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 nebius/nebius-physical-ai --skill guardrail-failuresgit clone --depth 1 https://github.com/nebius/nebius-physical-aiWrote 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/nebius/nebius-physical-ai/guardrail-failures)<a href="https://agentmods.dev/skills/nebius/nebius-physical-ai/guardrail-failures"><img src="https://agentmods.dev/badge/skills/nebius/nebius-physical-ai/guardrail-failures/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/nebius/nebius-physical-ai/guardrail-failures"><img src="https://agentmods.dev/badge/skills/nebius/nebius-physical-ai/guardrail-failures.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 Privilege Escalation · line 93 Commands invoke sudo or root privileges. Verify this elevated access is necessary and justified.Fix: Avoid sudo/root unless strictly required. Prefer least-privilege patterns. If elevation is needed, document the justification and scope.
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 | $0.00039 | $0.02774 |
| Opus 5 | $0.00019 | $0.01387 |
| Sonnet 5 | $0.00008 | $0.00555 |
| Haiku 4.5 | $0.00004 | $0.00277 |
Grade B, and why
guardrail-failures scanned grade B 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 yesterday.
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.
Asks for rootmediumPrivilege escalation
A mod that escalates privileges can change anything on the machine, not only the project.
| `test_workbench_image_k8s_prereqs` | An image lacks what SkyPilot needs on Kubernetes (python3, rsync, sudo with NOPASSWD, or the Isaac group/PATH rules). Update the Dockerfile and the shared install script together. | How it starts
The opening of the file, as written. The whole thing — 135 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Decoding A Guardrail Failure
The guardrail suite is ~2000 assertions over 43 files and runs in under a minute. It is static: it reads the repo and checks that surfaces which must agree still agree. A failure almost always means one of two things — you changed one side of a contract and not the other, or you added something the contract requires you to register.
Run the whole suite; it is cheap enough that narrowing is rarely worth it:
npa/.venv/bin/python -m pytest npa/tests/guardrails -q
Narrow once you know the file:
npa/.venv/bin/python -m pytest npa/tests/guardrails/test_tool_catalog_argv.py -q
Most of these tests carry a module docstring explaining the incident that caused them to exist. When the fix is not obvious, read it — it usually names the production failure the check prevents.
Triage Order
- Read the assertion message. These tests are written to name the fix, not just report a mismatch.
- Identify which side of the contract you changed. The guard is almost never wrong; it is reporting that a second file needs the same edit.
- Only weaken a pinned exemption list if the exemption is genuinely obsolete. Several lists are explicitly designed to shrink and never grow.
CLI, Catalog, Spec, And SDK Coherence
These fire when a tool's surfaces drift apart. This is the group a workbench change hits most.
| Guardrail | Fix when it fails |
|---|---|
test_tool_catalog_argv |
A toolRef argv names a CLI option that does not exist, or passes a value the option cannot mean. Correct the argv in catalog.py — see skills/atomic/toolref-argv-contract/SKILL.md. |
test_module_toolref_argv |
A python -m toolRef no longer parses against its module's own argparse. Sync the module parser and the catalog argv. |
test_three_tier_contract |
A new CLI group has neither a CapabilityContract nor a seam entry, or a CLI parameter is unreachable from a spec without a stated reason. Add the contract or record the spec_gap category. |
test_spec_declared_outputs |
A spec's declared outputs: path is not what the tool writes. Fix the spec, or the tool's result_uri_for() helper. |
test_spec_paths_are_not_repo_relative |
A stage argv points at a path that exists in this checkout. Stages run in a pod; use an s3:// URI or an in-image absolute path. |
test_shown_workflow_catalog |
A catalog entry no shipped spec reaches, or a spec outside the npa.workflow catalog. Add a reference spec or list the entry in PUBLIC_REUSABLE_TOOLREFS. |
test_workflow_schema_chaining |
A state depends on another but reads a schema that one does not produce. Align the producing and consuming stages. |
test_typer_command_calls |
A Typer command is called as a plain function without @resolve_typer_defaults. Add the decorator to the callee. |
test_internal_cli_entrypoint |
An internal argv no longer matches python -m npa. Build it with internal_cli_argv(). |
test_help_to_markdown |
The generated-docs converter mishandles wrapped Rich help. Fix scripts/_help_to_markdown.py. |
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.
- yesterday Changed · +2 lines 7f2e8ea5b156
- 4d ago Changed · +3 lines 3c6c447b42d1
- 7d ago Changed · +1 lines b35525310663
- 12d ago First seen · 129 lines · 39 tokens per session scan B 4cba2d5f7737
guardrail-failures is a skill published in the GitHub repository nebius/nebius-physical-ai (29 stars, last pushed today), licensed Apache-2.0. It adds 39 tokens to every session and 2,774 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it B with 1 finding (asks for root). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
systematic-debugging
4-phase root cause debugging: understand bugs before fixing.
langsmith-observability
LLM observability platform for tracing, evaluation, and monitoring. Use when debugging LLM applications, evaluating model outputs against datasets, monitoring production systems, or building systematic testing pipelines for AI applications.
experimental-code-coverage-local-debugger
Runs code coverage locally via Universal Test Runner (UTR) or helper scripts, mimicking LUCI trybots. Activate when CQ tryjobs fail or underreport coverage, to test local GN/recipe repairs before uploading, or to debug hermetic crashes.
adversarial-reviewer
Adversarial code review that assumes bugs exist and hunts for them. Use when asked to review code, find bugs, audit for correctness, stress-test a PR, or when someone says "tear this apart" or "what's wrong with this". Give no benefit of the doubt — every line is guilty until proven innocent.
cli-e2e
Write, modify, or debug Docker-based Composio CLI end-to-end tests under ts/e2e-tests/cli, including binary invocation, fixture isolation, output assertions, and package manifests. Use for CLI E2E test suites only; use cli-command for CLI source implementation.
work
Deliver one maintainer-approved EmDash issue, choosing the bug-fix path for a defect and the direct implementation path for an enhancement or task.