Getting it into your agent
This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.
/plugin marketplace add choiyounggi/groundwork/plugin install guardrailsWrote 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/choiyounggi/groundwork/self-test)<a href="https://agentmods.dev/skills/choiyounggi/groundwork/self-test"><img src="https://agentmods.dev/badge/skills/choiyounggi/groundwork/self-test/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/choiyounggi/groundwork/self-test"><img src="https://agentmods.dev/badge/skills/choiyounggi/groundwork/self-test.svg" alt="Reviewed on agentmods" width="80" 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 5.1 | $0.00079 | $0.00261 |
| Opus 5.5 | $0.00032 | $0.00104 |
| Sonnet 5.5 | $0.00016 | $0.00052 |
| Haiku 4.5 | $0.00008 | $0.00026 |
Grade D, and why
self-test scanned grade D with 3 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Unrestricted tool accessmediumExcessive agency
A wildcard tool grant or "run any command" leaves no least-privilege boundary at all.
**not** execute any command. Recursive force deletehighDestructive command
rm -rf with a variable or a broad path is one typo away from removing the wrong tree.
description: Prove the guardrails Bash guard is live — runs a simulation that feeds representative dangerous commands (curl|sh, rm -rf, DROP TABLE, kubectl delete, cloud deletes, …) through the real guard and shows the b Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
description: Prove the guardrails Bash guard is live — runs a simulation that feeds representative dangerous commands (curl|sh, rm -rf, DROP TABLE, kubectl delete, cloud deletes, …) through the real guard and shows the b What it actually says
guardrails: self-test
Run the bundled self-test to show the guard is active. It injects dangerous command strings into the guard as hook data and prints each decision — it does not execute any command.
Do this
- Run:
bash "${CLAUDE_PLUGIN_ROOT}/scripts/self-test.sh" - Show the user the output table verbatim.
- Summarize: block-tier commands were denied, ask-tier would prompt, harmless commands passed — and nothing was executed.
Hard rule
Never run any of the dangerous example commands yourself. The whole point is that they travel through the guard as data, never to a real shell. If the script exits non-zero, report which case mismatched — do not "fix" it by running commands.
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 · 26 lines · 79 tokens per session scan D bd99a501fe82
self-test is a skill published in the GitHub repository choiyounggi/groundwork (3 stars, last pushed 3d ago), licensed MIT. It adds 79 tokens to every session and 261 once invoked, about $0.0003 per session on Opus 5.5. A static security scan graded it D with 3 findings (unrestricted tool access, recursive force delete, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
improve-code-quality
Guided journey from a working-but-untested vibe-coded prototype to a production-ready product with tests, clean structure, a business-rules boundary, and resilience at scale. Orchestrates nine skills phase by phase - working-with-legacy-code, clean-code, refactoring-patterns, software-design-philosophy…
working-with-legacy-code
Safely change and test untested codebases using Feathers' "Working Effectively with Legacy Code". Use when the user mentions "legacy code", "no tests", "untested codebase", "how do I test this", "seams", "characterization tests", "golden master", "sprout method", "afraid to change this code", "monster method"…
dotnet-test
This skill should be used when running .NET tests selectively with a build-first, test-targeted workflow. Use it for running tests with xUnit focus.
dotnet-verify
This skill should be used when working with Verify snapshot tests in .NET projects. Use when updating verified snapshots after intentional code changes, accepting new snapshots, discovering verify tests, or troubleshooting snapshot mismatches. Trigger phrases include "verify tests", "update snapshots", "accept…
playwright-harness
Drive a real browser with Playwright on any OS and gate on what you observe: write a script, launch Chromium (headless by default; a shared headed window when a human must watch or drive), drive the page, and assert on screenshots plus collected page errors. The operational trunk for browser work; specializations…
playwright-storybook-flows
Specialization of playwright-harness for AUTHORING and REPRODUCING flows. A flow is a sequence of steps a human or an agent can take inside an application to demonstrate a specific feature; it is written as a Storybook docs page in a conventional location, where a human reads it as a manual walkthrough (issue…