Run Your First Query¶

Once installed, there are three ways to interact with OLAV. Let's start with the simplest.

Feature Claims

ID	Claim	Status
C-L2-02	`olav list` lists all available Agents	✅ v0.10.0
C-L2-14	`olav` with no arguments launches TUI interactive mode	✅ v0.10.0
C-L2-15	`olav --session <id>` resumes a session	✅ v0.10.0
C-L2-16	`olav service web start` launches the Web UI	✅ v0.10.0
C-L2-08	`olav log list` shows operation history	✅ v0.10.0

CLI Mode: One-Liner Questions¶

The fastest way is to type a natural-language question directly on the command line:

olav "list all agents"
olav "what tables are in the database?"
olav "run a health check"

OLAV automatically routes your question to the most suitable Agent. The default quick Agent is ideal for fast lookups.

If you want to query information about the OLAV platform itself, use the built-in commands:

olav version                          # Check platform version
olav list                             # List all available Agents
olav --agent services "List registered services"   # View connected external services

You can also use --agent (or -a) to target a specific Agent directly:

olav --agent core "run df -h"                              # Execute a system command
olav --agent core "run this python: import sys; print(sys.version)"  # Execute Python code

Interactive Mode: Multi-Turn Conversations¶

Run olav without arguments to enter the interactive terminal (TUI), which supports multi-turn conversations:

olav                      # Start with the default Agent
olav --agent admin       # Start with a specific Agent
olav --no-splash          # Skip the startup banner

In interactive mode, you can ask questions naturally as in a chat, and the Agent maintains context:

> How many Agents are available?
> What did I do in the last hour?
> Show recent errors

The Welcome Screen Knows Your State¶

The TUI's first screen is not a random tip — it reflects what is actually true about your installation right now:

Real findings: "⚠ Embedding backend unavailable — memory/recall features will be limited. Run olav doctor for details." or "netops has no device data yet — import a snapshot to get your first answers."
Session continuity: returning users see "Welcome back — last time (2h ago): 'why is R1's BGP flapping?'" so you can pick up where you left off.

When everything is healthy and there's nothing to say, you get a regular usage tip instead.

Slash Commands in Interactive Mode¶

Command	Function
`/help`	Show all available commands
`/doctor`	Run the zero-LLM health check (scaffolding + LLM + embedding) without leaving the TUI
`/workspace <name>`	Switch to another agent workspace (e.g. `/workspace admin`)
`/clear`	Reset the current conversation and clear the screen
`/model <name>`	Temporarily switch the LLM model within the session (does not modify config files)
`/trace-review`	Manual one-shot review of recent failures (mostly superseded by always-on L1 capture — see Self-Improving Loop)
`/quit` `/exit` `/q`	Exit interactive mode

Special Syntax in Interactive Mode¶

Syntax	Function	Example
`@file.txt`	Inject file contents into your prompt	`@config.yaml What's wrong with this config?`
`!command`	Execute a shell command directly (bypassing the Agent)	`!ls -la exports/`

Self-improvement is automatic

OLAV captures every successful tool call into LanceDB memory as you use it (L1), and a curator periodically distils repeated patterns into reusable directives (L2). Both surface back into prompts via AutoRecall — so the system gets smarter from your usage without any manual review step. See Self-Improving Loop for details and how to extend it with team-specific directives.

Resuming a Previous Session¶

olav --session <session-id>

Sessions are stored in ~/.olav/sessions/ and expire after 24 hours by default.

Web UI: Browser Access¶

If you prefer a graphical interface or need to share with your team:

olav service web start

Then open http://localhost:2280 in your browser to use the same Agent capabilities as the CLI, with real-time streaming responses.

See Background Services → for details.

Getting to Know Your Agents¶

Run the following command to see which Agents are available in your current workspace:

olav list

Available Agents:

  • config
    Config & System Agent — Service registration, data import, workspace health checks
    Location: .olav/workspace/config/

  • core
    Core OLAV platform agent — Code execution, SQL queries, shell commands
    Location: .olav/workspace/core/

  • quick  (active)
    Core Agent — Fast data queries, CLI execution, knowledge base search
    Location: .olav/workspace/quick/

Each Agent specializes in different tasks:

Agent	What It Does Best	Use Cases
`core` (default)	Fast queries, single-step answers	"What tables are in the database?" "Show recent errors"
`config`	Manage service registration, workspace health checks	Connect a new API, diagnose workspace issues
`core`	Execute Python/SQL/Shell, web search	Data analysis, automation scripts, complex queries

Viewing Operation History¶

OLAV automatically logs every operation. View recent activity:

olav log list

Recent Audit Runs (last 24h):

  [2026-04-03 15:35:52] 7f2693b8  completed     agent=core
  [2026-04-03 15:35:25] 2e31144b  completed     agent=config
  [2026-04-03 15:34:20] 4e314755  completed     agent=quick

See Audit and Logs → for details.

Next: How It Works →