Code & AI agent sandboxes: read the guide

Drop

Open source toolSandbox

Linux sandboxing that doesn't get in your way

Homepage Docs Repository

Scanned 25 Sep 2026 · Schema v5 · features v18

Licence: Apache-2.0
Self-hosted; you pay for the infrastructure.
Free tier: Free to run
Isolation: Container
shares the host kernel
Regions: —
Not disclosed on the public surface.
Capabilities: 3 / 55
2 partial · 24 fail · 6 not assessed
Agent readiness: 5.5/10 · 1/6
read · drive, pre-login surface

Advertised for

AI coding agents (Claude Code run with dangerously skip permissions) · Untrusted third party packages (PyPI/npm installs, supply chain containment) · Interactive developer shells

Built for

Developers, then technical

· hero content is an asciinema terminal recording; onboarding is `curl` a binary + `drop init` / `drop run`
· vocabulary is namespaces, capabilities, pasta, gVisor, AppArmor/SELinux policies
· install requires hand-writing an AppArmor profile (Ubuntu 24+) or compiling an SELinux module (Fedora)

01 Runtime Each fact carries its source; dots show confidence. Open a row for the evidence.

Isolation & lifecycle

Isolation primitive

Containershares the host kernel

Persistence

Per-environment persistent state on the host disk. Each env has its own writable home, /var and an /etc overlay upper dir under ~/.local/share/drop/envs/<id>/, plus /tmp under /tmp/drop-$USER/. All of it survives across `drop run` invocations until `drop rm`. The project dir where `drop init` ran is bind-mounted read-write (its .git read-only), so project work lands directly in the real tree.

Snapshots / forks

None. There is no snapshot, clone or rollback of an environment.

Resources & limits

Storage

Own root filesystem via a mount namespace. /usr,/bin,/sbin,/lib* are read-only from the host, /etc is a read-only overlay, and home, /var and /tmp are per-env writable. Fresh /proc,/sys,/run,/dev with only null/zero/full/random/urandom devices. The TOML `mounts` list exposes extra host paths (read-only by default, `:rw` opt-in), and the per-run `--mount` flag adds more.

Limits & quotas

No resource limits (no CPU/memory/pids caps are documented or configurable. A cgroup namespace is used for isolation, not for limits). Documented limitations: terminal only (no GUI by default), no sound/extra devices, setuid programs don't run, and programs that need user namespaces themselves (Podman, Snap-installed programs) are unsupported. Linux amd64/arm64 only.

GPU

Access

Deploy interface

cli

SDK languages

None

Network & secrets

Egress controls

Coarse, local-focused. `net.mode = off` cuts all network. The default `isolated` mode gives a separate netns via pasta, full internet access, and DEFAULT-DENY to host localhost services, with per-port allowlists (tcp_host_ports / udp_host_ports). Inbound is opt-in port publishing (tcp/udp_published_ports, incl. an `auto` mode). There is NO remote domain/IP allowlist and no egress proxy.

Secrets

Hide-by-default, no management. The real home (incl. ~/.ssh) is replaced by a per-env home. Only allowlisted env vars cross (`exposed_vars`, globs allowed), plus `set_vars`. Selected dotfiles are mounted read-only, and `blocked_paths` masks specific paths. The docs warn the user to review base.toml so no secret-bearing files or vars are exposed. There is no keystore integration or credential proxy: any secret an agent needs must be placed inside the sandbox.

Bring your own cloud

Placement & use

Jurisdiction

Advertised workloads

AI coding agents (Claude Code run with dangerously skip permissions)Untrusted third party packages (PyPI/npm installs, supply chain containment)Interactive developer shells

02 Capabilities 3 pass · 2 partial · 24 fail · 55 of 61 assessed

Supported — Documented on the public surface and verified.
Partially supported — Exists with caveats: gated, thin, or enterprise-only.
Not supported — Checked and not offered.
Not assessed — Scanned before this capability was tracked. Not a lack.

Compute & storage 1/5

Persistent diskStorage that survives restart/stop (not ephemeral-only).

Supported

Object storageManaged S3-compatible / blob object store the platform PROVIDES to a user's workloads (buckets, presigned URLs). DISTINCT from persistent-disk (block/volume attached to an instance) and from a self-host dependency (source.self_host.deps[]=["object-storage"] is an infra dep to RUN an OSS tool — the opposite direction). SCOPE varies and matters: native first-class (Railway Buckets, Cloudflare R2) vs a sibling SKU / extension (Fly Tigris, Bunny Storage, DigitalOcean Spaces) — record which in the evidence quote / note. Admitted v15.

Not applicable

Snapshots / forkingSnapshot or copy-on-write fork of a running instance (filesystem-only counts). Whether the snapshot also restores memory + running processes is recorded separately by memory-snapshot (v19).

Not supported

Live-state (memory) snapshotsA snapshot / checkpoint / pause captures a running instance's MEMORY and PROCESS state and restores it, so running processes continue where they left off instead of a freshly booted instance with the same files. Counts whether user-triggered (snapshot, checkpoint, pause-to-storage) or used to restore pre-initialised state (memory snapshots for fast cold starts). Distinct from snapshots-forks (any snapshot or fork, including filesystem-only; a product passing that key may still fail this one) and from scale-to-zero (an idle-cost property; suspension may or may not keep memory). Actor multiplexing (see scale-to-zero) records its live-state half here — only when the suspended actor's MEMORY/process state is what comes back, not a freshly booted worker with the actor's files.

Not assessed

Scale to zeroIdle instances suspend at no / negligible cost. ACTOR MULTIPLEXING (many suspended stateful actors sharing a small pool of pre-warmed workers, woken on request — Agent Substrate, AX, kagent; Rivet's hibernating actors) has NO key of its own (proposal REJECTED v20) — score the idle-cost half HERE and the live-state half under memory-snapshot; the oversubscription ratio / density is a performance characteristic the registry does not key.

Not applicable

GPU computeGPU-backed instances available.

Not supported

Multi-regionCompute placeable in more than one geographic region.

Not applicable

Sandbox templates / custom imagesBuild reusable sandbox/VM images from an OCI image or Dockerfile (a base you fork instances from).

Not supported

Cloud-init / ignition configCustomise a VM / microVM / instance at boot via cloud-init (or ignition) user-data — the same boot-metadata interface as AWS/Azure/GCP. Inject packages, users, SSH keys, files, and run-commands declaratively at launch. Distinct from sandbox-templates (a PREBUILT image you fork instances from) — this is PER-LAUNCH declarative customisation, not a baked image.

Not applicable

Desktop / computer-use sandboxA virtual desktop (GUI) for LLM computer-use — browser + desktop automation, often via noVNC/VNC.

Not supported

Selectable isolation runtimeThe user chooses the isolation runtime per workload / environment / sandbox from more than one option (e.g. microVM vs gVisor vs plain container, native namespaces vs gVisor, gVisor vs VM). Records the CHOICE; the isolation_primitive facet still records the best-evidenced primary. A platform that switches runtime automatically without a user choice (e.g. gVisor forced for GPU) does not pass on that alone.

Not assessed

Networking & delivery 0/3

Automatic HTTPSTLS endpoints provisioned automatically per workload.

Not applicable

Built-in reverse proxyPlatform routes external traffic to workload services.

Not applicable

Custom domainsBring-your-own domain via CNAME/DNS.

Not applicable

Public preview URLsShareable per-deploy / per-branch URLs.

Not applicable

Direct public IPv6Per-instance stable public IPv6 for raw TCP/UDP (beyond a managed HTTPS proxy) — e.g. mosh / raw sockets.

Not applicable

Egress controlsConfigurable OUTBOUND network policy — egress firewall / allowlist / egress proxy on traffic LEAVING a workload. Scoped to OUTBOUND only (v15): private-network REACHABILITY between a user's own services is private-networking, not this. Static-egress-IP belongs here (an outbound concept).

Partially supported

Private networkingA platform-internal private network between a user's OWN services — service-to-service connectivity, VPC, service mesh, and/or service discovery, not exposed to the public internet. DISTINCT from egress-controls (OUTBOUND firewall/allowlist), reverse-proxy (PUBLIC inbound ingress), and direct-ipv6 (per-instance PUBLIC addressing). Admitted v15.

Partially supported

Email send / receiveSend and/or receive email from the workload.

Not supported

Edge KV storeA globally-replicated key-value store read from WITHIN an edge runtime at the PoP (low-latency, eventually-consistent) — Cloudflare Workers KV, Deno KV, Fastly KV Store, Akamai EdgeKV, Vercel Edge Config. Distinct from first-party-db (a managed DATABASE service you provision and connect to) and from http-query-api (HTTP access to an external store) — edge-kv is colocated with the edge compute and read at the edge.

Not applicable

Developer access 1/4

Public APIDocumented programmatic API.

Not supported

OpenAPI specMachine-readable OpenAPI / Swagger spec published.

Not supported

CLIOfficial command-line tool.

Supported

SDKOfficial client library/libraries.

Not supported

SSH accessDirect SSH into workloads.

Not applicable

Filesystem APIRemote file operations over the API (list/read/write/watch/copy/upload) without SSH.

Not applicable

Git deployDeploy from a Git repo / push-to-deploy.

Not applicable

Private container registryHosted private image registry.

Not applicable

E2B-compatible APIDrop-in compatibility with the E2B sandbox API/SDKs — existing E2B client code runs against this product by changing the endpoint/domain + key, with no business-code changes. E2B itself passes as the reference implementation. Distinct from public-api / sdk (the product has its own API — this key says it also speaks another vendor's) and from self-host-oss (a self-hosted E2B-compatible server is scored on both).

Not assessed

HTTP data APIQuery / read / write the stored DATA over plain HTTP/REST (or GraphQL) WITHOUT the native protocol driver (Postgres wire / Redis RESP) — agent- and edge-friendly data access via fetch(). The DATA plane; distinct from public-api (a control-plane / management API). Auto-generated-from-schema REST/GraphQL (PostgREST, pg_graphql) is a flavour of this.

Not applicable

AI & agents 0/10

Built-in coding agentShips a first-party AI coding agent in the product.

Not supported

Agent skillsInstallable / declarable agent skills or tools.

Not supported

AI app builderPrompt-to-app / vibe-coding generation surface.

Not supported

MCP serverExposes a Model Context Protocol server for agents.

Not supported

MCP clientActs as an MCP client / consumes MCP tools.

Not supported

MCP-server hostingRun/host MCP servers as a workload on the platform (distinct from exposing a first-party MCP control server — that is mcp-server).

Not supported

LLM gatewayRoutes LLM API calls through the platform (multi-provider).

Not supported

BYOK for LLMsBring-your-own LLM provider keys.

Not supported

Model inference API (first-party)A FIRST-PARTY hosted LLM/model inference API — the provider serves its OWN catalogue of (usually open-weight) models behind a (typically OpenAI-compatible) endpoint, billed per token. Distinct from llm-gateway (a multi-PROVIDER proxy/router for YOUR calls to external providers like OpenAI/Anthropic) and from byok-llm (bring your own provider keys) — here the platform IS the inference provider, serving its own models.

Not applicable

Autonomous QA agentA first-party AUTONOMOUS AI agent that QA-TESTS a running app / implementation — exercising it like a real user (clicking through flows, checking behaviour) and reporting bugs/defects (often back to a coding agent, or before a human review). Distinct from built-in-agent (a code-AUTHORING agent — this one VERIFIES the result, it does not write the feature; a product can ship both), from ai-guided-code-review (organises a diff for a HUMAN to review — not an autonomous tester), and from observability (runtime metrics/logs, not active testing).

Not supported

Credential-injection proxyThe platform holds real secrets OUTSIDE the workload and injects them on the wire (an egress / broker proxy swaps a placeholder or phantom token for the real credential on matching outbound requests), so code running in the sandbox — e.g. an untrusted agent — never sees or can exfiltrate the real key. Distinct from secrets-management (storage + injection INTO the workload as env vars/files — the workload holds the secret) and from egress-controls (outbound allow/deny POLICY; a product can have an allowlist without injecting credentials, and vice versa).

Not supported

Preinstalled coding agentThe platform installs and launches a THIRD-PARTY coding agent (a harnesses.yml entry — Claude Code, Codex, OpenCode, Pi ...) inside a workspace/sandbox it provisions, with the agent's auth wired in (your own account or API key), so you start an agent session rather than an empty box. Distinct from built-in-agent (a FIRST-PARTY agent the vendor ships) and from harnesses[] (which records that a harness CAN target the product, not that the product runs it for you).

Not assessed

Data 0/0

First-party databasePlatform-provided managed database.

Not applicable

Database branchingCopy-on-write branch of a DATABASE (schema + data) as a dev-lifecycle primitive — branch your data like code, usually instant via a metadata operation. Distinct from snapshots-forks (forks a running compute INSTANCE / VM state) and from preview-environments (a full per-PR environment that may USE a DB branch). For a data product a CoW branch maps HERE, not to snapshots-forks.

Not applicable

Vector searchFirst-class storage, indexing, and similarity search of vector embeddings — a native vector type/index or a managed vector DB. The data-layer primitive for RAG / semantic retrieval.

Not applicable

Ephemeral instant DB (no-signup)Provision a working, usable instance via a single command / HTTP call with NO account, signup, or card — typically time-boxed (expires unless claimed). A zero-onboarding path an agent can hit directly.

Not applicable

Platform & operations 0/3

Scheduled tasksCron / scheduled job primitive.

Not supported

Durable workflowsDurable execution / long-running workflow primitive.

Not applicable

Preview environmentsA full EPHEMERAL environment spun up per PR / branch — compute + addons + (often) a DB branch — torn down on merge. More than preview-urls (which is just a shareable URL for one deploy).

Not applicable

Built-in observabilityFirst-party metrics / logs / tracing / APM / profiling included in the platform (no third-party stack required to see what a workload is doing).

Not supported

Managed CI/CD runnersPlatform-hosted, drop-in CI/CD runner VMs (GitHub Actions / GitLab), typically faster and cheaper than the CI vendor's own hosted runners, billed per runner-minute; isolated + ephemeral per job. Distinct from git-deploy (push-to-deploy of an APP) and scheduled-tasks (cron) — this is managed CI COMPUTE you point your existing pipelines at.

Not supported

Project template marketplaceA marketplace / gallery of prebuilt PROJECT STARTER templates (apps or websites) a user clones or launches a new project from and then customises — the "start from a template" surface of a builder / PaaS. Distinct from sandbox-templates (a COMPUTE base IMAGE you fork runtime INSTANCES from — Dockerfile / OCI), from app-builder (the prompt-to-app GENERATION surface — a product can have both), and from one-click OSS-app catalogues / helm-addon-marketplace (deploy a FINISHED app as-is, not a starter you build on).

Not applicable

Organisation & security 1/4

Built-in authPlatform-provided authentication / IAM for workloads.

Not supported

SSO / SAMLEnterprise single sign-on.

Not supported

Secrets managementFirst-class secrets storage / injection.

Not supported

TeamsMulti-user team accounts.

Not applicable

SharingShare a running workload/resource via link (optionally gated).

Not applicable

BYOCRuns inside the customer's own cloud account.

Not applicable

Deploy into customers' VPCsA SaaS vendor uses the platform to deploy and operate ITS software inside ITS CUSTOMERS' cloud accounts / VPCs (nested or "customer BYOC"), managed from the vendor's control plane. Distinct from byoc (your own workloads in your own cloud account).

Not assessed

Workload identityThe platform gives each workload its own verifiable identity (OIDC token, SPIFFE SVID, cloud IAM federation) that it exchanges for short-lived credentials, so no long-lived static secret is stored or injected. Distinct from secrets-management (stores / injects static secrets) and credential-injection-proxy (swaps real secrets in on the wire).

Not assessed

Self-host / OSS coreOpen-source core that can be self-hosted.

Supported

Proposed, not verified

Capabilities the scan saw but the registry does not track yet. A key is admitted only once two distinct products exhibit it. These are not scored.

Host-distribution passthrough: the sandbox reuses the host's installed OS and toolchain (read-only /usr, /bin, /lib, /etc overlay) instead of an imagehost-toolchain-passthrough
“Unlike Docker/Podman, Drop uses your existing distribution, so there is no container setup work: every program you've already installed is available in the sandbox.” droprun.sh
Selectable isolation runtime per environment/run (native namespaces vs gVisor user-space kernel, switchable back and forth)selectable-isolation-runtime
“Drop supports two runtimes: `native` [...] `gvisor` [...] Both runtimes support the same config options and create identically configured sandboxes. The runtime can be changed back and forth for existing Drop environments.” droprun.sh
Rootless operation (no root / setuid helper / daemon required to create the sandbox)rootless
“Drop doesn't require root and can't execute any operation that the current user is not allowed to execute.” droprun.sh

03 Agent readiness Pre-login surface · counts, not a grade

Can an agent read it?

parse-ability · docs, metadata, machine-readable surfaces

5.5/10

5 full · 1 partial · 55% of applicable checks

llms.txt
Not supported
“https://droprun.sh/llms.txt → 404 (also checked /docs/llms.txt via the docs tree: not present)”
llms-full.txt
Not supported
“https://droprun.sh/llms-full.txt → 404”
robots.txt AI policy
Partially supported
“https://droprun.sh/robots.txt is the single line 'User-agent: *' with no Disallow and no AI-UA mention, i.e. default-allow → half”
JSON-LD
Not supported
“No <script type=application/ld+json> on the homepage. Only itemprop microdata (name/description/image).”
Sitemap
Supported
“https://droprun.sh/sitemap.xml: 11 URLs covering home + all 8 docs pages”
Meta tags
Supported
“<title>Drop sandbox for Linux</title>, a descriptive meta description, og:title/og:url/og:image and a twitter:card with a description. og:description is absent, but twitter:description carries the same text.”
Markdown negotiation
Not supported
“Accept: text/markdown on / returns text/html, and /docs/tour.md → 404. (The markdown sources are public in the repo's website/content/, but the site does not serve them.)”
Content efficiency
Supported
“Homepage ≈6.5% readable text / HTML bytes (2,320 / 35,955), server-rendered Hugo”
Docs discoverable
Supported
“'Docs' in the header nav (/docs) plus an 'Install Drop' hero button. Every docs page is linked from the static HTML with no JS needed.”
Machine-readable pricing
Not applicable
“Free Apache-2.0 tool with no pricing page. There is nothing to price.”
No anti-bot wall
Supported
“Served via Cloudflare (cf-cache-status: HIT). No challenge on any fetched page.”

Can an agent drive it?

act-ability · APIs, SDKs, MCP, credentials, webhooks

1/6

1 full · 0 partial · 17% of applicable checks

Public API
Not supported
“No HTTP/socket control API. The surface is the `drop` CLI + TOML config.”
OpenAPI spec
Not supported
“No spec. None referenced.”
MCP server
Not supported
“No Model Context Protocol mention anywhere on droprun.sh or in the README. /.well-known/mcp → 404.”
SDK packages
Not supported
“No SDKs. Go packages are under internal/ (not importable).”
Auth documented
Not applicable
“No hosted API, so there is nothing to authenticate against.”
Self-serve credentials
Not applicable
“No accounts or keys exist. Install is a curl of a static binary, which is fully agent-executable, but there is no credential surface.”
Webhooks
Not applicable
“No service to emit events.”
CLI tool
Supported
“Official `drop` CLI (init/run/ls/rm/update), static binaries for amd64/arm64 on every release”
Error model
Not applicable
“No API.”
Rate limits documented
Not applicable
“No API.”
Agent config files
Not supported
“No AGENTS.md/CLAUDE.md/skills in the repo (the .gitignore explicitly ignores CLAUDE.md and .claude), and no /.well-known/agent.”

04 Coding harnesses Which agents and IDEs are documented to run here

Advertised integrations 1

Claude Code Anthropic
Supported
“This tour demonstrates key characteristics of Drop. We will install and run Claude Code within Drop to work on a project stored in `~/project`. [...] (drop)alice@zax:~/project$ claude --dangerously-skip-permissions”
Source droprun.sh
The only harness named anywhere on the surface: a full docs tour installing Claude Code inside an env and running it in skip-permissions mode.

Plausible compatibility 0

Derived, not stored: the product passes SSH access, so any CLI coding agent is installable inside it. This is plausible compatibility, not a supported integration.

No general shell access documented, so nothing is derived. Absence is not unsupported.

05 Compliance & regions Standards held, with ordinal rigor; never a score

Certifications held 0

No attestations found on the public surface. Empty means nothing is advertised, after checking, not that nothing was checked.

Regions —

No region list captured.

06 Open source Apache-2.0 · fetched by machine, not quoted from marketing

Licence & repository

Licence: Apache-2.0OSI
Primary language: Go
Stars: 299
Forks: 9
Created: 25 Jul 2025
Archived: No

Activity

Last commit: 25 Sep 2026
Latest release: v0.2.1 · 24 Aug 2026
Issues open / closed: 6 / 5 · 45% closed

9 tagged releases from v0.1.2 (2026-03-19) to v0.2.1 (2026-08-24): roughly monthly, with two bursts of back-to-back fixes (v0.1.2→v0.1.3 in 4 days, v0.2.0→v0.2.1 in 4 days). Pre-1.0. Each release ships static drop-linux-amd64 / drop-linux-arm64 binaries.

Fetched via gh on 25 Sep 2026.

Self-hosting

Methods: binarysource-build
You operate: passt

This IS the only way to run it. Drop is a local CLI with no hosted tier, so cost = free under Apache-2.0 + your own Linux machine. Linux only (amd64/arm64). It needs the distro `passt` package for networking and, for the gVisor runtime, a separately installed `runsc`. Ubuntu 24+ needs a hand-written AppArmor profile to allow unprivileged user namespaces, and Fedora needs a custom SELinux module for pasta. Neither is automated. No distro packages (no apt/dnf/brew). Install is `curl` of the release binary or `go install`.

Documentation signals 6/8

SupportedQuickstart
SupportedAPI reference
SupportedSelf-host guide
SupportedChangelog
Not supportedOpenAPI
Not supportedllms.txt
SupportedExamples
SupportedArchitecture doc

07 Company & presence Facts only, not scored

Company

Headquarters: Kraków, Poland (maintainer location; no company)
Founded: 2025
Team size: Not disclosed
Hiring: No (individual project)

Notable people

Jan Wróbel: sole author. Previously built wwwhisper (auth add-on for Heroku/OpenShift) and co-founded Shapespark (3D web visualization), per mixedbit.org. Does published security research on user-namespace escapes (CVE-2019-5736 replay against Drop, 2026-06-12).

Activity

Last release: v0.2.1 — 2026-08-24 github.com
Last blog post: 2026-06-12 — '/proc/self/exe overwrite from within a user namespace' (maintainer's blog, about Drop) mixedbit.org

Presence 4

Links found on the public surface scanned; not an exhaustive search.

08 Sources & method How this entry was produced

Public surface only

Everything here was read from droprun.sh's unauthenticated pages, docs, pricing and repository on 25 Sep 2026. Nothing behind a login was used.

Statuses, not scores

Capabilities are pass, partial or fail against a written rubric (features v18). “Not assessed” means the scan predates the key.

Confidence is separate

Each fact carries its own confidence. Low-confidence facts are shown, never hidden, so you can weigh them.

Schema: scan v5
Features registry: v18
Keys tracked today: 61
Assessed on this entry: 55

Something wrong on this page?

Nearby in the atlas

All Sandbox products →