Skip to content

chore(next): sync main into the 10.x beta line - #1911

Closed
rachitm022 wants to merge 19 commits into
nextfrom
main
Closed

rachitm022 wants to merge 19 commits into
nextfrom
main

Conversation

@rachitm022

Copy link
Copy Markdown
Collaborator

Summary

Release sequencing

This PR must land before publishing another SDK 10 beta. A push to next triggers the dedicated Release Please (next) workflow, which should propose 10.0.0-beta.18 from .release-please-manifest.next.json.

The stable 10.0.0 proposal in #1898 must remain blocked until the 10.x line is ready for GA; publishing it from main now would omit the 312 commits already present on next.

Consumer follow-up

After the beta is published, Scope3's embedded sales agent should move its exact adcp pin from 10.0.0b17 to the published beta. No adcp-reporting-storage bump is required by #1910.

Tracks Scope3 AI-21766.

bokelley and others added 19 commits November 7, 2026 23:37
…is contracted (#1867)

#1857 refused any ledger whose reporting_configurations has offering_id.
The rolling-safe upgrade on next (#1856) adds that column in an expand
step but keeps the four-column generation key that 0.1.x writes, until
the operator contracts. So 0.1.0b5 would refuse to restart mid-rollout,
though it otherwise runs on the expand state unchanged.

Refuse only when the configuration primary key includes offering_id,
i.e. once the ledger is contracted. That is the same test the packaged
SQL's key check makes.

Refs #1856

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* chore(main): release adcp-reporting-storage 0.1.0-beta.5

* chore: normalize prerelease version to PEP 440

---------

Co-authored-by: aao-ipr-bot[bot] <279263549+aao-ipr-bot[bot]@users.noreply.github.com>
…s resolve on ubuntu-26.04 runners (#1890)

GitHub's ubuntu-latest moved to ubuntu-26.04, whose system zone database
drops the legacy links (US/Pacific, PST8PDT). zoneinfo then raises for them,
so tests/test_reporting_period_date.py::test_legacy_zone_aliases_resolve_to_the_canonical_date
fails on every 9.x run. Pin tzdata 2026.6 in the dev extra and the dev group,
as next already does, so zoneinfo falls back to the IANA package.

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
…ated invocation that carries both members (#1889)

* fix(server): read the A2A profile's input member, and refuse parameters when the profile is activated

The default A2A parser read only an invocation's `parameters` member. A client
following the AdCP A2A profile (@adcp/sdk 16.7.0 on any A2A 1.0 agent) sends
`{skill, input}`, so handlers received `{}`: `context` was not echoed and
versioned tools failed with VERSION_UNSUPPORTED.

The parser now reads `parameters` when present, otherwise `input`, and never
merges them, so a caller that does not activate the profile is dispatched
exactly as before. A request that activates the profile (A2A-Extensions, or the
Message's extensions, names https://adcontextprotocol.org/extensions/adcp/v3)
must carry exactly one DataPart with an `input` object; `parameters`, alone or
beside `input`, is refused with INVALID_REQUEST before handler entry, as the
profile requires. The raw-body receipt path reads the same member.

Backport of 3f97994 from next.

Refs #1887

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* test(server): cover the A2A input member and the profile's parameters refusal over the real app

Refs #1887

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(server): treat a hand-built A2A context as not activating the profile

ADCPAgentExecutor.execute is public, and a context built by hand (an
adopter's executor test) may lack RequestContext.requested_extensions or
Message.extensions. The profile check now reads both defensively instead of
raising AttributeError.

Refs #1887

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(server): keep dispatching an activated parameters-only A2A invocation on 9.x, with a deprecation warning

9.x is the stable line, so a patch must not start refusing requests that work
today. An invocation that activates the AdCP A2A profile but carries only
`parameters`, a missing or non-object `input`, or a DataPart count other than
one is still dispatched as before, and a deprecation warning names the profile
rule. Only an activated invocation carrying both `parameters` and `input` is
refused, since it is ambiguous and no compliant client sends it. adcp 10
refuses every non-conforming activated shape.

Refs #1887

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* test(server): cover the 9.x warn-and-dispatch shapes for an activated A2A invocation

Refs #1887

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: aao-ipr-bot[bot] <279263549+aao-ipr-bot[bot]@users.noreply.github.com>
BREAKING CHANGE: Webhook challenge helpers now require seller_agent_url and event_types, and proof challenges require an RFC 9421 WebhookSender; pass legacy delivery authentication alongside that sender.
…ebhook-payload-fix

fix(webhooks)!: comply with AdCP 3.2 challenge schema
…count-readiness-fix

fix(reporting): resolve dynamic status account readiness
@aao-secretariat

Copy link
Copy Markdown

Ladon cannot review this PR until merge conflicts are resolved.

@rachitm022

Copy link
Copy Markdown
Collaborator Author

Superseded by a conflict-resolved synchronization branch based on next; the direct main head cannot resolve the divergent-line conflicts.

@rachitm022 rachitm022 closed this Nov 15, 2026

This branch was previously deployed

1 inactive deployment
github-pages — 39a5a0a7 Deployed Nov 15, 2026 by rachitm022 via docs #1019
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.

2 participants