GitLab

Closes #1957

The OAuth path forwards client requests to chatgpt.com/backend-api/codex/responses,
where applyCodexOAuthTransform forces store=false (chatgpt.com's codex backend
rejects store=true). Reasoning items emitted under store=false are NEVER
persisted upstream, so any rs_* reference that a client carries forward in a
subsequent input[] array triggers a guaranteed upstream 404:

    Item with id 'rs_...' not found. Items are not persisted when `store` is
    set to false. Try again with `store` set to true, or remove this item
    from your input.

sub2api wraps this as 502 "Upstream request failed" and the conversation
breaks on every multi-turn /v1/responses request that uses reasoning + tools
(reproducible with gpt-5.5; gpt-5.4 happens to dodge it because the upstream
does not emit reasoning items for that model).

Affected clients include any that follow the OpenAI Responses API spec and
replay prior assistant items verbatim — in practice this hit OpenClaw and
similar agent harnesses on every turn ≥2 with tool use.

The fix: in filterCodexInput, drop input items with type == "reasoning"
entirely. The model never reads reasoning summary text from input (only
encrypted_content can carry reasoning context across turns, and chatgpt.com
under store=false does not emit it), so this is a no-op for the model itself
and a clean removal of unreachable upstream lookups.

Scope is intentionally narrow:
  * Only OAuth account requests (account.Type == AccountTypeOAuth) reach
    applyCodexOAuthTransform / filterCodexInput.
  * API-key accounts going to api.openai.com/v1/responses are unaffected
    (store=true works there, rs_* persists, multi-turn already works).
  * Anthropic / Gemini platform groups go through different transforms and
    are unaffected.
  * /v1/chat/completions is unaffected (no reasoning items).
  * item_reference items (different type) are unaffected — only type ==
    "reasoning" is dropped.

Verification:
  * Existing tests pass: go test ./internal/service/ -run Codex|Tool|OAuth
  * New regression test asserts reasoning items are dropped under both
    preserveReferences=true and preserveReferences=false.
  * End-to-end repro on gpt-5.5 multi-turn + tools: pre-patch 502, post-patch
    200. Repro on gpt-5.4 unchanged. Three-turn deep loop on gpt-5.5 passes.

对称参照 Claude BetaPolicy 的 fast-mode 过滤实现，新增针对 OpenAI 上游
service_tier 字段（priority / flex，含客户端 "fast" → "priority" 归一化）的
pass / filter / block 三态策略，覆盖全部 OpenAI 入口 + admin 配置入口。

后端核心
- 新增 SettingKeyOpenAIFastPolicySettings、OpenAIFastPolicyRule、
  OpenAIFastPolicySettings 配置模型，含规则的 service_tier × action × scope
  × 模型白名单 × fallback action 维度。
- SettingService.Get/SetOpenAIFastPolicySettings；缺失时返回内置默认策略
  （所有模型的 priority 走 filter，whitelist 为空，fallback=pass）。设计
  依据：service_tier=fast 是用户级开关，与 model 字段正交，默认锁定特定
  model slug 会留下"用 gpt-4 + fast 透传 priority 上游"的绕过路径。JSON
  解析失败不再静默 fallback，slog.Warn 记录脏数据，便于运维定位。
- service_tier 归一化（trim + ToLower + fast→priority + 白名单 priority/flex）
  与策略评估（evaluateOpenAIFastPolicy）作为唯一真实来源，HTTP / WS 共用。
  抽出纯函数 evaluateOpenAIFastPolicyWithSettings，配合 ctx-bound settings
  快照（withOpenAIFastPolicyContext / openAIFastPolicySettingsFromContext），
  WS 长会话入口预取一次后所有帧复用，避免每帧打到 settingService。

HTTP 入口（4 个）
- Chat Completions、Anthropic 兼容（Messages，含 BetaFastMode→priority 二次
  命中）、原生 Responses、Passthrough Responses 全部接入
  applyOpenAIFastPolicyToBody，filter 走 sjson 顶层删除 service_tier，block
  返回 403 forbidden_error JSON。
- 4 入口统一使用 upstream 视角的 model（GetMappedModel +
  normalizeOpenAIModelForUpstream + Codex OAuth normalize 后的 slug），
  避免 chat/messages/native /responses/passthrough 因为 model 维度不同
  造成 whitelist 命中差异。
- 在 pass 路径也把客户端 "fast" 别名归一化为 "priority" 写回 body，
  否则 native /responses 与 passthrough 入口会把 "fast" 原样透传给上游
  导致 400/拒绝（chat-completions 入口的 normalizeResponsesBodyServiceTier
  此前已具备同等行为）。

WebSocket 入口
- 新增 applyOpenAIFastPolicyToWSResponseCreate：严格匹配
  type="response.create"，仅处理顶层 service_tier；filter 用 sjson 删字段，
  block 返回 typed *OpenAIFastBlockedError。
- ingress 路径在 parseClientPayload 内调用，block 命中先 Write Realtime
  风格 error event 再返回 OpenAIWSClientCloseError(StatusPolicyViolation
  =1008)，依赖底层 WebSocket Conn.Write 的同步 flush 保证 error 先于
  close。
- passthrough 路径在 RunEntry 前对 firstClientMessage 应用策略，并通过
  openAIWSPolicyEnforcingFrameConn 包装 ReadFrame 对每个 client→upstream
  帧执行策略；后续帧无 model 字段时回退到 capturedSessionModel。
  filter 闭包内同时侦测 session.update / session.created 帧的 session.model
  字段刷新 capturedSessionModel，封堵"首帧 model=gpt-4o（pass）→
  session.update 改为 gpt-5.5 → 不带 model 的 response.create fallback
  到 gpt-4o"的 mid-session 绕过路径。
- passthrough billing：requestServiceTier 在策略 filter 之后再从
  firstClientMessage 提取，filter 命中时 OpenAIForwardResult.ServiceTier
  上报 nil（default tier），与 HTTP 入口（reqBody 来自 post-filter map）
  / WS ingress（payload 来自 post-filter bytes）的语义一致。
- 错误事件 schema：{event_id: "evt_<32hex>", type: "error",
  error: {type: "forbidden_error", code: "policy_violation", message}}，
  与 OpenAI codex 客户端 error event 解析兼容。

Admin / Frontend
- dto.SystemSettings / UpdateSettingsRequest 新增
  openai_fast_policy_settings 字段（omitempty），bulk GET/PUT 接入。
- Settings 页 Gateway 页签新增 Fast/Flex Policy 表单卡片：
  service_tier × action × scope × 模型白名单 × fallback action 全字段配置。
- 前端守门：openaiFastPolicyLoaded 标志仅在 GET 真带回字段时才允许回写，
  避免 rollout/错误把默认规则覆盖成空；saveSettings 回写循环 skip 该字段，
  由专用刷新逻辑处理；仅 action=block 时发送 error_message，匹配后端
  omitempty 行为。

测试
- HTTP 路径：openai_fast_policy_test.go 覆盖默认配置（whitelist=[]，所有
  模型 priority filter）/ block 自定义错误 / scope 区分 / filter 删字段 /
  block 不改 body / block 短路上游 / Anthropic BetaFastMode 触发 OpenAI
  fast policy 等场景。
- WebSocket 路径：openai_fast_policy_ws_test.go 覆盖
    helper 单元（filter / fast→priority 归一化 / flex 透传 / block typed
    error / 无 service_tier 字节不变 / 非 response.create 帧不动 / 空 type
    帧不动 / event_id+code 字段断言 / 非字符串 service_tier 容错）+
    pass 路径 fast 别名归一化回归 +
    ingress 端到端（filter 后上游不含 service_tier / block 后客户端先收
    error event 再收 close 1008 且上游 0 写）+
    passthrough capturedSessionModel fallback 用例（whitelist 策略下首帧
    建立、缺 model 命中 fallback、缺少 fallback 时的 leak 文档化）+
    passthrough session.update / session.created 旋转 capturedSessionModel
    的 mid-session 绕过回归 +
    passthrough billing post-filter ServiceTier 与 idempotent filter 回归。
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

  - 修复返利不到账的根因：tryClaimAffiliateRebateAudit 中 PostgreSQL 参数类型推断冲突
  - 补全 OAuth 注册路径（LinuxDo/OIDC/WeChat/Pending Flow）的邀请码绑定
  - 前端 OAuth 注册页面传递 aff_code 参数
  - 新增返利冻结期机制：可配置冻结时间，到期后自动解冻（懒解冻）
  - 新增返利有效期：绑定后 N 天内有效，过期不再产生返利
  - 新增单人返利上限：超出上限部分精确截断
  - 增强返利流程 slog 结构化日志，便于排查问题
  - 已邀请用户列表增加返利明细列

PR #1914 unconditionally applied the full mimicry pipeline to all OAuth
accounts, including real Claude Code CLI clients. This replaced the
client's long system prompt (~10K+ tokens with stable cache_control
breakpoints) with a short ~45 token [billing, CC prompt] pair, which
falls below Anthropic's 1024-token minimum cacheable prefix threshold.
The result: every request created a new cache but never hit an existing
one.

Fix: restore the Claude Code client detection gate so that real CC
clients bypass body-level mimicry (system rewrite, message cache
management, tool name obfuscation). Non-CC third-party clients
(opencode, etc.) continue to receive full mimicry.

Also harden the detection logic:
- Make UA regex case-insensitive (align with claude_code_validator.go)
- Validate metadata.user_id format via ParseMetadataUserID() instead of
  just checking non-empty, preventing third-party tools from spoofing
  a claude-cli/* UA with an arbitrary user_id string to bypass mimicry

- gofmt: realign AffiliateDetail struct tags in affiliate_service.go
- ineffassign: remove dead seenCompleted assignment before return in account_test_service.go

- 在系统设置「功能开关」中新增邀请返利总开关，默认关闭；
  关闭态：菜单隐藏、注册忽略 aff、新充值不返利，但已有 quota 仍可转余额
- 支持管理员为指定用户设置专属邀请码（覆盖随机码，全局唯一）
- 支持管理员为指定用户设置专属返利比例（覆盖全局比例，可单条/批量调整）
- 在系统设置邀请返利卡片内嵌入专属用户管理表格（搜索/编辑/批量/删除），
  删除采用项目通用 ConfirmDialog，会同时清除专属比例并把邀请码重置为系统随机码
- /affiliate 用户页新增「我的返利比例」卡片与动态使用说明，让用户直观看到
  分享后能拿到多少（同源 resolveRebateRatePercent 计算，与实际充值一致）
- 新增数据库迁移 132 添加 aff_rebate_rate_percent 与 aff_code_custom 列
- 新增 admin 路由组 /api/v1/admin/affiliates/users/* 共 5 个端点
- AffiliateService 改为只依赖 *SettingService，去除冗余的 SettingRepository
- 邀请码格式校验放宽到 [A-Z0-9_-]{4,32}，兼容旧 12 位系统码与新自定义码
- 补充单元测试与集成测试覆盖新方法、冲突路径与边界值

将 vansour/sub2api#1555 的 OpenAI compact 能力建模手工移植到当前 main：账号
级 compact 状态/auto-force_on-force_off 模式、compact-only 模型映射、调度器
tier 分层（已支持 > 未知 > 已知不支持）、管理后台 compact 主动探测，以及对应
i18n/状态徽章。普通 /responses 流量行为不变，无数据库迁移。

The hardcoded codex CLI version (0.104.0) causes upstream rejection
when using gpt-5.5 with compact, as the server treats the request
as an outdated client and returns 400/502.

Update codexCLIVersion, codexCLIUserAgent, and openAICodexProbeVersion
to 0.125.0 to match the current Codex CLI release.

Fixes #1933, #1887, #1865
Related: #1609, #1298, #849

- staticcheck QF1001: apply De Morgan's law to the OAuth-mimic header
  passthrough guard (`!(a && b)` → `a != ... || !b`).
- unused: drop `isClaudeCodeRequest`, which became dead after PR #1914
  switched both `/v1/messages` and `/count_tokens` paths to unconditional
  `account.IsOAuth()` mimicry. The lowercase helper `isClaudeCodeClient`
  is kept (still referenced by `TestIsClaudeCodeClient`).

- Drop SetAffiliateService setters and ProvideAuthService /
  ProvidePaymentService / ProvideUserHandler wrappers in favor of direct
  Wire constructor injection. AffiliateService has no back-edge to
  Auth/Payment/User, so the indirection was never required.
- Change RegisterWithVerification's variadic affiliateCode to a fixed
  parameter; adjust all call sites.
- Validate aff_code length and charset in BindInviterByCode before any
  DB lookup, eliminating timing-side-channel and useless DB roundtrips
  on malformed input.
- Make affiliate cache invalidation synchronous; surface Redis errors
  via the project logger instead of swallowing them in a detached
  goroutine.
- Add an integration test guarding cross-layer tx propagation in
  AccrueQuota and a unit test pinning the aff_code format rules.

fix(apicompat): recognize web_search_20250305 / google_search in Responses to Anthropic tool conversion

Root cause of persistent third-party detection: sub2api's
buildUpstreamRequest transparently forwards client headers via
allowedHeaders whitelist (addHeaderRaw) before applying mimicry
overrides. When third-party clients (opencode, etc.) send their own
anthropic-beta / user-agent / x-stainless-* / x-claude-code-session-id
values, these get appended to the request alongside our injected
headers, creating an inconsistent header set that Anthropic detects.

Parrot's build_upstream_headers constructs exactly 9 headers from
scratch and never forwards anything from the client. This is why
'same opencode version, some users work some don't' — different
opencode configs/versions send different header combinations.

Fix: when tokenType=oauth and mimicClaudeCode=true, skip the
client header passthrough loop entirely. The subsequent
applyClaudeCodeMimicHeaders + ApplyFingerprint + beta merge
pipeline constructs all necessary headers from our controlled values.

Also: remove systemIncludesClaudeCodePrompt gate — OAuth accounts
now unconditionally rewrite system (even if client already sent a
Claude Code-style prompt), ensuring billing attribution block is
always present.

Before: isClaudeCodeRequest() checked whether the client looks like a
real Claude Code CLI (UA, system prompt, X-App header, metadata format).
If it looked like Claude Code, all mimicry was skipped — the assumption
being that a real CLI needs no help.

Problem: third-party tools like opencode partially impersonate Claude
Code (sending claude-cli UA + claude-code beta + CC system prompt) but
miss critical details (billing attribution block, tool-name obfuscation,
cache breakpoints, full beta set). Some users' opencode instances pass
the isClaudeCodeRequest check, causing sub2api to skip mimicry entirely,
while Anthropic still detects the request as third-party.

This explains why 'same opencode version, some users work, some don't'
— it depends on which opencode features/config trigger the validator.

Fix: OAuth accounts now unconditionally run the full mimicry pipeline,
matching Parrot's behavior (Parrot never checks client identity).
This is safe because our mimicry is strictly more complete than any
third-party client's partial impersonation.

Changed:
  - /v1/messages path: remove isClaudeCode gate
  - /v1/messages/count_tokens path: same

The previous commit only wired stripMessageCacheControl,
addMessageCacheBreakpoints, and tool-name obfuscation into
applyClaudeCodeOAuthMimicryToBody (used by /chat/completions and
/responses). The native /v1/messages path and count_tokens path
have their own independent mimicry code blocks and were missed.

Now all three entry points share the same D/E/F pipeline:
  - /v1/messages (gateway_service.go forwardAnthropic)
  - /v1/messages/count_tokens (gateway_service.go countTokens)
  - OpenAI compat (applyClaudeCodeOAuthMimicryToBody)

Implements the remaining three parity items with Parrot cc_mimicry:

  D) Tool-name obfuscation
     - Dynamic mapping when tools.length > 5 (matches Parrot threshold).
       Fake names follow {prefix}{name[:3]}{i:02d} (e.g. 'manage_bas00').
       Go port of random.Random(hash(tuple(names))) uses fnv64a seed +
       math/rand; byte-exact reproduction is impossible (Python hash vs
       Go hash), but the two invariants that matter are preserved:
         * same input tool_names yield identical mapping (cache hit)
         * prefix pool is shuffled (names look distributed)
     - Static prefix map (sessions_ -> cc_sess_, session_ -> cc_ses_)
       applied as fallback, matching Parrot TOOL_NAME_REWRITES verbatim.
     - Server tools (web_search_20250305, computer_*, etc.) are NOT
       renamed; only type=='function' and type=='custom' tools are.
     - tool_choice.name is rewritten in sync (only when type=='tool').
     - Response side: bytes-level replace on every SSE chunk / JSON
       body at 6 injection points (standard stream/non-stream,
       passthrough stream/non-stream, chat_completions stream +
       non-stream, responses stream + non-stream). Reverse mapping
       applied longest-fake-name-first to prevent substring conflicts
       (parity with Parrot _restore_tool_names_in_chunk).
     - tool_choice is no longer unconditionally deleted in
       normalizeClaudeOAuthRequestBody — Parrot passes it through.

  E) tools[-1] cache_control breakpoint
     - Injected as {type:ephemeral, ttl:<DefaultCacheControlTTL>} when
       the last tool has no cache_control. Client-provided ttl is
       passed through unchanged (repo-wide policy).

  F) messages cache_control strategy
     - stripMessageCacheControl removes every client-provided
       messages[*].content[*].cache_control (multi-turn stability).
     - addMessageCacheBreakpoints then injects two stable breakpoints:
       (1) last message, and (2) second-to-last user turn when
       messages.length >= 4.
     - Combined with the system block breakpoint and tools[-1]
       breakpoint, this gives exactly the 4 breakpoints Anthropic
       allows per request.

Non-trivial implementation details to be aware of when rebasing:

  * Two new files, no upstream collision:
      gateway_tool_rewrite.go       (D + E algorithms)
      gateway_messages_cache.go     (F strip + breakpoints)
  * Two new feature calls bolted onto the tail of
    applyClaudeCodeOAuthMimicryToBody in gateway_service.go — rebase
    conflicts will be ~10 lines maximum.
  * Response-side injection points all wrap their existing write with
    reverseToolNamesIfPresent(c, ...), preserving original behavior
    when no mapping is stored (static prefix rollback still runs).
  * Non-stream chat/responses switched from c.JSON to
    json.Marshal + c.Data so bytes-level replace is possible.
  * Retry bodies (FilterThinkingBlocksForRetry,
    FilterSignatureSensitiveBlocksForRetry, RectifyThinkingBudget)
    only prune blocks — they preserve the already-obfuscated tool
    names, so no extra mapping re-application is needed.

Manual QA: end-to-end scenario verified with 6 tools (above threshold)
and tool_choice.type=='tool'. Obfuscation + restore roundtrip shown
in test logs; then removed the temp test file.

Tests (16 new):
  - buildDynamicToolMap stability + below-threshold guard
  - sanitizeToolName precedence (dynamic > static)
  - restoreToolNamesInBytes longest-first + static rollback
  - applyToolNameRewriteToBody skips server tools + syncs tool_choice
  - applyToolsLastCacheBreakpoint defaults to 5m + passes client ttl
  - stripMessageCacheControl + addMessageCacheBreakpoints in the
    1/4/string-content cases + second-to-last user turn selection
  - buildToolNameRewriteFromBody ReverseOrdered is desc-by-fake-length
  - fake name shape follows Parrot {prefix}{head3}{i:02d}

Three field-level alignments in normalizeClaudeOAuthRequestBody to
match real Claude Code CLI traffic byte-for-byte:

  1. temperature: previously deleted unconditionally; now passes
     through client value, defaults to 1 when absent (real CLI
     always sends temperature, default 1).

  2. max_tokens: defaults to 128000 when absent (real CLI default).

  3. context_management: when thinking.type is enabled/adaptive
     and the client did not provide context_management, inject
     {"edits":[{"type":"clear_thinking_20251015","keep":"all"}]}
     to mirror real CLI behavior.

tool_choice removal is unchanged (Claude Code OAuth credentials
do not allow client-supplied tool_choice).

Tests updated:
  - gateway_body_order_test.go: temperature/max_tokens are now
    expected in output; tool_choice still removed.
  - gateway_prompt_test.go: system array is now 2 blocks
    (billing + cc prompt), assertions adjusted.
  - gateway_anthropic_apikey_passthrough_test.go: same 2-block
    assertion.

Real Claude Code CLI always sends a 2-block system array:

  [0] {"type":"text", "text":"x-anthropic-billing-header: cc_version=X.Y.Z.{fp}; cc_entrypoint=cli; cch=00000;"}
  [1] {"type":"text", "text":"You are Claude Code...", "cache_control":{...}}

Before this commit, sub2api's mimicry path only produced block [1].
The missing billing block is one of the primary third-party detection
signals Anthropic uses for Claude-Code-scoped OAuth tokens.

New file gateway_billing_block.go ports the fingerprint algorithm
(byte-for-byte from Parrot cc_mimicry.py:compute_fingerprint):
pick chars at positions [4,7,20] of the first user text, then
`sha256(SALT + chars + cc_version)[:3]`.

  - claude/constants.go: CLICurrentVersion = "2.1.92" (must match UA)
  - gateway_billing_block.go: computeClaudeCodeFingerprint +
    buildBillingAttributionBlockJSON + extractFirstUserText
  - gateway_service.go: rewriteSystemForNonClaudeCode now emits both
    blocks in order; cch=00000 is filled in later by
    signBillingHeaderCCH in buildUpstreamRequest.

Downstream compat note: syncBillingHeaderVersion's regex
`cc_version=\d+\.\d+\.\d+` only matches the semver triple,
leaving the `.{fp}` suffix intact when rewriting in buildUpstreamRequest.

Real Claude CLI traffic sends cache_control as
`{"type":"ephemeral","ttl":"1h"}`. Our previous payload only
sent `{"type":"ephemeral"}`, which is a bytewise mismatch with
the official CLI and one more third-party detection signal.

Policy: client-provided ttl is always passed through unchanged.
Proxy-generated cache_control blocks default to 5m (vs Parrot's 1h)
to avoid burning the 1h cache budget on automatic breakpoints while
still aligning with the `ttl` field being present.

  - claude/constants.go: DefaultCacheControlTTL = "5m"
  - apicompat/types.go: new AnthropicCacheControl type with TTL field;
    AnthropicTool gains optional CacheControl pointer so the mimicry
    path can attach a cache breakpoint to tools[-1] later.
  - service/gateway_service.go: anthropicCacheControlPayload gains TTL;
    marshalAnthropicSystemTextBlock and rewriteSystemForNonClaudeCode
    emit ttl=5m by default.

The previous commit added FullClaudeCodeMimicryBetas() but the two
call sites in buildUpstreamRequest still hardcoded the old 3-token
subset. Anthropic now checks the complete set of beta tokens to
decide if a request qualifies as Claude Code. Wire them up:

  - /v1/messages mimic path: requiredBetas = FullClaudeCodeMimicryBetas()
  - /v1/messages/count_tokens mimic path: same + BetaTokenCounting

Haiku models keep the 2-token exemption (BetaOAuth + InterleaveThinking).

Before: the OpenAI-compat forwarders only called injectClaudeCodePrompt,
which prepends the Claude Code banner but leaves the rest of the body
in its original non-Claude-Code shape. The codebase already admits this
is insufficient (see the comment on rewriteSystemForNonClaudeCode in
gateway_service.go: "仅前置追加 Claude Code 提示词无法通过检测").

Effect: OAuth accounts served through /v1/chat/completions or /v1/responses
were detected as third-party apps and bled plan quota with:

    Third-party apps now draw from your extra usage, not your plan limits.

Fix:
  - apicompat.AnthropicRequest: add Metadata json.RawMessage so metadata
    survives the OpenAI->Anthropic->Marshal round trip; without it the
    downstream rewrite has no user_id to work with.
  - service: extract applyClaudeCodeOAuthMimicryToBody, a ParsedRequest-free
    variant of the /v1/messages mimicry pipeline
    (rewriteSystemForNonClaudeCode + normalizeClaudeOAuthRequestBody +
    metadata.user_id injection) so the OpenAI-compat forwarders can reuse it.
  - service: add buildOAuthMetadataUserIDFromBody + hashBodyForSessionSeed
    for the same reason (no ParsedRequest at the call site).
  - ForwardAsChatCompletions / ForwardAsResponses: replace the 3-line
    prompt-prepend with the full mimicry pipeline.
  - applyClaudeCodeMimicHeaders: set x-client-request-id per-request
    (real Claude CLI always does); missing/duplicated values are one more
    third-party fingerprint signal.

No change to the native /v1/messages path: it already called the full
pipeline, we only lift those helpers into a reusable function.

Tests:
  - go build ./... passes
  - go test ./internal/service/... ./internal/pkg/apicompat/... passes
  - lsp_diagnostics clean on all touched files
  - pre-existing failures in internal/config are unrelated (env-sensitive
    tests that also fail on upstream main)

Align Claude Code mimicry constants with the latest real CLI traffic
(see Parrot's src/transform/cc_mimicry.py). Anthropic now uses the full
set of anthropic-beta tokens to decide whether a request counts as
"official Claude Code"; requests missing tokens that real CLI ships
today are demoted to third-party usage:

  Third-party apps now draw from your extra usage, not your plan limits.

Changes:
  - claude/constants.go: add new beta tokens (prompt-caching-scope,
    effort, redact-thinking, context-management, extended-cache-ttl) and
    expose FullClaudeCodeMimicryBetas() for the OAuth mimicry path.
  - claude/constants.go: bump default User-Agent to claude-cli/2.1.92.
  - identity_service.go: bump defaultFingerprint User-Agent accordingly.

No behavioral change for clients that already send a newer UA (fingerprint
merge still prefers the incoming value).

In reconstructResponseOutputFromSSE, text content Marshal/Unmarshal
failure previously caused an early return that silently discarded
already-extracted image_generation_call outputs. Now serialization
errors are tolerated so image results still reach the client.