Hacktoberfest 2026: the issues maintainers tagged for October, open and beginner-friendly. Browse Hacktoberfest issues

Setup guide and security page: say that Method 2 (bare metal) has no container boundary

Open
  • Is something wrong?
#385 0 comments 0 reactions 0 assignees View on GitHub

Maintainers usually reply within 1 day

Nobody has claimed this yet.

Assessment

Difficulty
3/5
Estimated time
1-2 days
Newbie friendliness
66/100
Issue type
Documentation
Clarity
Clearly specified
Activity status
Active
Tech stack
python
Domain
documentation

Research direction

Start with docs/SETUP.md Method 2, then read the referenced statements in SECURITY.md, docs/CONFIGURATION.md, proxy/permissions.go, and the translated pages. Check docs/CLI.md for the permission-mode names and run pytest tests/contracts. Done means the container-boundary conditions, bare-metal caveat and security link are documented, and the credential-read row names the modes that prompt.

Written by the indexing model from the issue text.

Description

area/docs area/sandbox documentation status/ready

What needs doing

docs/SETUP.md has "Method 2: Bare Metal" (a preview method for
Linux): all services run as local processes, without containers. In
that method the sandbox service is a local process too. So a command
that the model runs is a process on the user's machine, as the user.

Three places describe the boundary with no condition:

  • SECURITY.md: "shell commands run in the sandbox container".
  • docs/CONFIGURATION.md: "run_command runs a real shell inside
    the sandbox container, which is the actual safety boundary".
  • The comment on the trust modes in proxy/permissions.go:
    "trusted: commands run in the isolated sandbox container".
  1. One paragraph in Method 2 of docs/SETUP.md: this method has no
    container boundary. Say what still stands before a command (the
    command rule, the permission prompts) and what does not (the
    read-only root, the project folder as the only writable mount,
    the container's limits).
  2. One condition in SECURITY.md and in the sentence of
    docs/CONFIGURATION.md: the description holds for the container
    setups.
  3. The same lines in the translated pages.
  4. One more sentence of docs/CONFIGURATION.md lacks a condition of
    another kind. The row ATLAS_ALLOW_CREDENTIAL_READS says that shell
    commands "ask for approval instead". That holds in the modes
    default and accept-edits. In yolo a command is not asked about
    (docs/CLI.md, the table of permission modes). Name the modes in
    that sentence.

Why

A reader of the security page should know which setup its promise is
for.

Done when

  • The three places say it, and Method 2 links to the security
    page.
  • The row ATLAS_ALLOW_CREDENTIAL_READS names the modes in which a
    command asks.
  • pytest tests/contracts passes.
Dominant language
Python
Stars
2.1k
Forks
197
Avg merge
14h 1m
Merged PRs (30d)
201

Getting set up

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

More from inferstep/ATLAS

All issues in inferstep/ATLAS

Similar issues

More Python issues

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.