# Skill case study

Replace every placeholder with observed facts before publishing. An untested
step stays marked "not tested". A registry listing or metadata score is not a
successful run. Never include credentials, private input, or other people's data.

## Task and input

- User task:
- Input file or public fixture (include license and source):
- Expected output and acceptance criteria:

## Reproduction

- Skill author and repository:
- Skill source file and immutable commit:
- Agent and version:
- Operating system and relevant dependencies:
- Date tested (UTC):
- Required accounts, permissions, cost, and configuration:
- Exact installation steps actually performed:
- Prompt and execution steps:

## Observed result

- Outcome: passed / failed / partial / not tested
- Output artifact or public evidence link:
- Which acceptance criteria were met:
- Failed steps and limitations:
- Measured time or cost (state the measurement method):
- Reproduction notes:

## Sources and attribution

- Original author and project documentation:
- OpenAgentSkill listing, if useful to readers:
- Other dependencies and licenses:

The author chooses whether and how to cite OpenAgentSkill. A link is never a
condition of being listed, reviewed, or receiving a score. Use a normal text
citation where it helps readers; do not claim certification from a listing.
