feat: add local-agent.sh wrapper + local concurrency config

Co-Authored-By: Virgil <virgil@lethean.io>

config/agents.yaml

@@ -14,6 +14,7 @@ concurrency:
   claude: 1
   gemini: 1
   codex: 1
+  local: 1
 
 # Rate limiting / quota management
 # Controls pacing between task dispatches to stay within daily quotas.

scripts/local-agent.sh

@@ -0,0 +1,109 @@
+#!/bin/bash
+# Local agent wrapper — runs Ollama model on workspace files
+# Usage: local-agent.sh <prompt>
+#
+# Reads PROMPT.md, CLAUDE.md, TODO.md, PLAN.md from current directory,
+# combines them into a single prompt, sends to Ollama, outputs result.
+
+set -e
+
+PROMPT="$1"
+MODEL="${LOCAL_MODEL:-hf.co/unsloth/Qwen3-Coder-Next-GGUF:UD-IQ4_NL}"
+CTX_SIZE="${LOCAL_CTX:-16384}"
+
+# Build context from workspace files
+CONTEXT=""
+
+if [ -f "CLAUDE.md" ]; then
+    CONTEXT="${CONTEXT}
+
+=== PROJECT CONVENTIONS (CLAUDE.md) ===
+$(cat CLAUDE.md)
+"
+fi
+
+if [ -f "PLAN.md" ]; then
+    CONTEXT="${CONTEXT}
+
+=== WORK PLAN (PLAN.md) ===
+$(cat PLAN.md)
+"
+fi
+
+if [ -f "TODO.md" ]; then
+    CONTEXT="${CONTEXT}
+
+=== TASK (TODO.md) ===
+$(cat TODO.md)
+"
+fi
+
+if [ -f "CONTEXT.md" ]; then
+    CONTEXT="${CONTEXT}
+
+=== PRIOR KNOWLEDGE (CONTEXT.md) ===
+$(head -200 CONTEXT.md)
+"
+fi
+
+if [ -f "CONSUMERS.md" ]; then
+    CONTEXT="${CONTEXT}
+
+=== CONSUMERS (CONSUMERS.md) ===
+$(cat CONSUMERS.md)
+"
+fi
+
+if [ -f "RECENT.md" ]; then
+    CONTEXT="${CONTEXT}
+
+=== RECENT CHANGES (RECENT.md) ===
+$(cat RECENT.md)
+"
+fi
+
+# List all source files for the model to review
+FILES=""
+if [ -d "." ]; then
+    FILES=$(find . -name "*.go" -o -name "*.php" -o -name "*.ts" | grep -v vendor | grep -v node_modules | grep -v ".git" | sort)
+fi
+
+# Build the full prompt
+FULL_PROMPT="${CONTEXT}
+
+=== INSTRUCTIONS ===
+${PROMPT}
+
+=== SOURCE FILES IN THIS REPO ===
+${FILES}
+
+Review each source file listed above. Read them one at a time and report your findings.
+For each file, use: cat <filename> to read it, then analyse it according to the instructions.
+"
+
+# Call Ollama API (non-streaming for clean output)
+RESPONSE=$(curl -s http://localhost:11434/api/generate \
+    -d "$(python3 -c "
+import json
+print(json.dumps({
+    'model': '${MODEL}',
+    'prompt': $(python3 -c "import json,sys; print(json.dumps(sys.stdin.read()))" <<< "$FULL_PROMPT"),
+    'stream': False,
+    'options': {
+        'temperature': 0.1,
+        'num_ctx': ${CTX_SIZE},
+        'top_p': 0.95,
+        'top_k': 40
+    }
+}))
+")" 2>/dev/null)
+
+# Extract and output the response
+echo "$RESPONSE" | python3 -c "
+import json, sys
+try:
+    d = json.load(sys.stdin)
+    print(d.get('response', 'Error: no response'))
+except:
+    print('Error: failed to parse response')
+"