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 Netcracker/qubership-ai-packages --skill blackbox-test-designgit clone --depth 1 https://github.com/Netcracker/qubership-ai-packagesWrote 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/netcracker/qubership-ai-packages/blackbox-test-design)<a href="https://agentmods.dev/skills/netcracker/qubership-ai-packages/blackbox-test-design"><img src="https://agentmods.dev/badge/skills/netcracker/qubership-ai-packages/blackbox-test-design/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/netcracker/qubership-ai-packages/blackbox-test-design"><img src="https://agentmods.dev/badge/skills/netcracker/qubership-ai-packages/blackbox-test-design.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 | $0.00195 | $0.06294 |
| Opus 5.5 | $0.00078 | $0.02518 |
| Sonnet 5.5 | $0.00039 | $0.01259 |
| Haiku 4.5 | $0.00019 | $0.00629 |
Grade A, and why
blackbox-test-design 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 10d 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 — 378 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Designing questions for a black box
Use this skill when the behavior you need is owned by something you cannot read: a legacy service, a vendor binary, a reference implementation behind a network boundary, a stand you can only send requests to. You hold a model of it (an executable spec, a port, a mock, a mental model in a design document) and a pile of recorded answers (goldens, snapshots, logs). Every answer costs something: a stand run, a person's time, a release cycle. The job is to spend those answers on the questions most likely to show that the model is wrong.
This differs from ordinary test authoring in one way that changes everything: the expected value is unknown. You cannot assert it, so a test here is a question, and its value is how many wrong models its answer rules out. A question whose answer every live model predicts the same way is wasted, however thorough it looks.
This skill decides which questions to ask. Turning a question into a test, with its level, its assertion, and its
failure message, is test-authoring's subject, and so is a harness or comparator once its behavior is decided; load
that skill too before writing any of them. Where the expected value can be stated, from a specification or from code
you can read, the task is ordinary test authoring and test-authoring alone applies.
The sections below are techniques, ordered roughly by cost. Most projects need several of them, because each one is blind in a place another one sees.
0. Before anything: state the model and its provenance
Write down, or find, the model's decisions as a list: each place where the reference could have behaved in more than one way, the choice the model made, and the recorded answer that forced it (or "inferred" when none did). This list is what every technique below attacks. A decision with no answer behind it is a guess, and guesses are where the model is most likely wrong.
Settle the goal first: parity (reproduce the reference, quirks included) or intended behavior (reproduce what the reference was meant to do, and list the quirks as deliberate differences). The same answer is a bug to copy under one goal and a bug to fix under the other, and the questions worth asking differ.
What ships with it
1 file 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.
- 10d ago First seen · 378 lines · 195 tokens per session scan A cd36677c8452
blackbox-test-design is a skill published in the GitHub repository Netcracker/qubership-ai-packages (4 stars, last pushed today), licensed Apache-2.0. It adds 195 tokens to every session and 6,294 once invoked, about $0.0008 per session on Opus 5.5. 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 2026-10-08.
Other skills, from other repositories
research-engineer
An uncompromising Academic Research Engineer. Operates with absolute scientific rigor, objective criticism, and zero flair. Focuses on theoretical correctness, formal verification, and optimal implementation across any required technology.
eval-dataset-design
A guide to creating reliable test sets for AI agents. Each evaluation task includes the request, the simulated user's behaviour, a reset starting state, and a success check that can be independently verified.
tika-eval-compare
Compare extracts from two Tika builds over a corpus to detect regressions in content, encoding, exceptions, and embedded-document handling. Use for "compare before/after extracts", "eval this change against the corpus".
cua-driver
Use Cua Driver for desktop or browser tasks that are awkward or unavailable through Bash/APIs, or when the user explicitly wants GUI interaction: app testing, visual bug reproduction, form filling, calendar entry, screenshots, and demo recording. Also covers Cua setup; not OpenAI Codex Computer Use or web research.
jetson-validate-image
Use after jetson-flash-image to run static BSP checks, on-target smoke/regression tests on a flashed DUT, or both. Not for build or flash steps. Triggers: validate bsp, on-target validation.
atmos-validation
Validate Atmos projects, components, arbitrary JSON Schema inputs, EditorConfig, and GitHub Actions; use affected-file selection and native CI annotations.