Skip to content

docs(schema): reserve the caaf extension name and uid 240 - #492

Open
ncdingari wants to merge 1 commit into
agntcy:mainfrom
ncdingari:docs/reserve-caaf-extension
Open

docs(schema): reserve the caaf extension name and uid 240#492
ncdingari wants to merge 1 commit into
agntcy:mainfrom
ncdingari:docs/reserve-caaf-extension

Conversation

@ncdingari

Copy link
Copy Markdown
Contributor

docs(schema): reserve the caaf extension name and uid 240

What

One row in schema/extensions.md, reserving the extension name caaf and uid 240.

| Clinical Agent Assurance            | caaf | **240** | Clinical agent assurance for health AI   |

Documentation only. No schema files are added, nothing under schema/extensions/ is created, and no build, test or generated output is affected. Column widths match the existing rows, so the table needs no reformatting.

Why this file

schema/extensions.md states its own purpose: "to keep track of and avoid collisions in Extension names & ids." Registering before publishing is what makes that work. caaf and 240 are both currently unused on main.

uid 240 was picked to sit clear of dev (999) and of the low range native extensions are likely to want. I have no attachment to the number — see the question below.

What caaf is

Clinical Agent Assurance: assurance attributes for agents that operate on protected health information. Data-handling profile, authorization scope, autonomy tier, human oversight, evaluation policy. Maintained by SciEncephalon AI.

Unlike a bare reservation, the extension exists and loads. Measured against this branch with the extension placed in-tree:

task test:schema                              7 of 7 specs, SUCCESS
SCHEMA_EXTENSION=extensions mix test          366 passed

domain caaf/care_transitions   uid 2400101  category clinical_care
domain caaf/population_risk    uid 2400105  category clinical_care
module caaf/clinical_assurance uid 2400101  category clinical

Worth noting this is only verifiable as of #488. Before that fix, no in-tree extension could pass task test:schema at all, so "validates against the metaschemas" was not a claim anyone could actually check.

I am not asking to add those files here. This PR is the registry row only; the extension will be published separately once the name is settled.

A question about process, which matters more than my row

#471 asked for the same thing — a namespace reservation, agentminds_io — on 27 April 2026. It has been open four months with zero comments and zero reviews. It took a different approach, adding a README.md under schema/extensions/ rather than editing the registry file.

So there are two conventions in play and no signal about which is right, and a contributor who followed one of them has been waiting since April.

Rather than add a second unanswered request, I would rather ask directly:

  1. Is a PR the intended way to reserve an extension name and uid? If maintainers would rather allocate these some other way, say so and I will withdraw this and follow that route.
  2. If it is, which file? This PR edits schema/extensions.md, which documents itself as the collision registry. Reserve agentminds_io extension namespace #471 adds a README. Whichever you prefer, I will match it, and I am happy to update this PR to the other shape.
  3. Is 240 acceptable, or should extensions allocate from a specific range? Happy to take any number.

If the answer to (1) is yes, #471 has been waiting considerably longer than I have and should probably go first.

Note on ordering

#489 also adds a row to this table, reserving example / 998 for the reference extension. The two will conflict textually depending on merge order; whichever lands second, I will rebase. No coordination needed on your side.

Registers the caaf extension in the extensions registry so the name and
uid cannot collide with a future extension.

caaf (Clinical Agent Assurance) declares assurance attributes for agents
that operate on protected health information: data-handling profile,
authorization scope, autonomy tier, human oversight, and evaluation
policy. It is maintained by SciEncephalon AI.

Signed-off-by: Narahara Chari Dingari <chari@sciencephalon.com>
@ncdingari
ncdingari requested a review from a team as a code owner August 27, 2026 16:15
@codecov

codecov Bot commented Aug 27, 2026

Copy link
Copy Markdown

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant