agentleFS
AhmiDarrowAhmiDarrow/RemedyAI/skills/gauntlet-security/SKILL.md

gauntlet-security

Skill5 starsChanged 23 days agoApache-2.0View on GitHub

Security gauntlet for Remedy — unit jail/SSRF/auth suite plus controlled live red-team probes. Use when security audit, red-team, gauntlet, harden, jail, SSRF, auth bypass, or before ship after security-sensitive changes.

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 4 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Before release if security surface moved

    line 27
  2. Distinctive:

    Pass: all green. Fail: fix before live probes.

    line 59
  3. Distinctive:

    Start release or isolated API first.

    line 68
  4. Distinctive:

    Safe by design: no disk wipe, no real destructive shell.

    line 76
  5. Very specific:

    If new findings: document in docs/ dated note + fix or track.

    line 97

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Security gauntlet

Goal

Prove jails, auth, SSRF, and host boundaries still hold after changes. Threat model: same Windows user ≈ owner power; harden against misconfig and agent tool abuse, not full user compromise.

Artifacts: docs/REDTEAM_*.md, docs/SECURITY_AUDIT_*.md, docs/_redteam_live_results.json, scripts/_redteam_live_probes.py.

When to use

  • Security-sensitive PR (auth, shell, paths, media, computer host, webhooks)
  • User says gauntlet / red-team / harden / jail check
  • Before release if security surface moved

Preflight

  1. Repo root = RemedyAI; venv ready (uv sync / .venv).
  2. Know which API is under test (release :7400 vs isolated :7410).
  3. Set env if non-default:
$env:REMEDY_API = "http://127.0.0.1:7400"   # or 7410
$env:REMEDY_HOME = "$env:USERPROFILE\.remedy"

Gate A — Unit security suite (always)

From repo root:

uv run pytest -q --tb=short `
  tests/test_api_auth.py `
  tests/test_web_fetch_ssrf.py `
  tests/test_project_write_jail.py `
  tests/test_project_scan_jail.py `
  tests/test_zip_import_security.py `
  tests/test_secret_store.py `
  tests/test_secret_acl_no_everyone.py `
  tests/test_access_scope.py `
  tests/test_agency_hardening.py `
  tests/test_tool_policy.py `
  tests/test_provider_sanitize.py

Pass: all green. Fail: fix before live probes.

Also useful neighbors when you touched them:

  • tests/test_hidden_process.py, tests/test_computer_use.py (host auth)
  • tests/test_media_api.py, tests/test_uninstall_wipe_paths.py

Gate B — Live probes (API must be up)

Start release or isolated API first. Then:

.\.venv\Scripts\python.exe scripts/_redteam_live_probes.py

Expect JSON summary on stdout; often writes docs/_redteam_live_results.json.

Safe by design: no disk wipe, no real destructive shell; probes auth, path jail, SSRF helpers, computer host shape, webhooks.

Gate C — Shell write jail spot checks

.\.venv\Scripts\python.exe scripts\_prove_write_jail.py
# optional multi-run:
.\.venv\Scripts\python.exe scripts\live_project_write_jail_10x.py

Gate D — Report

Summarize for the user:

AreaResult
Unit jails / auth / SSRFpass/fail
Live probespass/fail + notable findings
Residual riskhonest (owner power, loopback bootstrap, etc.)

If new findings: document in docs/ dated note + fix or track; do not ship known P0 (unauth host, jail bypass).

Anti-patterns

  • Live probes without unit suite green
  • Probing production non-loopback hosts
  • Enabling dual messengers “to test security”
  • Claiming “secure” because pytest passed while live host was down

Related skills

  • soak-product — broader product health after harden
  • project-etiquette — ship only after gates
  • change-safety — blast radius for security surfaces
  • self-dev-loop — full self-dev sequence

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.