zeus: Skill for Claude Code

Zeus Test Gen — AI Test Flow Generation

Generate platform-specific end-to-end test scenarios from Zeus planning artifacts. AI (zeus-tester agent) authors the test cases. Humans do not write scenarios manually.

Preconditions

• .zeus/{version}/task.json exists and has at least one task.
• .zeus/{version}/prd.json exists and has at least one story.
• .zeus/schemas/test-flow.schema.json exists.
• .zeus/{version}/config.json exists.
• generate-tests.sh is executable.
• claude CLI is installed and reachable.

Optional arguments

• --version vN — target Zeus version folder (default: main)
• --platforms android,chrome,ios — subset of platforms to generate (default: all three)

Deterministic workflow

1) Check preconditions

Verify all required files exist. If any is missing, print a clear error and abort.

for f in .zeus/{version}/task.json .zeus/{version}/prd.json .zeus/schemas/test-flow.schema.json; do
  [[ -f "$f" ]] || { echo "❌ Missing: $f"; exit 1; }
done

2) Show generation plan

Print a preview before executing:

Platform(s) to generate : android, chrome, ios
Task source             : .zeus/{version}/task.json  (N tasks)
Story source            : .zeus/{version}/prd.json   (N stories)
Output directory        : .zeus/{version}/tests/

Require user confirmation to proceed (yes/no).

3) Run generate-tests.sh

bash .zeus/scripts/generate-tests.sh \
  --version {version} \
  --platforms {platforms}

Stream output to the user in real time.

4) Validate generated files

For each output file:

jq empty .zeus/{version}/tests/{platform}.test.json && echo "✓ valid JSON"

Report scenario counts per platform.

5) Print coverage report

After generation succeeds, output a coverage table:

Platform  | Scenarios | Tasks covered | Coverage
----------|-----------|---------------|----------
android   |         8 |           5/6 |      83%
chrome    |         6 |           5/6 |      83%
ios       |         7 |           5/6 |      83%