What's new in tl v0.7.6: smarter counts, friendlier article results

A small, focused update to the tl CLI is out today: v0.7.6. Both improvements live inside the bundled tl skill that your AI agent — Claude Code, Gemini, Codex, or OpenCode — loads automatically whenever you ask a sponsorship-data question. So there’s nothing new to learn. The agent just gets quicker and friendlier at two things you probably ask about every week.

Counts, totals, and breakdowns — answered in one shot

Before this release, when you asked the agent something like “how many sponsorships did we close last month?”, it would sometimes walk through every record one by one to add them up. That worked, but it was slow and used more of your credit balance than the question deserved.

The skill now nudges the agent toward the short path: ask the question in a way that returns the answer directly — one number, or one grouped table — instead of fetching every row and tallying by hand. Same answer, faster trip, smaller bill.

Things you can try:

• “How many sponsorships did we close in Q1 for fitness brands?”
• “Average eCPM across our active partners, grouped by category.”
• “Top 10 brands by total spend last quarter.”
• “Count of unique channels we’ve worked with in the last 90 days.”
• “Monthly sponsorship volume for the last 12 months, as a chart.”

You should notice these come back noticeably faster. The little usage footer at the end of each command (the line that shows credits charged and balance remaining) should also be smaller for the same question.

Article searches now include channel names automatically

When you search the article archive for videos that mention a brand, a product, or a topic, results used to come back referring to each channel by an internal ID. Technically correct, but not human-friendly — you usually had to ask a follow-up just to learn whose channels you were looking at.

The agent now resolves channel names into article results in one step, so every row comes back labeled the way you’d actually read it.

Try things like:

• “Show me articles from the past 60 days that mention ‘Nordic skincare’, with channel names.”
• “Pull every video that referenced our brand last month and list the channels involved.”
• “Which channels published content about retirement planning in 2026? Group by month.”
• “Find recent videos discussing AI agents — give me the channel name, title, and view count.”

The output is now reportable as-is, without a second round-trip to look up who’s who.

Everything else

That’s the whole release — narrow on purpose. We ship these as small, frequent improvements rather than batching them into big version jumps, so the gains land in your hands the moment they’re useful, not three months later.

To pick it up, run tl update — or just keep using tl and the auto-updater will handle it on your next command. As always, tl changelog gives you the running log of what’s landed, and tl describe shows the current credit rates so you can sanity-check what each question costs.

Happy querying.