Olly Observability Agent Skill

Use this skill to interact with Coralogix's Observability Agent (Olly) via the cx olly CLI commands. Olly can analyze your observability data, answer questions about alerts, metrics, logs, and generate artifacts like charts and reports.

CLI Commands

Command	Purpose	Key flags
`cx olly ask "message"`	Send a message to the Observability Agent	`--chat-id`, `--mode`, `--model`, `--timeout`
`cx olly artifacts list`	List all generated artifacts	-
`cx olly artifacts get <id>`	Get artifact content by ID	-

Output format: append -o json or -o agents for machine-readable output.

Single-profile only: cx olly commands do not support multi-profile fan-out. Use -p <profile> to specify a single profile.

Chat Commands

Start a new conversation

cx olly ask "What alerts fired today?"

This creates a new chat and returns a response along with a Chat ID that you can use for follow-up questions.

Continue an existing chat

cx olly ask "Tell me more about the error rates" --chat-id <chat-id>

Use --chat-id to continue a conversation and maintain context from previous messages.

Interaction modes

Mode	Description
`fast`	Quick responses, less detailed analysis
`focus`	Balanced mode (default) - good for most queries

cx olly ask "Quick summary of recent errors" --mode fast

Model selection

Available models include gpt-5.2 (default), claude-sonnet-4-5, sonnet-4.6, gpt-5.4, claude-haiku-4-5.

cx olly ask "Explain this error" --model claude-sonnet-4-5

Timeout

For complex queries that may take longer, increase the timeout (default: 900 seconds):

cx olly ask "Deep analysis of last week's incidents" --timeout 1800

Artifacts

Olly can generate artifacts like charts, tables, and reports. Artifact IDs appear as links in the agent's response text.

List all artifacts

cx olly artifacts list
cx olly artifacts list -o json

Get artifact content

cx olly artifacts get <artifact-id>
cx olly artifacts get <artifact-id> -o json

The artifacts get command automatically:

Fetches artifact metadata
Downloads content from the presigned URL
Decompresses gzip content
Parses JSON and uses spill logic for large content
Saves non-JSON text to a temp file

Output behavior:

JSON content: Displayed directly, or spilled to file if large
Text content: Saved to temp file (e.g., /tmp/cx_results_artifact_<id>_<hash>.txt)

Workflow Examples

Investigate an issue

# Start investigation
cx olly ask "Why is the checkout service showing high latency?"

# Follow up with the chat ID from the response
cx olly ask "What changed in the last hour?" --chat-id abc-123-def

# Get any generated charts
cx olly artifacts list -o json | jq '.[0].id'
cx olly artifacts get <artifact-id>

Get JSON output for scripting

# Get response as JSON
cx olly ask "List top 5 error messages" -o json | jq '.response'

# Parse artifacts
cx olly artifacts list -o json | jq '.[] | {id, filename, created_at}'

Detailed analysis with specific model

cx olly ask "Perform root cause analysis for the outage on 2024-01-15" \
  --model claude-sonnet-4-5 \
  --timeout 1800

Key Principles

Chat IDs enable context - save the Chat ID from responses to continue conversations
Use -o json for scripting - pipe to jq for filtering and extraction
Artifact IDs are in response text - look for markdown links like [Chart](https://...artifact_view/<id>)
Single-profile only - cx olly does not support multi-profile queries
Large artifacts auto-spill - JSON content over the configured limit is saved to temp files

Related Skills

cx-telemetry-querying - for direct DataPrime/PromQL queries without AI agent assistance (covers logs, spans, metrics, RUM)
cx-alerts - for managing alert definitions

cx-olly