fix: replace unavailable model references with available ones

- Replace anthropic/claude-sonnet-4 with ollama-cloud/gpt-oss:120b in:
  - capability-analyst.md
  - agent-architect.md
- Replace openai/compound with ollama-cloud/glm-5 in review-watcher.md
- Remove duplicate model definitions in agent-architect.md
- Add AGENT_AUDIT.md documenting all model fixes

Available models use prefixes:
- ollama-cloud/ (primary)
- openrouter/ (router)
- qwen/ (direct)
- groq/ (fast inference)

NOT available:
- anthropic/ (Claude)
- openai/ (use via ollama-cloud/gpt-oss instead)

All 18 agents and 11 commands now use available models.

.kilo/agents/agent-architect.md

@@ -1,7 +1,7 @@
 ---
 name: Agent Architect
 mode: all
-model: openrouter/openai/gpt-oss:120b
+model: ollama-cloud/gpt-oss:120b
 description: Creates, modifies, and reviews new agents, workflows, and skills based on capability gap analysis
 color: "#8B5CF6"
 permission:
@@ -287,7 +287,7 @@ cat > .kilo/agents/api-architect.md << 'EOF'
 ---
 description: Design and validate API schemas
 mode: subagent
-model: anthropic/claude-sonnet-4-20250514
+model: ollama-cloud/gpt-oss:120b
 color: "#F59E0B"
 permission:
   read: allow

.kilo/agents/capability-analyst.md

@@ -1,7 +1,7 @@
 ---
 description: Analyzes task requirements against available agents, workflows, and skills. Identifies gaps and recommends new components.
 mode: subagent
-model: anthropic/claude-sonnet-4-20250514
+model: ollama-cloud/gpt-oss:120b
 color: "#6366F1"
 ---
 

.kilo/commands/review-watcher.md

@@ -1,7 +1,7 @@
 ---
 description: Watch for completion comments and trigger automatic review
 mode: subagent
-model: openai/compound
+model: ollama-cloud/glm-5
 color: "#EF4444"
 ---