Engineering
Evidence
All kitsJoin beta

Coming soon · A kit by Rob Race

See the engineering work your review will otherwise miss.

A local-first workflow that turns GitHub and Jira history into source-linked evidence you can inspect, correct, and use in a real conversation.

For self reviews, promotion packets, engineering-manager evidence, and consultant client updates.

Generated locally inside your agent session. Nothing to sign into, and no engineering data sent to us.

Alex · GitHub + Jira evidence● 8 people mapped
WeekMonthQ3Custom
Jira completed12observed transitions
PRs merged08authored work
PRs reviewed19other authors
Team review share14%complete mapped scope

Review distribution · 8 people

43% of observed participation came from two mapped people.
review comment
Retry-boundary review commentAlex asked that retries be bounded during provider outages.
high

01 / A defensible record

Not a score. A trail back to the work.

Engineering Evidence preserves the difference between authoring, reviewing, commenting, merging, and assignment. It makes the record useful without pretending activity alone proves impact, effort, or seniority.

01

Collect

Fetch canonical work from the sources you already have permission to read.

02

Reconcile

Keep source roles, stable IDs, dates, and coverage gaps intact.

03

See

Generate a private local dashboard and a source appendix.

04

Write

Start your review or promotion artifact from the same evidence.

02 / One record, three views

As your tools record it

GitHub reviews, pull requests, and Jira transitions.

As the skill reconciles it

Authors, reviewers, mergers, assignees, timestamps, and coverage gaps.

As you use it

dashboard.htmlevidence.jsonself-review.md

02 / Product evidence

Every useful claim keeps its receipts.

This is synthetic data, rendered in the shape of the local artifact the kit produces. The important detail is not the dashboard polish; it is that every row can lead back to a canonical source.

Observed fact

Alex reviewed three distinct pull requests authored by other people.

Review participation is visible as review evidence — not rebranded as a productivity score.

Review participation

5observed days
Trust boundary

Counts describe source-observed activity. They do not establish quality, impact, effort, seniority, or productivity.

03 / What you receive

A complete, inspectable workflow.

  • GitHub and Jira collection skills
  • Local HTML dashboard and normalized evidence file
  • Source appendix
  • Self-review and promotion-packet workflows
  • Installation guide for the agent you already use

Early access

Help shape the first useful version.

Early access is for engineers preparing a self review, promotion case, client report, or career conversation. Tell me what you are trying to make visible.

Join the GitHub-first beta