evaluate-card

name	evaluate-card
description	Evaluate Shadowverse card DSL behavior using eval-card.js scenario packets. Use when verifying card implementations, fixing DSL, or marking cards pass/fail in card-manifest.json.

Evaluate Card DSL

When to use

• User asks to verify a card works correctly
• After authoring or changing DSL in packages/sve-engine/data/card-defs/sets/<SET>.json
• During expansion rollout (BP01, BP02, …)

Workflow

1. Build engine (if needed):

   npm run build:engine
   

2. Run eval packet:

   node src/scripts/eval-card.js <cardNo>
   

   Output: eval-reports/<cardNo>.md plus stdout.

3. Read the packet:

   • Official card text vs resolved DSL
   • Scenario results (action trace, assertions, state diff)
   • Open questions (missing scenarios, unresolved choices)

4. Compare behavior to text using the embedded rubric:

   • Timings (fanfare, last words, evolve, activate, spell)
   • Costs, targets, conditions (including Necrocharge)
   • Optional branches and once-per-turn limits

5. If fail — fix DSL in the correct set file: packages/sve-engine/data/card-defs/sets/<SET>.json

6. If scenario gap — add YAML under: packages/sve-engine/scenarios/cards/<cardNo>-<slug>.yaml

7. Re-run until scenarios pass:

   npm run eval:scenarios
   node src/scripts/eval-card.js <cardNo>
   

8. Update manifest:

   • Edit packages/sve-engine/data/card-manifest.json for the card:
     • evalStatus: pass | fail | blocked | keyword-only
     • dslStatus: verified | authored | keyword-only | missing
     • blockers: engine primitives still needed
     • lastEvalAt: ISO date
   • Or regenerate base manifest: npm run generate:manifest

Scenario authoring

See packages/sve-engine/docs/dsl-authoring.md for DSL schema.

Example scenario actions:

• play: CARDNO — play from hand
• resolveTrigger: fanfare
• selectZone: { zone: cemetery, cardNo: CARDNO }
• activate: CARDNO
• engageFollowers: [ALLY1, ALLY2]

Batch commands

node src/scripts/eval-card.js --set BP01 --inventory
node src/scripts/parse-effects-to-dsl.js --set BP01
npm run coverage:cards

Verdicts

Verdict	Meaning
pass	Text matches DSL and scenarios
fail	DSL wrong; fix set file
blocked	Needs new engine primitive (add to manifest blockers)
keyword-only	No effects beyond printed keywords