fix: infra: agents-llama (local-Qwen dev agent) is hand-added to docker-compose.yml — move into lib/generators.sh as a flagged service (#769)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

docs/agents-llama.md

@@ -0,0 +1,42 @@
+# agents-llama — Local-Qwen Dev Agent
+
+The `agents-llama` service is an optional compose service that runs a dev agent
+backed by a local llama-server instance (e.g. Qwen) instead of the Anthropic
+API. It uses the same Docker image as the main `agents` service but connects to
+a local inference endpoint via `ANTHROPIC_BASE_URL`.
+
+## Enabling
+
+Set `ENABLE_LLAMA_AGENT=1` in `.env` (or `.env.enc`) and provide the required
+credentials:
+
+```env
+ENABLE_LLAMA_AGENT=1
+FORGE_TOKEN_LLAMA=<dev-qwen API token>
+FORGE_PASS_LLAMA=<dev-qwen password>
+ANTHROPIC_BASE_URL=http://host.docker.internal:8081   # llama-server endpoint
+```
+
+Then regenerate the compose file (`disinto init ...`) and bring the stack up.
+
+## Prerequisites
+
+- **llama-server** (or compatible OpenAI-API endpoint) running on the host,
+  reachable from inside Docker at the URL set in `ANTHROPIC_BASE_URL`.
+- A Forgejo bot user (e.g. `dev-qwen`) with its own API token and password,
+  stored as `FORGE_TOKEN_LLAMA` / `FORGE_PASS_LLAMA`.
+
+## Behaviour
+
+- `AGENT_ROLES=dev` — the llama agent only picks up dev work.
+- `CLAUDE_AUTOCOMPACT_PCT_OVERRIDE=60` — more aggressive compaction for smaller
+  context windows.
+- `depends_on: forgejo (service_healthy)` — does **not** depend on Woodpecker
+  (the llama agent doesn't need CI).
+- Serialises on the llama-server's single KV cache (AD-002).
+
+## Disabling
+
+Set `ENABLE_LLAMA_AGENT=0` (or leave it unset) and regenerate. The service
+block is omitted entirely from `docker-compose.yml`; the stack starts cleanly
+without it.