← SkillSafe
OKR Desk

For team leads at the end and the start of a cycle

OKR DeskOutcomes, not features. Committed is pass or fail. Guardrails never averaged away.

Paste your team's OKRs. Free in your browser: every key result read for its baseline, target, deadline and source; deliverables posing as key results, activity counts, numbers in the objective, growth with no guardrail, scores tied to pay - and once the actuals are in, each key result scored the way its type demands. A run rewrites the set into outcome-based OKRs, or grades the closed cycle and drafts the next one from what it taught.

Try a sample:

1 Your OKR set free

Paste or drop a .txt, .md or spreadsheet .csv export (Objective and Key Result columns), or

2 The check and the scores free

Against the rules of the two OKR skills. Nothing leaves your browser for this part.

Where the problems are

The scorecard

Findings

3 Grade the cycle, or write the set

NOT RUN YET

Findings

The scan's calls

    Saved sets

    How it works

    The check reads the set line by line and applies the writer's rules from foundation-okr-writer (outcomes, not output; a baseline, target, deadline and source for every key result; 1 to 3 objectives with 2 to 4 key results; a guardrail for anything that pushes growth or speed; no compensation coupling) and, when actual values are present, the scoring conventions of measure-okr-grader: the type decides the score, guardrails are reported on their own, a window that has not closed is deferred, and partial coverage of a commitment is not a pass.

    Writing and grading are done by a language model from your set and the page's check. The page holds the reply to it: every error and warning answered by rule id; a rewritten set re-checked by the same rules; every baseline or target the reply calls "given" really in what you pasted, every other number raised as a question; every score matching the page's arithmetic unless the reply disputes it. Your OKR tracker stays the source of truth.