Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add instructions/tastyeffectco/sandboxd/agents-mdgit clone --depth 1 https://github.com/tastyeffectco/sandboxdWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/instructions/tastyeffectco/sandboxd/agents-md)<a href="https://agentmods.dev/instructions/tastyeffectco/sandboxd/agents-md"><img src="https://agentmods.dev/badge/instructions/tastyeffectco/sandboxd/agents-md.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.02187 | $0.02187 |
| Opus 5 | $0.01094 | $0.01094 |
| Sonnet 5 | $0.00437 | $0.00437 |
| Haiku 4.5 | $0.00219 | $0.00219 |
Grade B, and why
sandboxd AGENTS.md scanned grade B with 2 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Sends data to an external URLmediumData exfiltration
A POST to an outside endpoint may be telemetry or may be exfiltration; either way the mod talks to somewhere, and you should know where.
curl -s -XPOST $API/v1/agents/anthropic/api-key -d '{"api_key":"sk-ant-..."}' Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
curl -s http://127.0.0.1:9090/healthz # -> ok How it starts
The opening of the file, as written. The whole thing — 158 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md — operating sandboxd
A complete, self-contained runbook for an AI agent (or a human) to install, run,
use, and remove sandboxd with no outside knowledge. Commands are
copy-pasteable. Human-readable docs: README.md,
ARCHITECTURE.md.
What this is
A single-host service that creates isolated Linux dev containers ("sandboxes"),
each with an HTTP/HTTPS preview URL for a dev server running inside it. It runs
entirely on Docker: a Go control plane (sandboxd) + Traefik, both in
containers; each sandbox is a sibling container. Sandboxes stop when idle and
wake on the next request. Workspaces persist on disk.
Prerequisites
- Linux host with Docker Engine + the Compose plugin (
docker compose). - Ability to run Docker (either your user is in the
dockergroup, or you havesudo; the scripts auto-detect and usesudoif needed). - Assumes a standard Docker daemon. (If the daemon uses
userns-remap, the shipped--userns=hostdefaults keep it working — see ARCHITECTURE.md.)
Install
git clone https://github.com/tastyeffectco/sandboxd.git
cd sandboxd
./install.sh
install.sh is idempotent. It: checks Docker, copies .env.example→.env,
builds the base image (sandboxd-base:0.3.0) and the control plane, creates the
data dir, and runs docker compose up -d. The base-image build takes a few
minutes the first time, then caches.
Configuration lives in .env (all keys documented in .env.example). The two
you may change before installing:
HTTP_PORT(default80) — set to e.g.8088if port 80 is taken; preview URLs then include it.SANDBOXD_API_BIND(default127.0.0.1:9090) — where the API is published.
Verify it's up:
curl -s http://127.0.0.1:9090/healthz # -> ok
curl -s http://127.0.0.1:9090/readyz # -> ready
Core API
Base URL = http://${SANDBOXD_API_BIND} (default http://127.0.0.1:9090).
Auth is off by default (local). If you set SANDBOXD_API_AUTH_DISABLED=false
SANDBOXD_API_TOKENS=name:secret, add-H "Authorization: Bearer secret".
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 3d ago First seen · 158 lines · 2,187 tokens per session scan B 703cbef8add8
sandboxd AGENTS.md is an instructions file published in the GitHub repository tastyeffectco/sandboxd (925 stars, last pushed 5d ago), licensed MIT. It adds 2,187 tokens to every session, about $0.0109 per session on Opus 5. A static security scan graded it B with 2 findings (sends data to an external url, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other instructions, from other repositories
MultiGen CLAUDE.md
Instructions for lingyuanli/MultiGen, covering claude.md, architecture, development commands, full stack (docker) and api (local).
docker-nixuser AGENTS.md
Instructions for grigio/docker-nixuser, covering development guide, ci publishing (critical), multi-platform ci builds, flake auto-update and docker image.
aicontainer AGENTS.md
Instructions for stefanoginella/aicontainer, covering agents.md, repository layout, image tag pinning (read before editing the cli), version drift / update notifications and project identity and host-side validation.
aicontainer CLAUDE.md
Instructions for stefanoginella/aicontainer, a project described as: Sandboxed devcontainer for running Claude Code, Codex, and OpenCode in bypass / auto-approve mode.
appwrite AGENTS.md
AGENTS.md instructions for appwrite/appwrite, covering appwrite, commands, stack, layout and libraries.
microsandbox AGENTS.md
AGENTS.md instructions for superradcompany/microsandbox, covering agents.md, scope, project map, design principles and backward compatibility review.