N replica distributed terminal state machines. It works!
Tons of hard work going into this by the team @almonk @dizzyup @pearkes. Hope we can bring you something you can use soon. ❤️ Happy to answer questions in the replies.
Quick Stashbase update: we’ve added personal credentials for coding agents.
You can connect your own Linear or API credentials and let Agents use them without exposing raw keys to the agent or putting them in the codebase.
Personal credentials stay separate from shared project secrets.
stashbase.dev/agents
stashbase.dev retweeted
A coding agent can use my Linear account without seeing my API key.
In this demo, I add a personal Linear credential to Stashbase and use it from Codex. The key stays outside the codebase and agent context, and is injected only when the request is made.
This is what we’re building at Stashbase.
stashbase.dev