Hacktoberfest 2026：维护者为十月标记出来的 issue，仍然开放、适合新手。 浏览 Hacktoberfest issue

Record missing audit events for 13 authorization lifecycle transitions

未关闭
  • 哪里不对？
#876 1 条评论 0 个 reaction 已指派 0 人 在 GitHub 查看

维护者通常 1 天内回复

还没有人认领这个 Issue。

评估

难度
4/5
预计耗时
3-5 天
新手友好度
43/100
Issue 类型
功能
描述清晰度
描述清楚
活跃度
活跃
技术栈
typescript
领域
backend-api-design, databases, security

调研方向

Start with server/src/plugins/route-groups/connections.ts, the plugin store writes, and the historical ledger server/tests/pending-authorization-audit-ledger.test.ts at the linked commit; then inspect current behavior tests for each transition. Add persisted audit assertions for the 13 listed gaps, checking actor, attempt correlation, and safe payload while avoiding secret identifiers and duplicate ending events. Done means every transition has a behavior test that verifies its persisted audit event.

由索引模型根据 Issue 内容生成。

描述

PR #872 removes a source-scanning audit ledger that recorded known gaps but did not prove events fired. This issue preserves its 13 debt entries, as requested in review. These are existing gaps, not new regressions from the test cleanup.

Source ledger: pending-authorization-audit-ledger.test.ts at c19d0653. Most transitions are in server/src/plugins/route-groups/connections.ts; store writes are in server/src/plugins/plugin-store.

Dusty confirmed all 13 gaps are still current. The three items marked Priority below should be addressed first: consent can remain live at the vendor with only a console line as the record.

Missing lifecycle audit coverage
  • Claim (claimAuthorizationAttempt, B5e): record who started which consent attempt.
  • Expiry overwrite/re-claim (same call, B5e): record the previous attempt's ending before ON CONFLICT replaces it.
  • Discard after vendor mint refusal (arcade-refused-attempt-not-cleared, B5e): record the refusal/ending, not only cleanup-failure logging.
  • Priority — Discard after successful mint cannot be recorded (arcade-mint-not-recorded, B5e): record that a consent flow may remain live at the vendor.
  • Mint recorded (recordAuthorizationMint, authId: minted.authorizationId, B5e): audit the attempt being bound to a vendor authorization without exposing its secret identifiers.
  • Vendor reports failed consent (failAuthorizationAttempt, settled.status === "failed", B5e): record the terminal verdict.
  • Completed consent has the wrong identity (failAuthorizationAttempt, settled.userId !== attempt.vendorUserId, B5e): record the attribution refusal.
  • Context changed and vendor withdrew retired grant (markAuthorizationWithdrawn, B5e): audit the attempt's verdict; account-disconnection rows alone do not describe it.
  • Priority — Person no longer permitted to act; grant not withdrawn (markAuthorizationUnreconciled, arcade-completion-not-permitted, B5e): record the closed attempt and potentially live vendor consent.
  • Priority — Context changed and withdrawal unconfirmed (markAuthorizationUnreconciled, B5e): record the unresolved grant rather than only a console line.
  • App/server removal cascade (pending_authorizations_server_id_mcp_servers_id_fk, B5e): identify affected in-flight attempts; configuration-change rows currently describe released grants only.
  • Actor removal cascade (pending_authorizations_actor_id_users_id_fk, B5e): identify attempts that die with the actor, including potentially live vendor consent.
  • Cancel loses race to completion (That authorization finished before it could be cancelled, B2c): audit the refused cancellation (409) even though no pending row is deleted.
Acceptance

Exercise each transition through behavior tests and assert the persisted audit event, actor, attempt correlation, and safe payload. Do not log bearer consent URLs or live authorization IDs. Keep the existing successful completion and cancellation events; avoid duplicate attempt-ending events when a cancelled attempt is encountered by a late mint/completion. The original ledger waived those two late-return branches because cancellation had already been audited.

Re-check each gap against current code when implementing; the list preserves the historical findings rather than claiming a new runtime reproduction of all 13.

主要语言
TypeScript
星标
7.1k
派生
962
平均合并
1 天 6 小时
30 天内合并 PR
143

环境准备

  • 提供 Dockerfile 或 Docker Compose 文件
  • 有 Pull Request 模板
  • 没有贡献指南

从这里开始

  1. 先读完整个 Issue，再读项目的贡献指南。
  2. 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
  3. Fork 仓库，在一个分支上完成修改。
  4. 提交 Pull Request，并在描述里引用这个 Issue 编号。

CopilotKit/OpenBot 的其他 Issue

查看 CopilotKit/OpenBot 的全部 Issue

相似的 Issue

更多 TypeScript Issue

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。