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 skills/aduermael/herm/device-interactionnpx skills add aduermael/herm --skill device-interactiongit clone --depth 1 https://github.com/aduermael/hermWrote 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/aduermael/herm/device-interaction)<a href="https://agentmods.dev/skills/aduermael/herm/device-interaction"><img src="https://agentmods.dev/badge/skills/aduermael/herm/device-interaction.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 | $0.00023 | $0.01775 |
| Opus 5 | $0.00012 | $0.00888 |
| Sonnet 5 | $0.00005 | $0.00355 |
| Haiku 4.5 | $0.00002 | $0.00178 |
Grade A, and why
device-interaction 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 5d 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.
Copies of this mod
1 near-identical copy found in the catalogue:
- device-interaction — 95% identical, 4 lines differ
How it starts
The opening of the file, as written. The whole thing — 133 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Device Interaction
TRIGGER when: user asks to verify/test/check if the app works on device, after implementing a UI-affecting feature that needs device verification, user says "does it work", "test this", "check on device", user reports UI doesn't work as expected, need to debug touch/interaction issues. DO NOT TRIGGER when: user asks about unit tests only, build-only requests without device testing, code review without device testing, simulator configuration questions, changes that don't affect UI (e.g. comments, refactors, non-UI logic).
For the Main Agent
This is a SUBAGENT skill. Invoke it via the Agent tool when device verification is needed. If there is an open session for that work, provide that session identifier to a subagent for exclusive use by that subagent.
Agent tool:
- subagent_type: "general-purpose"
- description: "Verify login feature works"
- prompt: "Using the device-interaction skill, verify that the login feature works correctly on session <session-identifier>. Launch the app, capture screenshot and UI hierarchy, check that the login button is visible and tappable, and report if the implementation is working correctly."
After implementing a UI-affecting feature, invoke this skill to verify the implementation works on a device.
Session Lifecycle
DeviceInteractionStartSession (do this early, runs in the background)
→ DeviceInteractionInstallAndRun (after each code change; includes building)
→ DeviceEventSynthesize (interact + observe, repeatable)
→ DeviceInteractionEndSession (when done — keeping sessions open is resource-heavy)
DeviceInteractionStartSession tool
Device Discovery
When opening a new device interaction session, pass a device identifier to select a device, or omit it to use the current destination. Pass any non-matching value to get a list of available targets.
DeviceInteractionInstallAndRun tool
Optional Parameters
commandLineArguments— arguments passed to the app at launch. Use$(inherited)as a token to preserve the scheme's existing arguments (e.g.["$(inherited)", "--reset-state"]to add an extra argument at the end).environmentVariables— key/value pairs set in the app's environment at launch. Use"$(inherited)"as a key to preserve the scheme's existing environment variables (e.g.{"$(inherited)": "", "DEBUG_MODE": "1"}).
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.
- 5d ago First seen · 133 lines · 23 tokens per session scan A 0494967290e8
device-interaction is a skill published in the GitHub repository aduermael/herm (232 stars, last pushed 23d ago), licensed MIT. It adds 23 tokens to every session and 1,775 once invoked, about $0.0001 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 skills, from other repositories
Sanad Client Tester
Comprehensive testing and interactive verification protocol for the Sanad Flutter client including unit, widget, E2E, and agent-driven UI testing through the Dart VM Service.
test-xcode
Build and test iOS apps on simulator using XcodeBuildMCP.
e2e-testing
Guide for running end-to-end tests of the Qwen Code CLI, including headless mode, MCP server testing, and API traffic inspection. Use this skill whenever you need to verify CLI behavior with real model calls, reproduce user-reported bugs end-to-end, test MCP tool integrations, or inspect raw API request/response…
agent-reproduce-align
Use after a Codex or Claude Code feature has been implemented in Qwen Code to run the selected reference agent and Qwen Code under the same scenario, capture HTTP and terminal traces, compare request bodies, tool/function schemas, outputs, and iterate until the reproduced behavior is close enough.
run-integration-tests
Build, pack, and run .NET MAUI integration tests locally. Validates templates, samples, and end-to-end scenarios using the local workload.
webapp-testing
Start/reuse a local app, wait for readiness, inspect rendered state/console/network, act from observed selectors, and verify with evidence.