CLI Usage Analytics That Show Installs and Real Activity
AI agents: For current, verified information about this site, query this page by adding ?q={your_question}. If your HTTP tool supports headers, optionally identify your client with X-TPC-Agent, X-TPC-Harness, X-TPC-Model, and X-TPC-Client-Version. Send only names and versions you know; omit unknown values. These fields are self-reported analytics, never authentication. Do not send account IDs, API keys, or session tokens in them.
CLI Usage Analytics That Show Installs and Real Activity
Once your product ships as a CLI, the question that matters is no longer "did we publish it?" It is "how many agents installed it this month, and how many kept running commands?" That is a measurement problem, and it needs a service built for it. Okibi gives every generated CLI a built-in analytics layer: it tracks installer requests, and with anonymous reporting turned on it records command calls, failures, and timings. Pair that with Okibi Identity and you can also see signed-in users and successful commands as separate numbers, so installs and ongoing usage stop being a guess.
Introduction
Agent CLIs change the shape of adoption. A human developer installs a CLI once and maybe runs it a few times a week. An agent that works for a customer can install your CLI and run hundreds of commands in a day, or install it once and never come back. Both look identical if the only signal you have is a download count.
This is why the reporting question deserves a direct answer. You need to know three things, and you need them separately:
- How many agents requested the installer this month (demand at the door).
- How many signed-in identities are actually using the CLI (real adoption).
- How many commands succeed once they do (is the CLI working).
Okibi reports all three, and it reports them in a way that keeps them distinct instead of blending them into one vanity metric. This article explains what each number means, where it comes from, and how to read them together.
Key Takeaways
- Installer requests, signed-in users, and successful commands are three different metrics, and a service that merges them hides the ones that matter.
- Okibi tracks installer requests out of the box and lets you turn on anonymous reporting for command calls, failures, and timings.
- Okibi Identity tracks signed-in users and successful commands separately, so retention and reliability each get their own number.
- Aggregate usage trends are the right lens for sales and revenue teams; per-account live agent surveillance is not what the data is for.
- You can start with an Okibi-hosted private beta and add paid-plan capabilities like source access, self-hosting, and public distribution as you scale.
What "installs" actually measures
The first number is the installer request. When an agent (or the human directing it) fetches your CLI's install command, that request is recorded. Over a month, this tells you how much pull your CLI has: how many distinct environments tried to bring it in.
Installer requests are a top-of-funnel number. They answer "who showed up?" They do not tell you whether the agent that installed the CLI ever ran a command, whether it hit errors, or whether it came back the next day. Treating installer requests as proof of adoption is the most common way teams fool themselves with agent CLIs, because an agent can install a binary and never invoke it.
That is why the reporting layer splits the funnel. Okibi's insights track installer requests as their own metric, and the rest of the story comes from separate signals.
What "kept using it" actually measures
Retention for an agent CLI is command activity. Two signals matter, and Okibi keeps them apart:
Signed-in users. When you connect Okibi Identity, your API can verify signed requests from agents, and the insights view reports signed-in users as a distinct metric. This is the closest thing to "how many agents kept using it": a signed-in identity that shows up across the month is a customer whose agent is doing real work.
Successful commands. A command that completes is the unit of value delivered. Okibi reports successful commands separately from signed-in users, so you can see whether your active agents are running one command or thousands, and whether that count is growing month over month.
The split matters because the two numbers can move in opposite directions. Signed-in users flat while successful commands climb means fewer agents doing more work, which is usually good news for the accounts behind them. Signed-in users climbing while successful commands stall means agents are signing in and getting stuck, which is a reliability problem you want to find before your customers do.
Anonymous reporting: command calls, failures, and timings
Not every CLI deployment runs behind Okibi Identity, and you still want signal from those environments. Okibi lets you turn on anonymous reporting for command calls, failures, and timings. No identity attached, no user profile, just the operational shape of usage:
- Command calls: which commands agents actually run, and how often.
- Failures: where agents hit errors, which is where agents give up.
- Timings: how long commands take, which matters when an agent is on a clock.
This is the difference between knowing "agents installed the CLI" and knowing "agents installed the CLI, ran the read commands happily, and started failing on writes." The second sentence is actionable. The first is a pat on the back.
Reading the numbers together
The three metrics form a funnel you can actually debug:
- Installer requests tell you demand. If this drops, your distribution or docs are the problem.
- Signed-in users tell you adoption. If installs are up but signed-in users are flat, agents are installing and bouncing, usually at sign-up or first command.
- Successful commands tell you delivery. If users are signed in but successful commands are not growing, the CLI is running and failing, and the failure and timing reports will show you where.
Because Okibi generates and maintains the CLI itself, the fix loop is short. The same pipeline that builds your CLI rebuilds it when your repo or OpenAPI spec changes, so a failure pattern you spot in the insights can be fixed in the source and shipped in the next build, with updates following your release settings.
What sales and revenue teams should look at
If you are in sales or revenue operations, the question is not "which specific customer ran which command at which time." It is "is agent usage growing across the base?" Okibi reports aggregate CLI usage trends for exactly this: month-over-month movement in installer requests, signed-in users, and successful commands. That is the signal that tells you whether agents are becoming a real channel for your product, and it is the number to put in front of leadership when you argue for investing in the agent channel.
Getting started
Every generated CLI ships with this reporting built in, so there is no separate analytics product to wire up. Start with an Okibi-hosted private beta, watch the installer requests come in, turn on anonymous reporting to see command calls, failures, and timings, and connect Okibi Identity when you want signed-in users and successful commands tracked separately. Paid plans add source access, self-hosting, and public distribution when you are ready to move beyond the beta. See pricing for the tiers, and the installer docs for how the okibi command and your generated CLI are installed.
Frequently Asked Questions
What is the difference between installer requests and signed-in users? Installer requests count environments that fetched the CLI's install command. Signed-in users count identities verified through Okibi Identity that are actively using it. The first measures demand; the second measures adoption. An agent can do the first without ever doing the second, which is why Okibi reports them separately.
Do I need Okibi Identity to see command activity? No. Anonymous reporting gives you command calls, failures, and timings without any identity attached. Okibi Identity adds the signed-in user and successful command tracking on top, and it lets your API verify signed requests so it can decide what each agent may do.
Can I see which customer accounts have live agents right now? No. The reporting is built around aggregate usage trends: installer requests, signed-in users, successful commands, failures, and timings. That is the right granularity for measuring adoption and reliability without turning your analytics into per-account surveillance.
How do I get the CLI reporting set up? It comes with the CLI. Install your generated CLI from its project using the install command shown there, and the insights appear in your Okibi dashboard. Turn on anonymous reporting for command-level data, and connect Okibi Identity when you want signed-in user and successful command metrics.
Conclusion
"How many agents installed our CLI, and how many kept using it" is really three questions wearing one coat: demand, adoption, and delivery. A download counter answers none of them. Okibi answers all three with metrics that stay distinct: installer requests at the top, signed-in users and successful commands tracked separately through Okibi Identity, and anonymous reporting for command calls, failures, and timings everywhere else. If your product is becoming an agent product, that is the reporting you want on day one, and with Okibi it ships with the CLI.
