Observability & enforcement
See every session, skill, and PR.
Enforce the right behavior.
Full observability into how your team codes with agents — every session, every token, every skill, every PR — and the enforcement to make sure the right skills actually get used. Receipts, not vibes.
local & team
Your own observability.
Then your whole org's.
Solo, with the CLI, you get full observability into your own agent work. On a team, it rolls up across everyone.
- ▸Read any of your sessions in full — tokens, cost, timing, every tool call
- ▸See which of your skills ran, and what each one saved
- ▸Track your own spend by day, week, and month
- ▸Every teammate's sessions — and which skills ran on which PR, not just yours
- ▸What skills are used on what PR, across every repo
- ▸Where tokens go across the team — by model, by person, by repo
- ▸Org-wide adoption, compliance, and enforcement
[1] · session observability
Every session, live
and on the record.
See what’s running right now and read any past session in full — tokens and cost, timing and turns, every tool call, and the signals ZeroShot pulled out of it. One click turns a friction pattern into a skill.
- live status across the team — working, waiting, idle
- full detail: tokens, cost, span, turns, and tool activity
- a signal pipeline per session — and a one-click create skill
└ signal pipeline · 8 signals → create skill
[2] · usage observability
Where every token goes.
Spend by day, week, and month — by model, by person, and by repo — with a projection for where you’ll land, and the output-per-prompt that proves your skills are working. The bill stops being a surprise.
By model · last 30 days
[3] · skill observability & enforcement
Is the skill actually
getting used?
For every skill: coverage and adoption, who owns it, and how healthy it is — then set the rules: when it runs (automatically on a flow, or manually) and require it where it matters. A skill file no one runs changes nothing.
- coverage & adoption — what share of eligible sessions ran it
- rules for when each skill fires, by file glob or work type
- governance — owner, source, health, ready-to-enforce
When this skill runs
[4] · pr observability
Was this skill used on this PR?
ZeroShot enforces the required skills before a commit lands, and marks every PR compliant or not — with the evidence behind it. The skill no longer depends on whether someone remembered to run it.
[5] · the daily report
All of it, in Slack,
before your first coffee.
Every morning ZeroShot posts the overnight to #eng: a TL;DR that leads with speed and time saved, your rank vs the team, a snapshot of new and updated skills, and a per-person view of what everyone’s working on.
- overnight TL;DR — PRs shipped, speed, hours saved
- new & flagged skills — review or fix, one click
- where everyone’s stuck — with the skill that fixes it
Overnight your team shipped 8 PRs about 4× faster and skipped 15h of repeat work. 3 new skills are ready for review.
─── Saved this week ───Skills cut repeat work 90% — about $3,140 saved this month.
─── Active sessions · ● Live · 7 running ───/payments-reviewspotted across 11 PRsReview →/export-perfspotted across 6 ticketsReview →/migration-guardmissed a rollback 3×Fix →Where everyone's stuck: 3 sessions burned 40k+ tool calls on the export migration — add /migration-guard to fix.
the point, in one sentence
“For the first time I can actually see how my team codes with agents — who’s stuck, where the tokens go, and whether the skill actually ran.”
— Spencer Shulem, CEO