Skip to content

Usage

Core CLI loop

After installing the CLI and running kortix init + kortix ship, the day-to-day workflow is:

bash
# Start a new agent session with a plain-English prompt
kortix sessions new --prompt "Summarize this week's commits and open a change request"

# List pending change requests the agent has proposed
kortix cr ls

# Review a specific change request (shows the diff + agent rationale)
kortix cr show <cr-id>

# Merge a change request to main
kortix cr merge <cr-id>

# Talk to a running session's agent from your terminal
kortix chat

Real-world task examples

Research and report

bash
kortix sessions new --prompt "Research our top three competitors, summarize their pricing pages, and create a comparison table in reports/competitors.md"

Code generation with review gate

bash
kortix sessions new --prompt "Add input validation to the /api/users endpoint, write tests, and open a change request when done"

Scheduled automation (cron trigger)

Define a trigger in kortix.toml:

toml
[[triggers]]
name = "weekly-summary"
type = "cron"
schedule = "0 9 * * MON"
prompt = "Summarize last week's activity across GitHub, Slack, and Linear into weekly-summary.md and open a change request"

Then redeploy:

bash
kortix ship

The agent now runs automatically every Monday at 09:00 UTC.

Webhook trigger

toml
[[triggers]]
name = "on-pr-opened"
type = "webhook"
event = "github.pull_request.opened"
prompt = "Review the opened PR for security issues and post a comment with findings"

Parallel agents

You can run thousands of sessions in parallel against the same config. Each session is fully isolated on its own branch:

bash
kortix sessions new --prompt "Process row 1 of data.csv" &
kortix sessions new --prompt "Process row 2 of data.csv" &
kortix sessions new --prompt "Process row 3 of data.csv" &

All output flows back to the main branch through separate change requests, each requiring a human merge.

Project config file (kortix.toml)

toml
[project]
name = "my-company"
version = "1.0.0"

[[agents]]
name = "researcher"
persona = "agents/researcher.md"
skills = ["skills/web-research.md", "skills/report-writing.md"]

[[connectors]]
name = "github"
type = "github"

[[connectors]]
name = "slack"
type = "slack"

Agents are markdown persona files. Skills are reusable markdown know-how files. Connectors link to 3,000+ apps brokered server-side.

Switching between Cloud and self-hosted

bash
kortix hosts list          # see all configured hosts
kortix hosts use cloud     # target Kortix Cloud
kortix hosts use local     # target your self-hosted instance