GrepAI Health Check

Diagnose all GrepAI components and report status with indicators.

Workflow

Run each check sequentially. Collect results, then print a unified report.

1. Docker Services

Check by image ancestry rather than hardcoded container names:

docker ps --filter ancestor=ollama/ollama --format "{{.Names}}\t{{.Status}}\t{{.Ports}}"
docker ps --filter ancestor=qdrant/qdrant --format "{{.Names}}\t{{.Status}}\t{{.Ports}}"
docker ps --filter ancestor=postgres --format "{{.Names}}\t{{.Status}}\t{{.Ports}}"

Record: which services are running.

2. Ollama Connectivity

curl -s --max-time 5 http://localhost:11434/api/tags

Record: reachable? Parse response for available models.

3. Ollama Embedding Models

From the /api/tags response, list models that are embedding models:

• •mxbai-embed-large
• •nomic-embed-text
• •bge-m3
• •nomic-embed-text-v2-moe

Record: which embedding models are pulled.

4. PostgreSQL Connectivity

Check if a postgres container is running (from step 1). If so, get the container name and check:

docker exec {POSTGRES_CONTAINER} pg_isready -U grepai

Skip if config shows GOB or qdrant backend and no postgres container is running. Record: accepting connections?

5. pgvector Extension

docker exec {POSTGRES_CONTAINER} psql -U grepai -d grepai -tAc "SELECT extname FROM pg_extension WHERE extname='vector'"

Skip if config shows GOB or qdrant backend. Record: vector extension installed?

6. Qdrant Connectivity

Check Qdrant REST API on port 6333 (not 6334, which is gRPC):

curl -s --max-time 5 http://localhost:6333/collections

Skip if config does not use qdrant backend and no qdrant container is running. Record: reachable? Collection count.

7. grepai Config

Read: .grepai/config.yaml

If missing, report not initialized. Otherwise extract and report:

• •Provider + model
• •Storage backend
• •Endpoint URL

8. grepai Index Status

grepai status

Record: indexed files, chunks, last updated.

9. MCP Registration

claude mcp list

Or read ~/.claude/mcp.json and ./.claude/mcp.json. Record: grepai MCP server registered?

10. Watch Daemon

grepai watch --status

Record: running or not.

11. Workspace Config

grepai workspace list

If workspaces exist, get details for each:

grepai workspace show {NAME}

Optionally also read ~/.grepai/workspace.yaml for additional detail.

Record: configured workspaces, backend type, project count per workspace. Skip if no workspaces configured.

12. Workspace Status & Watcher

grepai workspace status {NAME}

And check workspace watcher:

grepai watch --workspace {NAME} --status

Record: per-workspace indexing health and watcher state. Skip if no workspaces configured.

Report Format

Print unified status report:

============================================================================
GrepAI Health Check
============================================================================

Infrastructure:
  {S} Docker           ollama: {STATUS}
  {S} Ollama           http://localhost:11434 — {STATUS}
  {S} Models           {MODEL_LIST} (or "none pulled")
  {S} PostgreSQL       {STATUS} (only if postgres backend, otherwise "skipped")
  {S} pgvector         {STATUS} (only if postgres backend, otherwise "skipped")
  {S} Qdrant           http://localhost:6333 — {STATUS} (only if qdrant backend, otherwise "skipped")

Configuration:
  {S} Config           .grepai/config.yaml — {PROVIDER}/{MODEL}
  {S} Storage          {BACKEND}

Index:
  {S} Files indexed    {COUNT}
  {S} Chunks           {COUNT}
  {S} Last updated     {TIMESTAMP}

Integration:
  {S} MCP server       {SCOPE} — {STATUS}
  {S} Watch daemon     {STATUS}

Workspaces: (only if workspaces configured)
  {S} {WORKSPACE_1}   backend: {TYPE}, {PROJECT_COUNT} projects — {STATUS}
     Watcher: {RUNNING|STOPPED}

============================================================================

Where {S} is one of: OK for healthy, FAIL for failed/missing, WARN for degraded/warning.

Troubleshooting Hints

After the report, if any component shows FAIL, print targeted fix suggestions: