Define what counts as a correct result for each test so outcomes are observable, comparable, and not left to intuition.
Trigger this skill when
You need disciplined test design rather than ad hoc case generation.
You need to justify why a test set is sufficient.
You want a reusable artifact that can feed automation later.
Expected inputs
requirements
test cases
business rules
data expectations
Deliverables
test oracles
expected outputs
observable checks
oracle assumptions
Operating procedure
For each test, define the observable correct outcome.
Specify exact values, state changes, error messages, side effects, or invariants as relevant.
Avoid vague expectations like “works” or “should be okay.”
Record any assumptions where the spec is silent.
Quality gates
Be explicit about infeasible, unreachable, or assumption-heavy cases.
Prefer lean, justified test sets over exhaustive-looking noise.
Keep expected outcomes observable and tied back to rules or logic.
Show why each test exists, not just the test data.
Handoff targets
negative-test-designer
traceability packs later
automation-testing packs later
Output style
Be explicit about uncertainty, infeasibility, and residual risk.
Prefer compact, explainable artifacts over bloated lists.
Tie tests back to rules, logic, or structure.
Make expected outcomes observable.
Failure modes to avoid
Do not leave expected results implicit.
Do not use subjective pass/fail language.
Do not ignore side effects or persistent-state changes.
Minimum output skeleton
## Summary
## Findings
## Structured outputs
## Coverage / rationale
## Assumptions
## Open questions
## Recommended next skill
1---2name: test-oracle-writer3description: test-oracle-writer4---5# test-oracle-writer67## Purpose8Define what counts as a correct result for each test so outcomes are observable, comparable, and not left to intuition.910## Trigger this skill when11- You need disciplined test design rather than ad hoc case generation.12- You need to justify why a test set is sufficient.13- You want a reusable artifact that can feed automation later.1415## Expected inputs16- requirements17- test cases18- business rules19- data expectations2021## Deliverables22- test oracles23- expected outputs24- observable checks25- oracle assumptions2627## Operating procedure281. For each test, define the observable correct outcome.292. Specify exact values, state changes, error messages, side effects, or invariants as relevant.303. Avoid vague expectations like “works” or “should be okay.”314. Record any assumptions where the spec is silent.3233## Quality gates34- Be explicit about infeasible, unreachable, or assumption-heavy cases.35- Prefer lean, justified test sets over exhaustive-looking noise.36- Keep expected outcomes observable and tied back to rules or logic.37- Show why each test exists, not just the test data.3839## Handoff targets40- negative-test-designer41- traceability packs later42- automation-testing packs later4344## Output style45- Be explicit about uncertainty, infeasibility, and residual risk.46- Prefer compact, explainable artifacts over bloated lists.47- Tie tests back to rules, logic, or structure.48- Make expected outcomes observable.4950## Failure modes to avoid51- Do not leave expected results implicit.52- Do not use subjective pass/fail language.53- Do not ignore side effects or persistent-state changes.5455## Minimum output skeleton56```md57## Summary58## Findings59## Structured outputs60## Coverage / rationale61## Assumptions62## Open questions63## Recommended next skill64```
Run npx skillmds@latest add 45ck/test-oracle-writer in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.
test-oracle-writer It is listed under Coding & Dev Tools on SkillMD.
This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.
This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.
Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.
45ck (@45ck) published this skill. Their other Agent Skills are listed on their SkillMD profile.