See, understand and audit the work of your agents
Hız ajanlarda,
izan sende.
İ.Z.A.N. gathers every coding agent of one person in one inbox and one campus. İZAN Server brings the same picture to the company: managers see what the agents do, understand where work waits and what it costs, and audit every decision.
npx @lyconiq/izan open edition, free

Why now
Agents have arrived. Oversight lags behind.
developers using AI agents, 2025 → 2026
Stack Overflow Developer Survey 2025; pulse survey, April 2026 (≈1,100 people)
rarely or never let an agent run fully on its own
Stack Overflow pulse survey, April 2026
of agentic AI projects are expected to be cancelled by the end of 2027
Gartner prediction, 25 June 2025
Two editions
An inbox for the developer, a city for management.
One inbox and one campus for every agent of one person
- Inbox: work waiting for you, running and done, in one place
- Campus: every session a building, with its state and cost
- Claude Code, GitHub Copilot CLI, Codex and Gemini CLI on one screen
- The whole process engine and the shipped
sdlc-liteprocess - Station: GitHub, Jira, Jenkins, Bitbucket, read-only
- Code, prompts and transcripts never leave your machine
npx @lyconiq/izan
See, understand, audit
- City view: every team’s campus on one map
- Dashboard, team and process reports: waits, rework, cost
- Server gates: the right role makes the decision
- Tamper-evident audit log and SIEM forwarding
- On-premises or in the cloud, the same product
- Company identity (OIDC, SAML) and vaults
What management gets
See, understand, audit.
See
The city view shows every team on one map: who is working, who is waiting, which item is at which stage.
Understand
Team and process reports give the wait at each gate, the rework, and the cost per stage and team, with a weekly trend.
Audit
At a server gate the right role decides. Every decision goes to a tamper-evident audit log and flows to your SIEM.

Privacy
Only numbers leave the machine.
The bridge never calls a model. Unless it is enrolled, it connects to nothing. Once enrolled, it sends only the records of a published schema.
| Travels to the server | Stays on the machine |
|---|---|
| Session ids (keyed), states, waits, costs | Prompts, transcripts, file contents |
| Gate decisions: who decided, and when | Command text, titles, folder and project names |
| Run summaries: items, stages, attempts, cost | The agent’s working files |
One team, six weeks, free.
The pilot runs on your own server. At the end you look at the same measures and decide.