Skip to content

feat(routing): make advisor narrowing visible, and optional - #482

Merged
MagicalTux merged 2 commits into
masterfrom
fix/advisor-eligibility-visibility
Sep 29, 2026
Merged

MagicalTux merged 2 commits into
masterfrom
fix/advisor-eligibility-visibility

Conversation

@MagicalTux

Copy link
Copy Markdown
Member

Fixes #479.

Claude Code declares the advisor tool on every request, so the advisor model's eligibility is paid by all of a client's traffic. When only some accounts could serve it, selection was confined to them and nothing said so: the only diagnostic fired at zero eligible accounts.

Two changes, neither altering the default routing:

  1. The narrowing is announced. Once per minute at most: Advisor model "claude-fable-5-1" narrows selection to 1 of 4 accounts — set advisorEligibility to "prefer" to route by request model instead. The counts are the accounts able to serve the request now and how many of those can serve the advisor as well; they are computed once in getActiveAccount, so the session-distribution and untagged-spread paths are covered too. GET /teamclaude/status carries the last reading as advisorNarrowing: { model, eligible, of, at } | null beside advisorEligibility.

  2. advisorEligibility: "strict" | "prefer" (top-level, default strict, applied live on reload). prefer routes a narrowed request by its main model only, as though it declared no advisor: every account that can serve the request is a candidate under the ordinary rotation, and whichever serves it runs the advisor call. Where every candidate can serve the advisor nothing changes, and zero-eligible still degrades exactly as before. Failover hops (pickAlternate) read the mode the same way. Any other value is strict with a one-time warning.

Docs in docs/configuration.md and docs/routing.md; config.example.json lists the key.

Tests: test/advisor-eligibility.test.js (16 tests: mode parsing, strict still pins, the log fires once per window and not when the sets are equal, both distribution paths, status readout, prefer spreads sessions and untagged requests, route pins, pickAlternate in both modes, zero-eligible degrade, live reload through a real server subprocess).

🤖 Generated with Claude Code

MagicalTux and others added 2 commits September 29, 2026 10:07
Claude Code declares the advisor tool on every request, invoked or not,
so the advisor model's eligibility is paid by all of a client's traffic.
When only some accounts could serve the advisor's model, strict
filtering (#98) confined the whole fleet's traffic to them and said
nothing: the one diagnostic fired at zero eligible accounts, not at one.

Selection now counts, once per request in getActiveAccount, how many of
the accounts that can serve the request can also serve its advisor, so
the session-distribution walks are covered as well as the plain one.
A strict subset is logged, throttled to once a minute like the degrade
line, and reported by getStatus() as `advisorNarrowing` beside the mode.

A new `advisorEligibility` config key chooses what happens then:
"strict" (the default, unchanged) keeps the filter; "prefer" routes a
narrowed request by its main model only, as though it declared no
advisor, so every account that can serve the request is a candidate and
whichever serves it runs the advisor call too. Unknown values are
"strict", said once. Applied live on reload like the neighbouring
routing knobs, and read the same way by pickAlternate so a failover hop
does not re-pin what the first attempt spread.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
@MagicalTux
MagicalTux merged commit 6b7a7aa into master Sep 29, 2026
5 checks passed
@MagicalTux MagicalTux mentioned this pull request Sep 29, 2026
MagicalTux added a commit that referenced this pull request Sep 29, 2026
Thirty-two commits since 1.1.21. Two change routing on an existing
config without an opt-in (#480, #481); the rest is opt-in, additive, or
display.

Behaviour changes
  #481 an API-key account's 401 is a cooldown, not a permanent `error`:
       1 min, then 5, 15 and 60 for every further rejection with no success
       in between; any 2xx/3xx resets it. The request still fails over and
       the client never sees the 401. OAuth accounts are unchanged
  #480 with session distribution on, requests carrying no session id stay
       within the top priority tier, so a fallback gateway no longer answers
       Claude Code's bootstrap and connector calls
  #470 a 200 whose SSE stream reports a provider failure before any output
       (`server_is_overloaded`, `response.failed`) fails over once, like a
       status-shaped failure would
  #465 a reload removes running accounts whose config entry is gone from
       disk, so `teamclaude remove` from another shell takes effect at once
  #460 `import` refuses an account whose token upstream has definitively
       rejected (401/403), even with `--name`; a 5xx or timeout still imports

Rename
  #483 the project is being renamed to TeamRouter (#72). This release accepts
       the new name everywhere the old one is read and changes nothing an
       install has on disk: `teamrouter` runs the same CLI, every
       `TEAMCLAUDE_*` variable is also read as `TEAMROUTER_*` (which wins when
       both are set), every `/teamclaude/…` control route also answers at
       `/teamrouter/…`, and `~/.config/teamrouter.json` is used when it exists

Features
  #441 per-account egress proxy (`accounts[].routing`: http, socks4/4a,
       socks5/5h) for refresh, probes and requests; `login --routing`,
       `teamclaude routing set/show/clear`, a connection check before it is
       relied on, and a short hold when the proxy is unreachable
  #427 `accounts[].allowExtraUsage: true` lets a paid extra-usage account
       serve once every account is past its threshold, instead of a 429
  #466 `accounts[].maxSpend`, a money cap judged against the month-to-date
       extra-usage spend upstream reports; the TUI shows what an account
       has billed
  #436 `autoRedeemResets` spends a free Codex rate-limit reset credit when
       the Codex pool runs dry (off by default)
  #482 `advisorEligibility: "strict" | "prefer"`; when the advisor model
       narrows selection to a subset of the fleet the log says so, and status
       carries the reading (`advisorNarrowing`)
  #478 `stripOverageHeaders` drops another org's per-organization billing
       headers from responses, for a pool spanning several orgs (#476)
  #471 `quota.unified5hSeenAt` / `unified7dSeenAt` in status: when upstream
       last stated each shared window
  #446 client and dimension usage for the last 5h and 24h in status and the
       dashboard, resumed across restarts
  #458 #459 #461 the dashboard sets the switch threshold, enables/disables and
       reprioritizes an account, and has a light theme remembered per browser
  #464 `l` in the TUI signs an account in `error` in again from the dashboard
  #457 status records which Codex limit meters each model
       (`quota.codexModelLimits`)
  #442 `quotaBarPercent` drops the percentage beside a TUI bar's countdown
  #451 `stripRequestFields` takes `content.<block type>` to drop content
       blocks a strict Anthropic-compatible upstream rejects
  #469 `proxy.mcp` schemas declare their item types, the write audit line
       records what happened, and the write queue has a depth (#447–#450)

Fixes
  #477 a refused WebSocket handshake whose headers all drop is relayed as a
       well-formed head instead of a blank line and body bytes
  #474 two members of one ChatGPT workspace are told apart by user id, so a
       second `login --codex` no longer replaces the first
  #469 a Codex Responses stream with no Content-Type is relayed as a stream
       and booked; thread repair on the global upstream; TUI settings and
       status gaps; a hint when a local login would have served
  #463 a Codex row with no session window draws one wide weekly bar

Tests
  #484 #485 #486 the suite asserts behaviour, not the scheduler: wall-clock
       upper bounds are gone, and subprocess tests spawn the server through
       `test-helpers/spawn-server.js`, which verifies the server it reached by
       `server.pid` (new in status) instead of trusting a port

Tooling
  #452 #453 #454 #455 docker workflow actions bumped
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.

Advisor eligibility silently pins the fleet when exactly one account can serve the advisor model

1 participant