@lythos/skill-arena 0.9.32 → 0.9.36

package/README.md

@@ -49,26 +49,26 @@ Note: Claude `-p` mode has known issues with web tools in Bun.spawn (deferred to
 ```bash
 bun add -d @lythos/skill-arena
 # or use directly
-bunx @lythos/skill-arena@0.9.32 <command>
+bunx @lythos/skill-arena@0.9.36 <command>
 ```
 
 ## Quick Start
 
 ```bash
 # Mode 1: Compare two skills on the same task
-bunx @lythos/skill-arena@0.9.32 \
+bunx @lythos/skill-arena@0.9.36 \
   --task "Generate auth flow diagram" \
   --skills "design-doc-mermaid,mermaid-tools" \
   --criteria "syntax,context,token"
 
 # Mode 2: Compare full deck configurations
-bunx @lythos/skill-arena@0.9.32 \
+bunx @lythos/skill-arena@0.9.36 \
   --task "Generate auth flow diagram" \
   --decks "./decks/minimal.toml,./decks/rich.toml" \
   --criteria "quality,token,maintainability"
 
 # Visualize results
-bunx @lythos/skill-arena@0.9.32 viz tmp/arena-<id>/
+bunx @lythos/skill-arena@0.9.36 viz tmp/arena-<id>/
 ```
 
 ## Commands
@@ -77,16 +77,16 @@ bunx @lythos/skill-arena@0.9.32 viz tmp/arena-<id>/
 
 ```bash
 # Print execution plan without running
-bunx @lythos/skill-arena@0.9.32 run --config arena.toml --dry-run
+bunx @lythos/skill-arena@0.9.36 run --config arena.toml --dry-run
 
 # Execute with per-side runs_per_side and statistical aggregation
-bunx @lythos/skill-arena@0.9.32 run --config arena.toml
+bunx @lythos/skill-arena@0.9.36 run --config arena.toml
 ```
 
 ### CLI-flag mode (backward compat)
 
 ```
-bunx @lythos/skill-arena@0.9.32 run \
+bunx @lythos/skill-arena@0.9.36 run \
   --task ./TASK-arena.md \
   --players ./players/claude.toml \
   --decks ./decks/run-01.toml,./decks/run-02.toml \
@@ -96,13 +96,13 @@ bunx @lythos/skill-arena@0.9.32 run \
 ### Scaffold mode (legacy, manual execution)
 
 ```
-bunx @lythos/skill-arena@0.9.32 scaffold --task "..." --skills a,b
+bunx @lythos/skill-arena@0.9.36 scaffold --task "..." --skills a,b
 ```
 
 ### Viz
 
 ```bash
-bunx @lythos/skill-arena@0.9.32 viz runs/arena-<id>/
+bunx @lythos/skill-arena@0.9.36 viz runs/arena-<id>/
 ```
 
 ## Skill Documentation
@@ -116,7 +116,7 @@ The agent-visible **Skill** layer documentation is here:
 Part of the [lythoskill](https://github.com/lythos-labs/lythoskill) ecosystem — the thin-skill pattern separates heavy logic (this npm package) from lightweight agent instructions (SKILL.md).
 
 ```
-Starter (this package) → npm publish → bunx @lythos/skill-arena@0.9.32 ...
+Starter (this package) → npm publish → bunx @lythos/skill-arena@0.9.36 ...
 Skill   (packages/<name>/skill/)     → build → SKILL.md + thin scripts
 Output  (skills/<name>/)             → git commit → agent-visible skill
 ```

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@lythos/skill-arena",
-  "version": "0.9.32",
+  "version": "0.9.36",
   "description": "Skill Arena — benchmark skill effectiveness with controlled-variable comparison",
   "keywords": [
     "ai-agent",

package/src/runner.ts

@@ -3,8 +3,9 @@ import { join, resolve } from 'node:path'
 import { tmpdir } from 'node:os'
 import { runAgentScenario, type AgentScenario } from '@lythos/test-utils/agent-bdd'
 import { useAgent } from '@lythos/test-utils/agents'
-// Optional: register claude-sdk adapter if the package is installed
+// Optional: register heavy adapters if their packages are installed
 try { await import('@lythos/agent-adapter-claude-sdk') } catch { /* package not installed */ }
+try { await import('@lythos/agent-adapter-deepseek-serve') } catch { /* package not installed */ }
 import { ArenaManifest, Player } from '@lythos/test-utils/schema'
 import type { ArenaManifest as ArenaManifestType, JudgeVerdict } from '@lythos/test-utils/schema'
 import { runComparativeJudge } from './comparative-judge'