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/tomascupr/sandstorm/agents-mdgit clone --depth 1 https://github.com/tomascupr/sandstormWrote 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/tomascupr/sandstorm/agents-md)<a href="https://agentmods.dev/instructions/tomascupr/sandstorm/agents-md"><img src="https://agentmods.dev/badge/instructions/tomascupr/sandstorm/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 | $0.00555 | $0.00555 |
| Opus 5 | $0.00278 | $0.00278 |
| Sonnet 5 | $0.00111 | $0.00111 |
| Haiku 4.5 | $0.00056 | $0.00056 |
Grade A, and why
sandstorm AGENTS.md scanned grade A with 0 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 4d 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.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 29 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Repository Guidelines
Project Structure & Module Organization
Core application code lives in src/sandstorm/. Key modules include main.py for the FastAPI app, cli.py for the ds CLI, sandbox.py for E2B orchestration, and slack.py / slack_routes.py for Slack support. Deployment entrypoints live in api/, documentation in docs/, runnable configs in examples/, and test coverage in tests/. Keep new modules inside src/sandstorm/ and mirror them with focused tests such as tests/test_auth.py.
Build, Test, and Development Commands
Use Python 3.11+ and uv.
uv sync --extra dev # install app + dev dependencies
ds serve --reload # run the FastAPI server locally
uv run pytest tests/ # run the test suite
ruff check src/ tests/ # lint
ruff format --check src/ tests/ # formatting check
uv run pyright src/sandstorm/ # type check
uv run python -c "from sandstorm.main import app" # smoke-test imports
If you use hooks locally, install them with uv run pre-commit install.
Coding Style & Naming Conventions
Follow Ruff defaults with the repo’s configured 99-character line length. Use 4-space indentation, snake_case for modules, functions, and variables, and PascalCase for classes. Keep FastAPI request/response models in models.py-style modules and prefer small, composable helpers over large route handlers. Run ruff check --fix src/ tests/ and ruff format src/ tests/ before opening a PR.
Testing Guidelines
Tests use pytest and live under tests/ with filenames matching test_<feature>.py. Prefer fixtures in tests/conftest.py and mock E2B or external API calls instead of hitting live services in unit tests. Some optional-path tests skip when Slack or OpenTelemetry extras are absent; install the relevant extras when working in those areas. For manual end-to-end work, use .env.example as the template and never commit real API keys.
Commit & Pull Request Guidelines
Recent history follows Conventional Commits: feat:, fix:, docs:, refactor:, test:, chore:. Keep commits scoped to one change. PRs should explain what changed, why it changed, and link the related issue (Fixes #123). Include screenshots or sample CLI/API output when changing user-facing behavior such as Slack flows, dashboard HTML, or streamed responses.
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.
- 4d ago First seen · 29 lines · 555 tokens per session scan A 8c9eb0bdd646
sandstorm AGENTS.md is an instructions file published in the GitHub repository tomascupr/sandstorm (447 stars, last pushed 4mo ago), licensed MIT. It adds 555 tokens to every session, about $0.0028 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other instructions, from other repositories
claude-code-docs CLAUDE.md
Claude Code instructions for thevibeworks/claude-code-docs, covering claude.md, repository purpose, fetcher, documentation resources and repository structure.
claude-architect CLAUDE.md
Instructions for timothywarner-org/claude-architect, covering claude.md, what this repo is, how this repo bootstraps (read before running anything), class-day lifecycle (the july 2026 delivery pattern) and architecture: how the pieces fit.
lume AGENTS.md
Instructions for CavinHuang/lume, covering working agreements, 改动工作流, format, gitmoji reference and rules.
harness CLAUDE.md
Instructions for mastersof-ai/harness, covering masters of ai harness, related docs, quick orientation, running locally and cli subcommands.
Traceplane AGENTS.md
Instructions for jaxxchen003/Traceplane, covering agents.md, 1. 实现顺序, 2. 绝对约束, 3. 保底字段 and 4. 模块边界.
Kilntainers CLAUDE.md
Instructions for Kiln-AI/Kilntainers, covering kilntainers: secure agent sandboxes, specs and commands & tools.