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/fergius-engineering/instincts/tests-with-teethnpx skills add Fergius-Engineering/instincts --skill tests-with-teethgit clone --depth 1 https://github.com/Fergius-Engineering/instinctsWrote 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/fergius-engineering/instincts/tests-with-teeth)<a href="https://agentmods.dev/skills/fergius-engineering/instincts/tests-with-teeth"><img src="https://agentmods.dev/badge/skills/fergius-engineering/instincts/tests-with-teeth.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 | $0.00025 | $0.00637 |
| Opus 5 | $0.00013 | $0.00318 |
| Sonnet 5 | $0.00005 | $0.00127 |
| Haiku 4.5 | $0.00003 | $0.00064 |
Grade A, and why
tests-with-teeth 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 4d 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 — 44 lines — stays where its author put it; the contents beside it link to each section on GitHub.
The rule
A test exists to fail when the thing it covers breaks. If you could delete the feature and the test stays green, it protects nothing. It's worse than no test, because it looks like coverage. Before accepting any test, run it through five questions.
This sharpens superpowers' test-driven-development: TDD's red step proves a brand-new test can fail; these questions hold for any test, including ones you didn't write.
Fires when
Writing a new test, reviewing a test in a diff, inheriting a suite, or about to treat a green run as proof that something works.
The five questions
- Removal — mentally delete the line the test targets. Does the test now fail? If no, it's hollow.
- Reach — list every guard and early return between setup and the assertion. Does the setup trip one of them first, so execution never reaches the target? Then the assertion is dead.
- Distinguishability — would the test pass for a wrong implementation too? Give two inputs different values so the test can only pass if the right one was used.
- Environment gap — what does the real feature rely on that the test harness lacks (a real database, say, or a network)? If the gap means the real path can't run, test at the lowest level that does run and label what it does and doesn't prove.
- Algorithm gap — is the novel logic (a sort, a filter, a dedup) tested directly? An end-to-end test can produce the right final output through a compensating bug. Test the tricky function on its own with inputs that would expose a wrong algorithm.
Worked example
You're testing a discount function. The test: applyDiscount(cart) returns a number >= 0. It passes. It's hollow. It passes whether the discount is 10%, 0%, or the function just returns the original total. Now make it distinguish: a $100 cart with a 10% code should return exactly $90. Delete the discount math and it returns $100, and the test fails. That version has teeth.
Hollow-test smells
| Test smell | Why it's hollow |
|---|---|
| Asserts "does not crash" | Not crashing isn't correctness. |
| Asserts a value >= 0 / not null | Passes for almost any implementation. |
| Setup uses an input that triggers an early return | The assertion is never reached. |
| Same value passed for two different inputs | Can't tell which one the code used. |
| Only an end-to-end path, novel logic never tested alone | A compensating bug can hide. |
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.
- 4d ago First seen · 44 lines · 25 tokens per session scan A 961b30daff81
tests-with-teeth is a skill published in the GitHub repository Fergius-Engineering/instincts (2 stars, last pushed 10d ago), licensed MIT. It adds 25 tokens to every session and 637 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 2026-08-31.
Other skills, from other repositories
rulesync
Generates and syncs AI rule configuration files (.cursorrules, CLAUDE.md, copilot-instructions.md) across 20+ coding tools from a single source. Use when syncing AI rules, running rulesync commands, importing or generating rule files, or managing shared AI coding configurations.
establishing-project-context
Use when the user asks to establish shared project language, or project work exposes a conflicting, renamed, or deprecated domain term that needs active semantic modeling. Routine small tasks stay on the fast path.
autoprompt
Explicit-only useful-first orchestration. Invoke /autoprompt to turn a mission into one executable roadmap, build dependency-safe lanes, and verify the result with independent reviewers. Never infer invocation from ordinary requests. Never resume from leftover artifacts without an explicit resume instruction.
loongsuite-pilot-insight
基于 LoongSuite Pilot / AI Coding Agent 日志生成事件洞察、组织洞察、数据质量、研发效能和 AI Native 使用类 SLS 报表时使用；包含 AI Coding 事件表语义，以及团队报表可选的部门维表、deptuser 组织关系、指标口径和公共 CTE，通常与 sls-dashboard-builder 一起使用。.
map-fast
Minimal workflow for small, low-risk changes — no planning, no learning.
alipay-webhooks
Receive and verify Alipay (Antom / Alipay+) webhook notifications. Use when setting up Alipay webhook handlers, debugging RSA256 Signature header verification, or handling payment events like notifyPayment, notifyCapture, notifyRefund, notifyAuthorization, and notifyDispute.