Skip to content
Where We're At

Building on the protocol

Lexicons and OAuth are some of the most confusing parts of ATproto. How far have you gotten with each?

Have you designed your own lexicon?

182 responses

Yes, and it is in production: 36.3%Yes, experimental so far: 25.3%Not yet, but I plan to: 16.5%No: 22%
  • Yes, and it is in production36.3%66 people
  • Yes, experimental so far25.3%46 people
  • Not yet, but I plan to16.5%30 people
  • No22%40 people
Share of the 182 people who answered. One answer each.

How do you publish your lexicon schemas?

144 responses

  1. As a com.atproto.lexicon.schema record50%, 72 people
  2. With a _lexicon DNS TXT record41.7%, 60 people
  3. They are not published39.6%, 57 people
  4. In a repository38.2%, 55 people
Share of the 144 people who answered. Several answers each, so shares can add up past 100%.

Have you implemented ATproto OAuth?

151 responses

Most builders have taken to OAuth, if needed.

Yes, shipped: 68.2%In progress: 16.6%No, I use app passwords: 6%My project does not need it: 9.3%
  • Yes, shipped68.2%103 people
  • In progress16.6%25 people
  • No, I use app passwords6%9 people
  • My project does not need it9.3%14 people
Share of the 151 people who answered. One answer each.

Which scopes does your app request?

138 responses

Granular scopes are gradually taking hold.

Granular permissions: 72.5%transition:generic: 8%Both, depending on the PDS: 7.2%Not sure: 12.3%
  • Granular permissions72.5%100 people
  • transition:generic8%11 people
  • Both, depending on the PDS7.2%10 people
  • Not sure12.3%17 people
Share of the 138 people who answered. One answer each.

What made implementing OAuth harder than it needed to be?

56 responses

  1. OAuth complexity21.4%, 12 people
  2. Sparse OAuth documentation17.9%, 10 people
  3. DPoP complexity16.1%, 9 people
  4. Local development friction16.1%, 9 people
  5. Session persistence issues10.7%, 6 people
  6. Lack of libraries7.1%, 4 people
  7. Differences between frontend and backend oauth5.4%, 3 people
  8. Lack of tutorials5.4%, 3 people
  9. Missing context for AI5.4%, 3 people
  10. Bluesky3.6%, 2 people
  11. Hosting3.6%, 2 people
  12. Library Design3.6%, 2 people
Show 29 more
  1. Missing packages3.6%, 2 people
  2. Permission Sets3.6%, 2 people
  3. Redirects3.6%, 2 people
  4. Testing3.6%, 2 people
  5. AI use (negative sentiment)1.8%, 1 person
  6. Boiler Plate Code1.8%, 1 person
  7. Bot detection issues1.8%, 1 person
  8. CIMD1.8%, 1 person
  9. Cryptography1.8%, 1 person
  10. C#1.8%, 1 person
  11. Flutter1.8%, 1 person
  12. Implementation is time-consuming1.8%, 1 person
  13. Inconsistent behavior across apps1.8%, 1 person
  14. jacquard1.8%, 1 person
  15. Lack of reference data1.8%, 1 person
  16. Lifetimes1.8%, 1 person
  17. Missing understanding from users1.8%, 1 person
  18. Missing verification tooling1.8%, 1 person
  19. Not serverless friendly1.8%, 1 person
  20. OAuth metadata cache1.8%, 1 person
  21. PaR1.8%, 1 person
  22. Passkeys1.8%, 1 person
  23. Inconsistent behavior across PDS hosts1.8%, 1 person
  24. Permission Set Display1.8%, 1 person
  25. Rust1.8%, 1 person
  26. Scopes1.8%, 1 person
  27. Setup1.8%, 1 person
  28. Spec compliance1.8%, 1 person
  29. Step up auth1.8%, 1 person
Free-text answers grouped into topics. Share of the 56 people who answered.

Will you build on ATproto spaces?

164 responses

Despite being in a pre-release stage, Spaces are highly anticipated.

I'm already experimenting with them: 31.7%I plan to once they ship: 36%Maybe, depending on the final design: 18.3%No, my project doesn't need them: 4.9%I don't know enough about them yet: 9.1%
  • I'm already experimenting with them31.7%52 people
  • I plan to once they ship36%59 people
  • Maybe, depending on the final design18.3%30 people
  • No, my project doesn't need them4.9%8 people
  • I don't know enough about them yet9.1%15 people
Share of the 164 people who answered. One answer each.

Does your app in the Atmosphere bring in money?

165 responses

Over half of apps from respondents are meant to be enjoyed for free.

Yes, it earns something: 6.7%Not yet, but I plan for it to: 27.9%No, and it is meant to stay free: 51.5%I do not have a project of my own: 13.9%
  • Yes, it earns something6.7%11 people
  • Not yet, but I plan for it to27.9%46 people
  • No, and it is meant to stay free51.5%85 people
  • I do not have a project of my own13.9%23 people
Share of the 165 people who answered. One answer each.

In a typical month, how much does it bring in?

10 responses

Asked only of people whose app already earns money

  1. Under $10030%, 3 people
  2. $100 to $1,00050%, 5 people
  3. $1,000 to $5,0000%, 0 people
  4. $5,000 to $20,00010%, 1 person
  5. More than $20,00010%, 1 person
Share of the 10 people who answered. One answer each.

What pays for the work?

152 responses

  1. My own money93.4%, 142 people
  2. Donations or sponsorships21.1%, 32 people
  3. My clients or customers7.2%, 11 people
  4. My employer7.2%, 11 people
  5. A grant5.3%, 8 people
  6. Revenue from the thing itself5.3%, 8 people
  7. Venture capital2.6%, 4 people
  8. Nothing2%, 3 people
  9. No cost0.7%, 1 person
  10. Scholarship0.7%, 1 person
Share of the 152 people who answered. Several answers each, so shares can add up past 100%.

Helpful resources