Skip to content

aviciot/slack_agent

Repository files navigation

DS_agent

ngrok config add-authtoken YOUR_TOKEN to run the service ngrok http 8090

1) Copy .env.example → .env and fill Slack tokens

2) Start ollama sidecar

docker compose up -d ollama

3) Pull a model (inside ollama container)

docker exec -it ollama bash -lc "ollama pull mistral"

4) Build & run API

docker compose up -d --build

5) Tail logs

docker compose logs -f api

for dummy db

/sql total sales in the last 7 days

/sql total sales in the last month

/sql top 3 products by revenue this month

/sql daily sales totals for the past 14 days

/sql average order value last week

If your /sql expects more explicit phrasing, use:

/sql compute the sum of sales.amount for the last 7 days

/sql list top 3 products by sum(sales.amount) in the current month

docker compose run --rm kb-seed python /app/tools/seed_queries.py

NL2SQL checks (to test the LLM path) /sql list the first 5 rows from v_wf_hierarchy_explorer

/sql top 10 workflows by number of sessions

/sql show 3 columns (subject_area, workflow_name, task_type_name) from session_run_statistics limit 5

/sql top 10 workflows by total sessions in the last 30 days

/sql failure rate (failed/total) by workflow in the last 30 days, top 10 by total volume

/sql average session duration (seconds) by workflow in the last 7 days, top 10

/sql daily total sessions for the last 14 days (date, sessions)

/sql top 10 workflows by sessions in the last 7 days where total sessions ≥ 100

/sql success rate by workflow in the last 7 days (success/total), show top 10 by volume

/sql top 10 workflows by max session duration (seconds) in the last 7 days

/sql workflows with error rate > 5% in the last 7 days, order by total sessions desc

/sql top 10 workflows by growth in sessions vs previous 7 days

/sql hourly sessions today (00:00–23:59)

in Slack:

Quick smoke tests /ask What columns are in session_run_statistics?

/ask Show 5 sample rows from workflow_run_statistics.

Simple analytics /ask Which 10 workflows have the most sessions? Return a table and a short takeaway.

/ask What’s the busiest subject_area? Give counts by subject_area (top 5).

/ask For v_wf_hierarchy_explorer, list 5 example rows and explain what each column represents.

“Figure it out” / agent-y prompts /ask Find any workflows with unusually high run counts compared to the median. Explain briefly.

/ask Are there any workflows that look like outliers by sessions per day? Show the table you used.

Debug / schema-aware /ask Do we have a timestamp column to filter recent runs? If yes, show 5 most recent session_run_statistics rows.

/ask Build the SQL to list workflow_name and total sessions, sorted desc. Then run it.

And a follow-up in the same thread (to test conversational memory):

/ask Now do the same, but only for the top 3 subject areas.

docker ollama

'{"model":"anindya/prem1b-sql-ollama-fp116","prompt":"warm","stream":false,"options":{"num_predict":1},"keep_alive":"10m"}' | Set-Content -Encoding ASCII warm.json curl.exe -s -X POST http://localhost:11434/api/generate -H "Content-Type: application/json" --data-binary "@warm.json"

See live usage

docker stats ollama

What models are currently LOADED in RAM (and by whom)

docker exec -it ollama ollama ps

What models are INSTALLED on disk

docker exec -it ollama ollama list

Inspect container limits (helps spot no memory cap / swap thrash)

docker inspect ollama --format='Memory={{.HostConfig.Memory}} Swap={{.HostConfig.MemorySwap}}'

docker exec -it ollama ollama stop mistral:latest docker exec -it ollama ollama stop llama3.1:8b-instruct-q4_K_M docker exec -it ollama ollama ps # should now be empty

Qwen coder

python sanity_ollama.py -m qwen2.5-coder:7b-instruct --threads 6 --show_ps

Mistral (pick the tag you have)

python sanity_ollama.py -m mistral:latest --threads 6 --show_ps

#Redis docker exec ds_redis redis-cli -a devpass FLUSHALL

docker compose exec redis redis-cli -a devpass FLUSHALL docker compose exec redis redis-cli -a devpass FT.DROPINDEX idx:qcache DD

docker compose exec api sh -lc "env | grep -E 'CATALOG|TABLE|JSON|DB_PATH'" docker compose build --no-cache api

docker compose run --rm --build kb-seed

docker compose exec redis redis-cli -a devpass KEYS "*"

If you also want to pull fresh base images (in case Python or deps changed):

docker compose build --no-cache api

And then restart:

docker compose up -d api

⚡ Pro tip: If you only changed requirements.txt, you can trigger just the dependency layer rebuild:

docker compose up -d --build api

tree

tree /f /a | findstr /v "\Test" | findstr /v "\V1" > tree_output.txt

get env variable from the dockers docker exec -it ds_agent_api env | Select-String PROVIDER

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors