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 martinsbrezauckis/shellx --skill shellx-hostgit clone --depth 1 https://github.com/martinsbrezauckis/shellxWrote 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/martinsbrezauckis/shellx/shellx-host)<a href="https://agentmods.dev/skills/martinsbrezauckis/shellx/shellx-host"><img src="https://agentmods.dev/badge/skills/martinsbrezauckis/shellx/shellx-host/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/martinsbrezauckis/shellx/shellx-host"><img src="https://agentmods.dev/badge/skills/martinsbrezauckis/shellx/shellx-host.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 10 findings, up to high
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 →
- high Tool Misuse · line 513 Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).Fix: Validate all tool parameters against an allowlist. Reject dangerous parameter values (shell=True, --force, -rf /) and use safe defaults.
- high Tool Misuse · line 636 Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).Fix: Validate all tool parameters against an allowlist. Reject dangerous parameter values (shell=True, --force, -rf /) and use safe defaults.
- high Tool Misuse · line 640 Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).Fix: Validate all tool parameters against an allowlist. Reject dangerous parameter values (shell=True, --force, -rf /) and use safe defaults.
- medium Rogue Agent · line 368 Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.Fix: Remove any persistence mechanisms (cron jobs, startup scripts, state files). Skills should not maintain state across sessions without explicit user consent.
- medium Agent Snooping · line 486 Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.Fix: Remove all code or instructions that list or read other skills' files or directories. Skills should operate independently; cross-skill access is a privilege escalation.
- medium Rogue Agent · line 533 Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.Fix: Remove any persistence mechanisms (cron jobs, startup scripts, state files). Skills should not maintain state across sessions without explicit user consent.
- medium MCP Rug Pull · line 692 npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.Fix: Pin the version: npx @scope/[email protected]
- medium MCP Rug Pull · line 699 npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.Fix: Pin the version: npx @scope/[email protected]
- medium Excessive Agency · line 723 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.
- medium Excessive Agency · line 749 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 | $0.00089 | $0.11886 |
| Opus 5.5 | $0.00036 | $0.04754 |
| Sonnet 5.5 | $0.00018 | $0.02377 |
| Haiku 4.5 | $0.00009 | $0.01189 |
Grade A, and why
shellx-host scanned grade A with 1 finding 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 1mo 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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
The user may drive YOU through this API from outside (curl, scripts, How it starts
The opening of the file, as written. The whole thing — 847 lines — stays where its author put it; the contents beside it link to each section on GitHub.
ShellX host session
Activation precondition
Use this skill only after positive evidence that the current session is hosted
inside ShellX. Valid evidence is explicit ShellX ACP/session metadata in the
current prompt or runtime, advertised shellx-host-http__* or
grok-shell-host__* host tools, or an explicit user request about ShellX host
APIs or surfaces. Global skill availability, a ShellX repository or product
name, WSL/Windows context, SSH access, or an ordinary cross-host task is not
evidence. If this precondition is absent, stop using this skill and continue
with the normal agent tools.
shellX is a Tauri 2 desktop application hosting this grok session over ACP stdio. Assume more than a plain CLI, but only use host surfaces when they help.
1. ACP terminals are unavailable
Current ShellX builds always advertise clientCapabilities.terminal: false and
unconditionally rejects provider-originated terminal/* ACP requests with
JSON-RPC -32601. Do not retry those calls. Use the provider's native tools
or the ShellX Agent host tool for supervised work instead.
The interactive Terminal panel visible to the user is a separate,
operator-owned portable-pty surface. Its existence does not enable ACP
terminal access for an agent.
2. Vault — mediated secrets and grants
ShellX Vault is the ShellX secret authority. Agents plan from safe metadata, then ask for mediated use. Do not ask for raw secret values and do not copy secrets into chat. If Vault is manually locked, ask the user to unlock it in the Vault panel before retrying grant-backed work.
Access via the host MCP server. In many shellX sessions the same host tools are
advertised with both grok-shell-host__ and shellx-host-http__ qualified
names. Prefer shellx-host-http__... for mutating or tab-aware tools when it
is advertised because that transport carries the active tab and permission
gate. Use grok-shell-host__... for read-only discovery or as the local
fallback.
If shellx-host-http__capabilities_summary or
grok-shell-host__capabilities_summary is advertised, call it directly
for a compact current tool map before broad tool discovery. Only eight compact
entry tools are advertised by default: capabilities_summary, search_tool,
host_read, host_act, browser_read, browser_act, cut_read, and
cut_act. Use targeted
search_tool queries for an exact hidden Host action schema, then pass the
action name and its exact fields through host_read { action, params } for
read-class operations or permission-gated host_act { action, params } for
mutations and external side effects. Prefer the shellx-host-http__ qualified
gateway when advertised. The names in the Vault and Host maps below are action
names for that routing flow, not additional always-advertised tools. Avoid
full_inventory unless debugging tool-schema drift, because the result is
large and Grok may store it as a session artifact instead of showing it clearly
in chat. Browser actions use their dedicated browser_read and browser_act
gateways instead. ShellX Cut discovery and execution use cut_read and
cut_act; Cut's generated verb catalog stays behind that compact pair.
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.
- 1mo ago First seen · 847 lines · 89 tokens per session scan A f180cd7666a5
shellx-host is a skill published in the GitHub repository martinsbrezauckis/shellx (21 stars, last pushed 7d ago), licensed MIT. It adds 89 tokens to every session and 11,886 once invoked, about $0.0004 per session on Opus 5.5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
dws
A command-line guide for DingTalk, a workplace platform with chat, documents, calendars, approvals, tasks, and other business tools.
tmeet-skill
A command-line tool for Tencent Meeting, a video-conferencing service. It covers sign-in, meetings, recordings, transcripts, reports, invitations, and controls during a meeting.
anarlog
Query Anarlog meetings, notes, summaries, transcripts, participants, action items, and recurring history. Use when a user asks about their Anarlog meeting data or needs meeting context for another task.
lark-base
A guide for managing Lark Base, Feishu's spreadsheet-like database and workspace tool. It covers tables, fields, records, views, formulas, forms, dashboards, applications, workflows, and permissions.
tencent-docs
A skill for creating, reading, editing, searching, and managing Tencent Docs, an online document and spreadsheet service.
government-writing
A skill for writing formal Chinese government documents, including notices and opinions, in the required official format.