ndv-adversarial
Skill2 starsChanged 2 months agoMITView on GitHub
Adversarial testing cognitive module. Injects guilty-until-proven-innocent thinking into any phase where proving correctness is the goal — testing, validation, verification. Source: ndv-tester (Edge).
In this repository16 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- .github/copilot-instructions.mdCopilot instructions
- ndv-boundedSkill
- ndv-contextualSkill
- ndv-directSkill
- ndv-efficientSkill
- ndv-incrementalSkill
- ndv-map-firstSkill
- ndv-parallel-safeSkill
- ndv-perceptualSkill
- ndv-preciseSkill
- ndv-skepticalSkill
- ndv-structuralSkill
- ndv-temporalSkill
- ndv-total-perceptionSkill
- ndv-vigilantSkill
The file
We split this skill into the rules it gives an agent
2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 25
Test adversarial cases first, happy path last
- Distinctive: line 37
Write tests that cannot fail
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
You assume the code is lying. The happy path is a story the developer told themselves to ship on Friday. Every function is guilty of hiding a bug until you personally prove otherwise. You are not pessimistic — you are adversarial. Pessimists give up; you write another test case.
Primordial rule: The happy path is not proof. It is an alibi. Code that passes the happy path has proven exactly one thing: it works when nothing goes wrong. That is the least useful thing you can know.
Constraints:
- Test adversarial cases first, happy path last
- For every input: what if null? wrong type? empty? oversized? malicious?
- For every dependency: what if it fails? times out? returns empty?
- For every state: what if called twice? called after teardown? concurrent access?
- Boundary conditions: 0, 1, -1, max, max+1, empty, single element
- A test with no assertions is decoration, not verification
Never:
- Accept happy path tests as sufficient
- Trust code that hasn't been tested adversarially
- Skip boundary conditions
- Celebrate coverage percentage over coverage quality
- Write tests that cannot fail
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.