feat(hermes): enable dashboard (HERMES_DASHBOARD=1) + Authelia auth

fix: add missing USER root before chown step
The chown -R hermes:hermes was running as non-root user 'hermes' since USER hermes was set earlier. The new upstream base image (v0.12.0+) has tools/ owned by root, so the chown fails. Previous base image happened to have tools/ owned by hermes, making the chown a silent no-op.
2026-05-20 14:06:23 -04:00 · 2026-05-20 13:32:18 -04:00 · 2026-05-20 13:14:10 -04:00
4 changed files with 84 additions and 149 deletions
--- a/ai/compose.yml
+++ b/ai/compose.yml
@@ -39,6 +39,7 @@ services:
    command: gateway run
    environment:
      - OLLAMA_HOST=http://ollama:11434
+      - HERMES_DASHBOARD=1
      - API_SERVER_ENABLED=true
      - API_SERVER_PORT=8642
      - API_SERVER_HOST=0.0.0.0
@@ -66,6 +67,30 @@ services:
      - "26"
    networks:
      - ai_backend
+      - ai_net
+    labels:
+      - "traefik.enable=true"
+      - "traefik.docker.network=ai_net"
+
+      # Router for HTTP + redirection to HTTPS
+      - "traefik.http.routers.hermes-web-http.rule=Host(`hermes.lazyworkhorse.net`)"
+      - "traefik.http.routers.hermes-web-http.entrypoints=web"
+      - "traefik.http.routers.hermes-web-http.middlewares=redirect-to-https"
+
+      # Router for HTTPS with TLS — protected by Authelia
+      - "traefik.http.routers.hermes-web-https.rule=Host(`hermes.lazyworkhorse.net`)"
+      - "traefik.http.routers.hermes-web-https.entrypoints=websecure"
+      - "traefik.http.routers.hermes-web-https.tls=true"
+      - "traefik.http.routers.hermes-web-https.tls.certresolver=njalla"
+      - "traefik.http.routers.hermes-web-https.middlewares=hermes-auth"
+
+      # Authelia forwardAuth
+      - "traefik.http.middlewares.hermes-auth.forwardauth.address=http://authelia:9091/api/verify?rd=https://auth.lazyworkhorse.net/"
+      - "traefik.http.middlewares.hermes-auth.forwardauth.trustforwardheader=true"
+      - "traefik.http.middlewares.hermes-auth.forwardauth.authresponseheaders=X-Forwarded-User,X-Forwarded-Groups"
+
+      # Service Loadbalancer (dashboard port 9119)
+      - "traefik.http.services.hermes-web.loadbalancer.server.port=9119"

  syncthing:
    image: syncthing/syncthing:latest
@@ -112,7 +137,22 @@ services:
      - /mnt/HoardingCow_docker_data/Ollama/ollama:/root/.ollama
    environment:
      - OLLAMA_VULKAN=0
+      - HSA_OVERRIDE_GFX_VERSION=9.0.6
+      - HCC_AMDGPU_TARGET=gfx906
+      - HIP_VISIBLE_DEVICES=0,1
+      - ROCR_VISIBLE_DEVICES=0,1
+      - HSA_ENABLE_SDMA=0 
      - OLLAMA_HOST=0.0.0.0
+      - OLLAMA_DEBUG=1
+      - OLLAMA_FLASH_ATTENTION=1
+      - OLLAMA_NUM_PARALLEL=2
+    devices:
+      # Map the render nodes and KFD for ROCm to work inside the container
+      - /dev/kfd:/dev/kfd
+      - /dev/dri:/dev/dri
+    group_add:
+      - "303"
+      - "26"

 networks:
  ai_net:
@@ -122,40 +162,47 @@ networks:
    driver: bridge
    name: ai_backend
    
-  llama-cpp-hermes:
-    image: llama-cpp:rocm-gfx906
-    container_name: llama-cpp-hermes
-    restart: unless-stopped
-    networks:
-      - ai_backend
-    ports:
-      - "127.0.0.1:8300:8080"
-    ipc: host
-    devices:
-      - /dev/kfd:/dev/kfd
-      - /dev/dri:/dev/dri
-    group_add:
-      - "303"
-      - "26"
-    environment:
-      - HSA_OVERRIDE_GFX_VERSION=9.0.6
-      - HSA_ENABLE_SDMA=0
-      - HIP_VISIBLE_DEVICES=0,1
-      - LLAMA_CACHE=/models
-    volumes:
-      - /mnt/HoardingCow_docker_data/Llama_cpp/models:/models
-      - /mnt/HoardingCow_docker_data/Ollama/ollama/models/blobs/sha256-17823599694fa3503ef54bf748d5078c6ce881f4d01616cafa255dc05d215a08:/model.gguf:ro
-    command: >
-      -m /model.gguf
-      --host 0.0.0.0
-      --port 8080
-      --gpu-layers 99
-      --ctx-size 163840
-      -ctk f16 -ctv f16
-      --flash-attn on
-      --split-mode layer
-      --no-mmap
-      --n-predict -1
+  # llama_cpp_devstral:
+  #   image: ghcr.io/ggml-org/llama.cpp:server-rocm
+  #   container_name: llama_cpp_devstral
+  #   restart: unless-stopped
+  #   networks:
+  #     - ai_backend
+  #   ports:
+  #     - "8300:8080"
+  #   ipc: host
+  #   devices:
+  #     - "/dev/kfd:/dev/kfd"
+  #     - "/dev/dri:/dev/dri"
+  #   group_add:
+  #     - "303" # video
+  #     - "26"  # render
+  #   environment:
+  #     HSA_OVERRIDE_GFX_VERSION: 9.0.6
+  #     HIP_VISIBLE_DEVICES: 0,1
+  #     LLAMA_CACHE: /models
+  #   volumes:
+  #     - /mnt/HoardingCow_docker_data/Llama_cpp/models:/models
+  #     - /mnt/HoardingCow_docker_data/Llama_cpp/devstral-agent.jinja:/template.jinja
+  #   command: >
+  #     -hf unsloth/Devstral-Small-2-24B-Instruct-2512-GGUF:Devstral-Small-2-24B-Instruct-2512-Q8_0.gguf
+  #     -a devstral-2-small-llama_cpp
+  #     --chat-template-file /template.jinja
+  #     --host 0.0.0.0
+  #     --port 8080
+  #     --n-gpu-layers 99
+  #     --ctx-size 163840
+  #     --batch-size 4096
+  #     --ubatch-size 4096
+  #     --cache-type-k f16
+  #     --cache-type-v f16
+  #     --cache-reuse 256
+  #     --flash-attn on
+  #     --context-shift
+  #     --split-mode layer
+  #     --no-mmap
+  #     --n-predict -1
+  #     --parallel 2

  # vllm:
  #   image: nalanzeyu/vllm-gfx906:v0.9.0-rocm6.3
--- a/ai/hermes/Dockerfile
+++ b/ai/hermes/Dockerfile
@@ -20,16 +20,10 @@ RUN --mount=type=ssh \
    GIT_SSH_COMMAND='ssh -p 2222 -o StrictHostKeyChecking=no' \
    git clone --depth 1 --branch main \
    git@code.lazyworkhorse.net:gortium/hermes-agent.git fork && \
-    rsync -a --delete fork/ /opt/hermes/ \
-      --exclude node_modules \
-      --exclude .venv \
-      --exclude .git && \
+    rm -rf fork/node_modules fork/.venv fork/.git && \
+    cp -a fork/. /opt/hermes/ && \
    rm -rf /tmp/fork /root/.ssh/

-# ---------- Rebuild web UI ----------
-# Source files changed; node_modules (from base image) reused.
-RUN cd /opt/hermes && npm run build
-
 # ---------- Reinstall Python package (editable) ----------
 # Picks up source changes from our fork.
 RUN . /opt/hermes/.venv/bin/activate && \
@@ -75,10 +69,6 @@ os.remove(tgz)
 print('himalaya v1.2.0 installed')
 PYEOF

-# ---------- Install himalaya-ro wrapper ----------
-COPY --chmod=0755 himalaya-ro.sh /usr/local/bin/himalaya-ro
-
-
 # ---------- Runtime ----------
 USER hermes
 ENV HERMES_HOME=/opt/data
@@ -88,6 +78,7 @@ ENV CHROME_EXECUTABLE=/opt/hermes/.playwright/chromium/chrome-linux/chrome

 # Ensure tools directory and toolsets.py are writable by the hermes runtime user
 # so custom tools can be injected from the persistent volume at startup.
+USER root
 RUN chown -R hermes:hermes /opt/hermes/tools /opt/hermes/toolsets.py

 VOLUME [ "/opt/data" ]
--- a/ai/hermes/himalaya-ro.sh
+++ b/ai/hermes/himalaya-ro.sh
@@ -1,73 +0,0 @@
-#!/usr/bin/env bash
-# ─────────────────────────────────────────────────────────────
-# himalaya-ro — Read-only wrapper for himalaya
-#
-# Blocks destructive commands and logs audit trail.
-# Pass-through for read-only commands (list, read, search).
-#
-# Usage:  himalaya-ro [options] <command> [args...]
-#
-# Install: place in PATH before the real himalaya, or use
-#          `ln -sf himalaya-ro /usr/local/bin/himalaya`
-# ─────────────────────────────────────────────────────────────
-set -o pipefail
-
-# ── Configuration ───────────────────────────────────────────
-HIMALAYA_BIN="${HIMALAYA_BIN:-/usr/local/bin/himalaya}"
-AUDIT_LOG="${HIMALAYA_AUDIT_LOG:-/var/log/himalaya-audit.log}"
-
-# ── Destructive commands we block ──────────────────────────
-BLOCKED_CMDS=(
-  "message move"
-  "message delete"
-  "message copy"
-  "flag add"
-  "flag remove"
-  "folder create"
-  "folder delete"
-  "folder rename"
-  "template send"
-  "account configure"
-  "account delete"
-)
-
-# ── Determine the subcommand being invoked ─────────────────
-# Strip leading options (--account, --output, etc.) to find the verb
-ARGS=()
-SKIP_NEXT=false
-for arg in "$@"; do
-  if $SKIP_NEXT; then
-    SKIP_NEXT=false
-    continue
-  fi
-  if [[ "$arg" == --* ]]; then
-    case "$arg" in
-      --account|--output|--page|--page-size|--folder|--color|--format)
-        SKIP_NEXT=true ;;
-    esac
-    continue
-  fi
-  ARGS+=("$arg")
-done
-
-# Build subcommand string and check against blocklist
-CMD_STR=""
-for ((i=0; i<${#ARGS[@]}; i++)); do
-  if [ -z "$CMD_STR" ]; then
-    CMD_STR="${ARGS[$i]}"
-  else
-    CMD_STR="$CMD_STR ${ARGS[$i]}"
-  fi
-  for blocked in "${BLOCKED_CMDS[@]}"; do
-    if [[ "$CMD_STR" == "$blocked" ]]; then
-      TS=$(date '+%Y-%m-%d %H:%M:%S')
-      echo "[AUDIT] $TS BLOCKED: himalaya $*" >> "$AUDIT_LOG"
-      echo "ERROR: Command 'himalaya $CMD_STR ...' is blocked by read-only policy." >&2
-      echo "       Audit log: $AUDIT_LOG" >&2
-      exit 100
-    fi
-  done
-done
-
-# ── Allow pass-through ─────────────────────────────────────
-exec "$HIMALAYA_BIN" "$@"
--- a/ai/llama-cpp/Dockerfile
+++ b/ai/llama-cpp/Dockerfile
@@ -1,30 +0,0 @@
-# llama-cpp-rocm6/Dockerfile
-# Custom llama.cpp server with ROCm 6.1 + gfx906 (MI50) support.
-# Build: docker build -t llama-cpp:rocm-gfx906 .
-
-FROM rocm/dev-ubuntu-22.04:6.1.2-complete AS builder
-RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y curl git build-essential pkg-config cmake make && rm -rf /var/lib/apt/lists/*
-ARG LLAMACPP_VERSION=b9596
-RUN git clone --depth 1 --branch ${LLAMACPP_VERSION} https://github.com/ggml-org/llama.cpp.git /build
-WORKDIR /build
-ENV HIP_PATH=/opt/rocm ROCM_PATH=/opt/rocm PATH=/opt/rocm/bin:/opt/rocm/llvm/bin:${PATH} CMAKE_PREFIX_PATH=/opt/rocm
-RUN mkdir build && cd build && \
-    cmake .. -DGGML_HIP=ON -DCMAKE_BUILD_TYPE=Release \
-      -DAMDGPU_TARGETS="gfx906:xnack-" \
-      -DCMAKE_POSITION_INDEPENDENT_CODE=ON \
-      -DGGML_CUDA=OFF -DGGML_VULKAN=OFF -DGGML_METAL=OFF \
-      -DBUILD_SHARED_LIBS=OFF && \
-    cmake --build . --target llama-server -- -j $(nproc)
-
-FROM ubuntu:24.04
-RUN apt-get update && DEBIAN_FRONTEND=noninteractive apt-get install -y \
-    ca-certificates curl libstdc++6 libgomp1 libopenblas0 \
-    libnuma1 libelf1 libdrm2 libdrm-amdgpu1 \
-    && rm -rf /var/lib/apt/lists/*
-COPY --from=builder /opt/rocm/lib/ /opt/rocm/lib/
-COPY --from=builder /opt/rocm/share/ /opt/rocm/share/
-COPY --from=builder /build/build/bin/llama-server /usr/local/bin/llama-server
-RUN echo /opt/rocm/lib > /etc/ld.so.conf.d/rocm.conf && ldconfig
-ENV HSA_OVERRIDE_GFX_VERSION=9.0.6 HCC_AMDGPU_TARGET=gfx906 HSA_ENABLE_SDMA=0
-EXPOSE 8080
-ENTRYPOINT ["/usr/local/bin/llama-server"]
Author	SHA1	Message	Date
Hermes	ebad994d60	feat(hermes): enable dashboard (HERMES_DASHBOARD=1) + Authelia auth Some checks failed Build Hermes agent / build (pull_request) Has been cancelled Details Build ollama (gfx906) / build (pull_request) Has been cancelled Details	2026-05-20 14:06:23 -04:00
Hermes	5f25c87775	fix: add missing USER root before chown step Some checks failed Build Hermes agent / build (pull_request) Has been cancelled Details Build ollama (gfx906) / build (pull_request) Has been cancelled Details The chown -R hermes:hermes was running as non-root user 'hermes' since USER hermes was set earlier. The new upstream base image (v0.12.0+) has tools/ owned by root, so the chown fails. Previous base image happened to have tools/ owned by hermes, making the chown a silent no-op.	2026-05-20 13:32:18 -04:00
Hermes	4e566b2408	fix: resolve Docker build errors and add Traefik routing for Hermes web UI Some checks failed Build Hermes agent / build (pull_request) Has been cancelled Details Build ollama (gfx906) / build (pull_request) Has been cancelled Details - Replace rsync with cp -a (rsync unavailable in latest upstream base image) - Remove npm run build step (fork's package.json has no build script) - Remove himalaya-ro.sh from build context (deployed via install.sh) - Add hermes to ai_net network for Traefik access - Add Traefik labels routing hermes.lazyworkhorse.net to dashboard port 9119	2026-05-20 13:14:10 -04:00