WorkSpark for AI assistants

Let your assistant handle the performance admin

Connect WorkSpark to Claude, Cowork, Codex, or OpenCode. Ask it to log what you shipped, draft your self-review, or check your OKRs — and it works from your real records, never invented ones.

Works withClaudeCoworkCodexOpenCode
Your assistant · WorkSpark connected

Log what I shipped this week.

Tool call · Reading latest work from GitHub
Tool call · Reading latest accomplishments in WorkSpark

Here's a draft of your accomplishments from this week's work:

  • Cut cold-start latency by 40%execute
  • Shipped the new onboarding flowcreate
  • Migrated the billing service to Postgres 17maintain

Create these entries?

Create all threeLet me edit first

Try asking

Things you stop doing by hand

Plain requests, in whatever assistant you already have open. No commands to memorise.

Log what I shipped today.

Creates the accomplishment, with title and category filled in for you.

Prep my self-review.

Pulls your entries for the cycle and drafts from them — flagging the thin ones.

Which OKRs are stale?

Finds key results with no recent check-in, before someone else notices.

Summarise my team's quarter.

A paragraph per person, built only from work they actually logged.

Who's been quiet this month?

Surfaces reports with no entries, so nobody's work goes unrecorded.

Is this entry good enough?

Coaching on specificity and evidence before the entry is saved.

The guarantee

Grounded in real records, never fabricated

Performance data is the one place a confident guess is unacceptable. Your assistant can only report what WorkSpark actually returned.

Reads real data

Every claim traces back to an accomplishment, review, or OKR record with an ID.

Confirms before writing

No entry, review draft, OKR update, or survey response is saved without your say-so.

Your permissions

The assistant sees exactly what you see — same account, same organisation, same access.

Three ways to enable it

Pick the one that matches your assistant

Two pieces have to be in place: a connection to WorkSpark, and a playbook that teaches your assistant how to use it. How you install them depends on where you work.

Plugin

Easiest

Claude & Cowork

One file bundles both pieces. Upload it, allow network access, sign in by asking.

  • Connection + playbook included
  • No terminal at any point
  • About two minutes
Install the plugin

Skill

Codex, OpenCode & others

One command adds the WorkSpark playbook to your harness. The connection installs alongside it.

  • One command to install
  • Works with any Skills-aware harness
  • About three minutes
Add the Skill

Manual

Install the CLI yourself

For managed machines, custom harnesses, or when you want to drive WorkSpark from your own terminal too.

  • Full control over the binary
  • Every install method covered
  • About five minutes
CLI install guide

Your assistant is already open. Give it your performance data.

Two minutes to enable, and nothing it tells you is made up.