Skip to content

QQ tracer: actual image understanding and original-group image results #1157

Closed

Description

@DoodleBears

v1.3 release reconciliation — 2026-10-11

Target release: v1.3. The implementation/evidence recorded below was merged into the existing #1156 QQ feature branch, not main. Added native blocker #1156 for default-branch stack promotion and current-product qualification; preserve the fulfilled #1153 foundation and all existing owner/evidence records. Do not reimplement the accepted media slice or close it merely because the child PR was merged to a dependency branch. This is a release/integration dependency, not a new claim or authorization to merge.


Parent

QQ group chat specification #1149

What to build

Mention the Bot with a synthetic image in a QQ group, inspect that image in the linked local Channel, have the Bot actually interpret visual content through a supported model/tool, and return an authorized image result through its own QQ app to the original group.

Use the existing checked source-media and canonical Attachment owners and shared Human display patterns. Image preview and actual model understanding are separate evidence. Extend only the necessary QQ acquisition/native output and model-input boundary in this slice.

Acceptance criteria

  • Real QQ image input preserves its native message, caption/mention, content association and authenticated account/conversation provenance through canonical intake. The Provider owns private URLs, access material and checked acquisition.
  • The linked Channel displays the authorized received image in its original message using shared media components; Inbox-only and mentions-only policy remain effective. Human preview alone creates no Bot Observation, admission or wake.
  • Use synthetic content requiring actual visual interpretation. A supported live model/tool demonstrably receives the image content and produces the correct interpretation; text captions, fixture outputs and Human screenshots alone do not qualify this criterion.
  • The Bot produces an authorized image result and returns it once through a checked native operation to the same QQ group using its own identity. Independently inspect the native image and receipt, not just a transport call or result URL.
  • Recheck authority before acquisition, model access, local serving and external output. Revocation/block/unbind and expired resources do not bypass checks through raw public URLs or another Bot's identity.
  • Existing bounds govern resource size, decode/acquisition budgets, cancellation and output. Known original metadata remains distinct from validated/derived metadata; unsupported formats and unavailable image-model capability appear as explicit states.
  • Canonical media facts retain message association across replay/refresh/restart without a separate image history store. Native unknown output is not auto-retried.
  • Public-seam regression covers source identity, mention scope, authorized acquisition/display/model access, budget failure, cancellation/revocation, result destination and checked receipts. Preserve existing provider media behavior.
  • Real QQ/model/Client acceptance includes original image display, meaningful visual interpretation and the independently inspected result image. Record model capability and exact Provider/SDK/DSH/artifact revisions.
  • Supply Client interaction evidence, affected bilingual guides/ledgers and required checks. Record any genuinely new model-media architecture seam and integrate its owning docs; submit reusable Provider work through the maintained fork and upstream.
  • An unavailable official image operation or model capability leaves the full image slice unfinished with a clear reason. This ticket neither qualifies Lark/Slack vision retroactively nor adds video support.

Blocked by

Shared media specification #1020 and completed Lark image-display work #1021 are prior art; unrelated provider media tickets are not blockers.

Activity

  1. added this to the v1.2 milestone on Oct 7, 2026
  2. added
    ready-for-agentFully specified, ready for an AFK agent
    area:fullstackEnd-to-end Host and Client implementation
    feature:messagingMessaging module and external providers
    scope:tracerHuman-testable implementation slice
    on Oct 7, 2026
  3. DoodleBears commented on Oct 8, 2026

    @DoodleBears
    ContributorAuthor

    Agent work claim

    • Agent: Codex
    • Task: codex/local/01a117bf-865d-72d1-a9c6-88658c59f844 (not a DSH Session)
    • Issue scope: QQ tracer: actual image understanding and original-group image results #1157 — one official QQ group image path through checked source media, canonical Attachment authority, actual Client preview, actual supported model/tool interpretation, and an authorized own-app image result in the original group.
    • Branch: codex/1157-qq-images
    • Started: 2026-10-08 08:45 UTC
    • State: active

    Stacked from qualified text/conversation work and the final restart correction (3b2be95). #1156's native Echo gap remains recorded in its own Draft PR and is not an image capability claim. Reuse the already approved public checked-contract → Source Event / Bot Inbox → Orchestrator → original-group result test boundary, extending it only to source-media authorization, canonical display/model access and checked image output. Preview alone does not qualify model understanding. No ordinary-message intake or outgoing voice is included.

    The maintained Provider starting point is 71fa2a2; SDK/native Host versions and actual image-model capability will be inspected before the smallest production path is extended. Any reusable Provider work will have its own fork branch and upstream contribution. No merge is authorized or performed.

  4. added
    agent:activeOne or more coding-agent tasks currently claim this issue; see claim comments for task IDs
    on Oct 8, 2026
  5. DoodleBears commented on Oct 8, 2026

    @DoodleBears
    ContributorAuthor

    QQ image qualification checkpoint

    Task: codex/local/01a117bf-865d-72d1-a9c6-88658c59f844; product c1db0d42, maintained Provider 876087ba.

    Text and image share one durable reply intent per Bot/source. A source that already settled text correctly refuses a different image payload; the successful live source chose image output first. Repeated identical requests return their stored result.

  6. DoodleBears commented on Oct 8, 2026

    @DoodleBears
    ContributorAuthor

    Agent work outcome

    • Task: codex/local/01a117bf-865d-72d1-a9c6-88658c59f844
    • State: completed for this image implementation/merge scope.
    • PR: #1189, merged at 6ca997f2 into dependency branch codex/1156-qq-own-identities, on the Human's explicit instruction. Default-branch promotion follows the existing PR stack.
    • Evidence: real native image intake, model read_image understanding, independent result image, checked native receipt and Human receipt confirmation; complete product Linux CI passes.
    • Remaining record: matched dark baseline and continuous recording were not captured; the Human accepted merge with the documented limits. Windows full-suite failures are not baseline-qualified; no full Windows success is claimed. Maintained Provider and upstream contribution remain separate artifacts.
    • Continuing sibling scope: file candidate #1193, awaiting native QQ file qualification. This outcome does not close the parent specification or declare the entire QQ integration complete.
  7. removed
    agent:activeOne or more coding-agent tasks currently claim this issue; see claim comments for task IDs
    on Oct 8, 2026
  8. modified the milestones: v1.2, v1.3 on Oct 11, 2026
  9. DoodleBears commented on Oct 13, 2026

    @DoodleBears
    ContributorAuthor

    Agent work outcome — accepted delivery merged

    • Agent: Codex
    • Task: codex/local/01a1342d-5ce4-7872-9f9e-492a9caee393 (not a DSH Session)
    • State: completed for delivery/acceptance and default-branch promotion. Earlier implementation outcomes and their contributors remain part of the record.
    • Product #1505 merged into main at b5285b610393df2531be6c386d15656bb45ca947. Reviewed head 41c3a1c7 passed full Linux verify and package preparation.
    • Maintained Provider #26 merged at b283be11 into its declared dependency branch, not the default branch; no Provider package/release or production cutover is claimed.
    • Human acceptance decision: native own Echo is deferred, not passed. No own-text-echo capability is advertised. Tencent contract inquiry #8 remains open.
    • Evidence and exact revision boundaries: fresh K qualifies recovered runtime 2edb56d5 / Provider 1eee207c / SDK1.0.4 / DSH RC2. The merge does not retroactively qualify the later main composition. Earlier capture/Windows limits remain disclosed.

    The already accepted image implementation (#1189) is now promoted to main. Fresh actual native read_image/image-copy/native receipt evidence and Human original-group inspection confirm blue / BLUE / code 68241. Image preview enlargement, Escape and focus-return pass. No other platform/model capability is inferred.

  10. moved this from Todo to Done in Roadmapon Oct 13, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    area:fullstackEnd-to-end Host and Client implementationfeature:messagingMessaging module and external providersready-for-agentFully specified, ready for an AFK agentscope:tracerHuman-testable implementation slice

    Type

    No type

    Projects

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions