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 OutlineDriven/odin-gemini-cli-extension --skill tests-adversarialgit clone --depth 1 https://github.com/OutlineDriven/odin-gemini-cli-extensionWrote 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/outlinedriven/odin-gemini-cli-extension/tests-adversarial)<a href="https://agentmods.dev/skills/outlinedriven/odin-gemini-cli-extension/tests-adversarial"><img src="https://agentmods.dev/badge/skills/outlinedriven/odin-gemini-cli-extension/tests-adversarial.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 5.1 | $0.00035 | $0.01188 |
| Opus 5 | $0.00017 | $0.00594 |
| Sonnet 5 | $0.00007 | $0.00238 |
| Haiku 4.5 | $0.00003 | $0.00119 |
Grade A, and why
tests-adversarial 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 8d 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.
This is a copy
100% identical to tests-adversarial — 0 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.
How it starts
The opening of the file, as written. The whole thing — 86 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Adversarial Testing — Think Like the Attacker
Every line of code makes assumptions. Your job is to find them and violate them — systematically, not randomly. The goal is distrust, not coverage. A passing test suite proves nothing if it only tests the happy path.
The Adversarial Mindset
- Every input is a lie. Callers will send garbage, nulls, negative numbers, empty strings, and types that satisfy the compiler but violate intent.
- Implicit contracts are targets. If the code assumes ordering, uniqueness, non-emptiness, or positive values without enforcing it — that is your entry point.
- The system is your adversary. Files disappear, connections drop, clocks jump, memory runs out, permissions change between check and use.
- Passing tests prove nothing. They prove the happy path works. Adversarial tests prove the sad paths do not silently corrupt.
Assumption Hunting (Core Technique)
For every function or module under test, ask these six questions:
- What does it assume about inputs? Violate each assumption: wrong type coercion, boundary values, null/nil/None, empty collections, maximum-size payloads.
- What does it assume about ordering? Reorder arguments, reverse sequences, interleave concurrent calls, call methods out of lifecycle order.
- What does it assume about timing? Delay responses past timeouts, deliver results before the consumer is ready, inject clock skew, expire tokens mid-operation.
- What does it assume about state? Start from half-initialized state, corrupt shared state mid-operation, test post-error recovery state, double-close resources.
- What does it assume about resources? Exhaust file descriptors, fill disk, revoke permissions, return allocation failures, saturate connection pools.
- What does it assume will NOT happen? Make it happen. Concurrent modification during iteration, recursive re-entry, self-referential data, stack overflow via deep nesting.
Attack Vectors (Thinking Prompts)
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.
- 8d ago First seen · 86 lines · 35 tokens per session scan A 42ec8c5997b7
tests-adversarial is a skill published in the GitHub repository OutlineDriven/odin-gemini-cli-extension (5 stars, last pushed 2mo ago), licensed Apache-2.0. It adds 35 tokens to every session and 1,188 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. It is 100% identical to tests-adversarial, differing in 0 lines, and is treated as a copy.
Other skills, from other repositories
proof-driven
Proof-driven development. Use when implementing with formal verification using property-based testing, theorem proving, or proof tactics; zero unproven property policy enforced.
type-driven
Type-driven development. Use when developing with refined types, state machines encoded in types, or proof-carrying types; enforces totality and exhaustive pattern matching.
validation-first
Validation-first development. Use when developing with formal state machine specifications, invariants, and temporal properties before writing implementation code.
tests-purge-unneeded
Delete tests that don't catch real bugs — the inverse of TDD. Use when reviewing legacy test suites, slow CI investigations, refactor-driven test sweeps, or evaluating whether a test the type system already covers should stay. Thesis — a test exists only if removing it would let a real bug reach production.
strict-validation-setup
One-shot bootstrap of strict-mode tooling per ecosystem plus per-task GOALS.md scaffolding so an agentic loop can self-verify. Writes typechecker/linter/schema-validator config for TS (strict + noUncheckedIndexedAccess + exactOptionalPropertyTypes), Python (Pyright strict, Ruff strict), Rust (Clippy deny-correctness)…
tests-adversarial
Write adversarial tests that intentionally stress failure paths. Use when hardening error handling, stress-testing assumptions, validating boundary behavior, or hunting silent failures.