agent-fed skill profile
Skills proved by work, not by you
Your AI agents watch you work and build your skill profile from real evidence. No self-assessment, no resume writing, no guessing.
Self-assessed profiles lie
npx @temet/cli audit --publish in Terminal.Your agents already know
When you work with an AI agent, every decision you make is visible. How you scope a task, how you react to errors, how you refine instructions, how you manage complexity. Your agent sees all of it. Temet turns that observation into a structured profile with concrete evidence from real sessions.
Evidence, not endorsements
A Temet profile does not say you are good at scope management because you checked a box. It says you held scope on 4 consecutive refactors, deferred CSS changes until logic was merged, and split a 400-line PR into 3 focused changesets. That is evidence. It is specific, timestamped, and extracted from real work.
The profile improves with every session
Each time you work with an agent that runs Temet, new observations are added. Skills get upgraded when the evidence supports it. Watchouts get resolved when the pattern disappears. The profile is not a snapshot. It is a living record of how you actually work and how that changes over time.
Publish a profile built from real work, not self-description.
FAQ
Can I edit what Temet extracts?
No. The profile reflects real work. That is the point. If something is wrong, the fix is to work differently, not to edit the profile.
Is my work data sent to a server?
The local audit runs on your machine. Only the extracted skills and patterns are published if you choose to make your profile public.
How is this different from LinkedIn endorsements?
LinkedIn endorsements are social signals with no evidence. Temet profiles are machine-extracted from real work sessions with concrete proof.
Next step
Use this guide in practice with Temet's audit, tracking, and profile workflow.
Connect your agentPublished March 22, 2026