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 skills add naveedharri/benai-skills --skill allow-teamgit clone --depth 1 https://github.com/naveedharri/benai-skillsWrote 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/skills/naveedharri/benai-skills/allow-team)<a href="https://agentmods.dev/skills/naveedharri/benai-skills/allow-team"><img src="https://agentmods.dev/badge/skills/naveedharri/benai-skills/allow-team/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/naveedharri/benai-skills/allow-team"><img src="https://agentmods.dev/badge/skills/naveedharri/benai-skills/allow-team.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 1 finding, up to medium
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- medium Excessive Agency · line 8 Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.Fix: Add human-in-the-loop confirmation for destructive, irreversible, or high-impact operations. Never auto-execute commands that modify files, send data, or alter system state.
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 5.1 | $0.00181 | $0.01845 |
| Opus 5 | $0.00090 | $0.00923 |
| Sonnet 5 | $0.00036 | $0.00369 |
| Haiku 4.5 | $0.00018 | $0.00185 |
Grade A, and why
allow-team 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 7d 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 — 102 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Allow Team
Takes a local Open WebUI or Odysseus and makes it reachable by other people, without asking them to install anything. The tunnel is the easy part. Confirming the login is actually enforced before the URL exists is the job.
Before you start
Run the check in references/environment-check.md first. This skill needs a shell on the user's own machine, because the thing being shared is running on it. If the environment is a sandbox or container, stop and tell the user to run this in Claude Code on the computer hosting the harness.
What this skill is doing
It puts a door on the public internet that leads into the user's laptop. The harness's own login is the lock on that door, and it is the only lock. That is the right design, because a second password in front of it just confuses the teammate about which credential to use. It also means the safety gate is not optional: if the login is off, misconfigured, or bypassable, there is nothing else standing between a stranger and the user's chat history.
Gate first. Tunnel second.
Steps
Track progress:
Task Progress:
- [ ] 1. Confirm a harness is running
- [ ] 2. Run the safety gate
- [ ] 3. Get ngrok ready
- [ ] 4. Get consent, with the exposure spelled out
- [ ] 5. Open the tunnel
- [ ] 6. Prove the lock holds from outside
- [ ] 7. Render the handover report
1. Confirm a harness is running
Run the preflight block in references/tunnel-steps.md. It detects Open WebUI (8080, /health) or Odysseus (7860, /api/version) and sets $HARNESS and $PORT. Do not assume a port; read it from what is actually listening.
If nothing is running, stop and offer /install-openwebui, or /local-ai-setup if they have not picked a harness yet. Do not install anything from inside this skill. If both are running, ask which one to share rather than opening two tunnels.
2. Run the safety gate
Go to references/safety-gate.md and run it in full. It reads the harness's real auth posture, Open WebUI or Odysseus, and decides whether this instance is safe to expose.
What ships with it
5 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 7d ago First seen · 102 lines · 181 tokens per session scan A 927cf70ad620
allow-team is a skill published in the GitHub repository naveedharri/benai-skills (61 stars, last pushed today), licensed MIT. It adds 181 tokens to every session and 1,845 once invoked, about $0.0009 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 2026-09-05.
Other skills, from other repositories
dell-idrac-bmc-ops
Bring a Dell PowerEdge BMC (iDRAC) onto the network headlessly, standardize its baseline settings, stage its firmware, and drive an unattended Proxmox/Linux install through it — including generation-specific Redfish/racadm gotchas, the identity-before-power-action rule, and the phantom-drive failure mode caused by…
homelab-runbooks
Use when powering on a sleeping DR node, converging DNS records for a new ingress vhost, or bringing up OpenBao/Terrakube JWT identity — homelab operational runbooks for power management, DNS, and secrets-engine bring-up.
proxmox-cluster-ops
Operate a Proxmox VE cluster safely — read-only inspection with pvesh/pct/qm/pvecm instead of hand-editing a live guest, node-by-node package updates that respect quorum, and the shape of joining a new node to an existing cluster. Use when inspecting cluster or guest state, planning a rolling update across cluster…
terrakube-ops
Operate self-hosted Terrakube (remote OpenTofu/Terraform plan/apply, state, and workspace locking) — the canonical login/plan/apply workflow, why a targeted apply is dangerous, workspace lock recovery, the offline-mirror gotcha, and the token-rotation trap. Use when running or observing a Terrakube plan/apply…
openbao-dynamic-aws-creds
Replace a static AWS access key on a workstation or CI runner with short-lived STS credentials minted on demand by OpenBao's (or Vault's) AWS secrets engine, via an AWS credentialprocess wrapper. Covers the architecture, bring-up order, verification, plugin-upgrade caution, and common failure modes. Use when a…
llm-router-ops
Operate a self-hosted OpenAI-compatible LLM router/proxy (e.g. LiteLLM) in front of one or more backends — the minimal client-wiring block for every client type, the context-window advertisement gotcha, the env-vs-persisted-config gotcha, and why an unauthenticated health probe should 401, not 200. Use when wiring a…