@alexeiled/claude-router 1.10.3 → 1.11.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/.claude-plugin/plugin.json +1 -1
- package/CHANGELOG.md +35 -0
- package/README.md +4 -3
- package/docs/activity-routing.md +13 -4
- package/docs/architecture.md +13 -9
- package/docs/configuration.md +9 -6
- package/docs/evaluation.md +1 -1
- package/docs/native-router.md +2 -2
- package/docs/router-pane-classifier.svg +49 -43
- package/docs/user-guide.md +57 -37
- package/hooks/native-router.mjs +129 -0
- package/lib/activity.mjs +14 -12
- package/lib/config.mjs +17 -0
- package/lib/display.mjs +3 -3
- package/lib/facts.mjs +82 -0
- package/lib/policy.mjs +38 -14
- package/lib/route.mjs +93 -7
- package/package.json +2 -2
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
{
|
|
2
2
|
"name": "router",
|
|
3
3
|
"displayName": "Router",
|
|
4
|
-
"version": "1.
|
|
4
|
+
"version": "1.11.0",
|
|
5
5
|
"description": "Model and effort routing for the main conversation by task difficulty and activity, advised by a prompt classifier (Jev, Clef, Clef Flash, OpenAI or local Ollama). Subagents get one model at spawn: every agent that does not pin one.",
|
|
6
6
|
"author": {
|
|
7
7
|
"name": "Alexei Ledenev",
|
package/CHANGELOG.md
CHANGED
|
@@ -1,5 +1,40 @@
|
|
|
1
1
|
# Changelog
|
|
2
2
|
|
|
3
|
+
## 1.11.0 (2026-10-13)
|
|
4
|
+
|
|
5
|
+
Clef Omni joins Clef and Clef Flash as a third Cloudflare classifier. It costs less than Clef, and in Cloudflare's benchmarks it leads the Clef family on intent classification.
|
|
6
|
+
|
|
7
|
+
### Added
|
|
8
|
+
|
|
9
|
+
- Clef Omni (`clef-omni`), Cloudflare's 30B mixture-of-experts model (3B active) on Workers AI, as a built-in classifier. Select its row on the Classifier tab, or set `"classifier": "clef-omni"`. It takes the same `system-one` request and the same Cloudflare API token and account ID as Clef, with a 3,000 ms deadline. Cloudflare's list price is $0.15 per million input tokens, between Clef Flash ($0.038) and Clef ($0.24); output is free.
|
|
10
|
+
- The user guide compares the three Clef models on price and on Cloudflare's published benchmarks. No model wins every task.
|
|
11
|
+
|
|
12
|
+
Router sends Clef Omni text only, not images, audio or video. Clef Omni has no activity probe yet, so its Classifier tab row shows no accuracy or p95 latency. Router has not measured its routing quality, and its probabilities have not been compared with Jev's.
|
|
13
|
+
|
|
14
|
+
## 1.10.5 (2026-10-11)
|
|
15
|
+
|
|
16
|
+
The route now follows the task while it runs, for the tier and the activity, in both directions. Before, a turn was
|
|
17
|
+
routed once, from its prompt, and a continuation prompt could only keep the route. A team worker whose `/goal` brief
|
|
18
|
+
read as `ops` ran on Haiku · high and kept it for 100+ requests of coding. See
|
|
19
|
+
[Changes during a task](docs/user-guide.md#changes-during-a-task).
|
|
20
|
+
|
|
21
|
+
### Added
|
|
22
|
+
|
|
23
|
+
- A turn whose own tool calls edit a code file moves at once to the tier's `code` route (`docs` for doc-only edits), shown as `↗ coding` with a toast. The tools are the evidence, so no classifier support is needed and the cache cost does not hold the move back; only the cash cap on a credits-billed model can, priced at the route the context fits. It only goes up and skips pinned turns and `activityRouting: off`.
|
|
24
|
+
- Classifier reviews inside a turn: every `policy.reviewEveryRequests` requests (default 4; `0` turns them off) the classifier is asked again in the background, with the turn's prompt and recent dialogue. A step never waits; the answer applies at the next step and can move the tier or the activity either way. Each review sends the classifier the same kind of text a new turn does.
|
|
25
|
+
- `policy.minDwellRequests` (default 8) and a checkpoint for cheaper routes on the running task: a downgrade or a cheaper activity waits for that many requests after the last model change and for the last test, build, lint or type-check command since then to pass. Until then the route stays: a tier shows `… down after a passed check` (reason `downgrade-held`), a held activity keeps the running one. A check is a test runner, type checker or linter, or a launcher running one (`npm test`, `go test`, `make lint`), alone or after `&&`; a piped check (`npm test | tail`), `;`, `|| true`, package names and paths do not count. A cheaper activity also needs `policy.downgradeMass` probability.
|
|
26
|
+
|
|
27
|
+
### Changed
|
|
28
|
+
|
|
29
|
+
- A downgrade lands on the cheapest tier the answer supports as a whole, not only on the chosen tier. A split answer (micro 0.63, low 0.3) on `high` now moves to `low` instead of staying until `micro` alone reaches 0.9, and a cold candidate can give way to a warm tier between it and the incumbent.
|
|
30
|
+
- A prompt the classifier scores as a continuation of the task no longer keeps the route by itself. It is decided like a review: up on the usual votes, down past the checkpoint. The `continuation` reason is gone. `policy.continuationMass` now only decides whether a prompt counts as the running task.
|
|
31
|
+
|
|
32
|
+
### Fixed
|
|
33
|
+
|
|
34
|
+
- Replies after an in-turn move are labelled with the activity they ran on, not the turn's first label; the classifier's answer stays as recorded for the stats.
|
|
35
|
+
|
|
36
|
+
The review interval, the dwell and the checkpoint are starting values, not tuned on data.
|
|
37
|
+
|
|
3
38
|
## 1.10.3 (2026-10-11)
|
|
4
39
|
|
|
5
40
|
### Fixed
|
package/README.md
CHANGED
|
@@ -3,19 +3,20 @@
|
|
|
3
3
|
[](https://github.com/alexei-led/claude-router/actions/workflows/ci.yml)
|
|
4
4
|
[](https://opensource.org/licenses/MIT)
|
|
5
5
|
|
|
6
|
-
**A Claude Code Mod that chooses a model and effort for each main-conversation turn, advised by a prompt classifier: Jev
|
|
6
|
+
**A Claude Code Mod that chooses a model and effort for each main-conversation turn, advised by a prompt classifier: Jev; Cloudflare's Clef, Clef Flash or Clef Omni; OpenAI; or a local Ollama model.**
|
|
7
7
|
|
|
8
8
|
The Mod changes only the model and effort in Claude Code's turn hook. Claude Code sends the request, streams the reply, runs tools, and reports usage. The router does not proxy Anthropic traffic or start a local server; an Ollama classifier calls an Ollama server that you run. Subagents get one model each when they start, chosen the same way: every agent whose definition names no model, says `model: inherit`, or sets `modelRouting: auto`. Their steps are never re-routed. See [Subagent routing](docs/configuration.md#subagent-routing).
|
|
9
9
|
|
|
10
10
|
## Supported classifiers
|
|
11
11
|
|
|
12
|
-
The active classifier is asked
|
|
12
|
+
The active classifier is asked at the start of each logical turn, for the tier and for the turn's activity, and again in the background every 4 requests while the turn runs (`policy.reviewEveryRequests`). Pick one on the pane's **Classifier** tab or in `router.json`; only the active one receives prompt text. With subagent routing `on` (the default), it is also asked once per routed subagent, with the task's description and prompt; set `"subagentRouting": "off"` to keep that text local.
|
|
13
13
|
|
|
14
14
|
| Classifier | Service | Credential | Prompt text goes to | How it is asked for the activity |
|
|
15
15
|
| ---------- | ------------------------------------------ | ----------------------------------- | ------------------------- | --------------------------------------------------------------------- |
|
|
16
16
|
| Jev | [typesafe.ai](https://typesafe.ai) | Jev API key | typesafe.ai | A third `choice` question in the same request |
|
|
17
17
|
| Clef | Cloudflare Workers AI, 27B | Cloudflare API token and account ID | Cloudflare | The same |
|
|
18
18
|
| Clef Flash | Cloudflare Workers AI, 9B | Cloudflare API token and account ID | Cloudflare | The same |
|
|
19
|
+
| Clef Omni | Cloudflare Workers AI, 30B MoE | Cloudflare API token and account ID | Cloudflare | The same |
|
|
19
20
|
| OpenAI | OpenAI, `gpt-6-luna` (public beta) | OpenAI API key | OpenAI (`api.openai.com`) | A second `choice` in `questions`, in the same request |
|
|
20
21
|
| Ollama | A local [Ollama](https://ollama.com) model | None | Stays on this machine | A second request: letters A to H, one token, if 300 ms or more remain |
|
|
21
22
|
|
|
@@ -54,7 +55,7 @@ sequenceDiagram
|
|
|
54
55
|
Mod-->>Dev: Status band and router pane
|
|
55
56
|
```
|
|
56
57
|
|
|
57
|
-
The active classifier labels
|
|
58
|
+
The active classifier labels each new logical turn, and the router checks the route again while the task runs: a turn labelled `ops` that edits code moves up to the `code` route at once, and a background classifier review every 4 requests can move the tier or the activity either way. A cheaper route on the same task waits for a dwell and a passed test, build or lint run. See [Changes during a task](docs/user-guide.md#changes-during-a-task). A local policy applies vote, context, failure, and cache-cost rules before the Mod changes the next step. Claude Code owns request construction, model credentials, tools, streaming, and the API cost ledger.
|
|
58
59
|
|
|
59
60
|
| Tier | Default model | Effort |
|
|
60
61
|
| -------- | ------------- | -------- |
|
package/docs/activity-routing.md
CHANGED
|
@@ -32,6 +32,13 @@ The router compares routes, not labels. If the new activity resolves to the rout
|
|
|
32
32
|
A move to a cheaper route must pay back its cache write first. A move to a stronger route needs the classifier's
|
|
33
33
|
support. The [user guide](user-guide.md#route-by-activity) has the full rules.
|
|
34
34
|
|
|
35
|
+
The label comes from the prompt, so it can be wrong. A prompt such as "run the worker brief" reads as `ops` and the
|
|
36
|
+
turn may then spend 80 requests writing code. The router therefore also watches the turn's own tool calls. Once the
|
|
37
|
+
turn edits a file, it moves to the `code` route of its tier (or `docs` for doc files) if that route is stronger. This
|
|
38
|
+
move only goes up. The cache cost does not hold it back; only the cash cap on a credits-billed model can. A later
|
|
39
|
+
classifier review in the same turn can move it down again, past the checks in
|
|
40
|
+
[Changes during a task](user-guide.md#changes-during-a-task). The next turn starts from the route the turn ended on.
|
|
41
|
+
|
|
35
42
|
## The default matrix
|
|
36
43
|
|
|
37
44
|
A `·` cell uses the base route.
|
|
@@ -51,7 +58,7 @@ Why each row:
|
|
|
51
58
|
scores well above Haiku on coding tests. Sonnet at `medium` does not, so it is not used.
|
|
52
59
|
- **`ops` at `medium`: Haiku · high.** Command runs last many requests, so Haiku pays back its cache write. Haiku is
|
|
53
60
|
weaker on multi-step tool work, so `ops` at `high` stays on Opus. Repeated tool errors move an `ops` turn to the
|
|
54
|
-
`high` route, Opus · xhigh.
|
|
61
|
+
`high` route, Opus · xhigh. An `ops` turn that starts editing code moves to Opus · medium.
|
|
55
62
|
- **`explore` and `docs`.** Moving them saved nothing in the replay. Haiku also scores lower on research
|
|
56
63
|
benchmarks: 50.1% on Humanity's Last Exam with tools against 63.0% for Opus at `medium`. They keep the base routes.
|
|
57
64
|
|
|
@@ -163,7 +170,9 @@ model** in the pane, which writes that for you. The cell keeps that route if you
|
|
|
163
170
|
again then:
|
|
164
171
|
|
|
165
172
|
```json
|
|
166
|
-
{
|
|
173
|
+
{
|
|
174
|
+
"activities": { "ops": { "medium": { "model": "opus", "effort": "medium" } } }
|
|
175
|
+
}
|
|
167
176
|
```
|
|
168
177
|
|
|
169
178
|
Two other directions, with what they trade:
|
|
@@ -187,8 +196,8 @@ Two other directions, with what they trade:
|
|
|
187
196
|
```
|
|
188
197
|
|
|
189
198
|
**Activity threshold.** An activity applies only when the classifier gives it at least this probability. Below it, a
|
|
190
|
-
new task goes back to the base route if that move passes the usual checks
|
|
191
|
-
|
|
199
|
+
new task goes back to the base route if that move passes the usual checks; the running task follows
|
|
200
|
+
[Changes during a task](user-guide.md#changes-during-a-task). The default is 60% (`policy.activityMass: 0.6`). Change it on the Policy tab under **Activities**,
|
|
192
201
|
or in `router.json`. There is no data to tune it on yet: the probe results record whether the
|
|
193
202
|
label was right, not the classifier's probabilities.
|
|
194
203
|
|
package/docs/architecture.md
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
# Architecture
|
|
2
2
|
|
|
3
|
-
Claude Code owns the model request. The router Mod asks the active classifier (Jev by default; Cloudflare's Clef or Clef
|
|
3
|
+
Claude Code owns the model request. The router Mod asks the active classifier (Jev by default; Cloudflare's Clef, Clef Flash or Clef Omni; OpenAI; or a local Ollama model) for tier advice and, unless `activityRouting` is `off`, the turn's activity, applies local routing policy, and changes only the next main-conversation step's model and effort. Subagent steps pass through unchanged; `subagentRouting` may choose a subagent's model once, at `agent.spawn` (see [Subagents](#subagents)).
|
|
4
4
|
|
|
5
5
|
## Request flow
|
|
6
6
|
|
|
@@ -30,9 +30,9 @@ sequenceDiagram
|
|
|
30
30
|
|
|
31
31
|
The Mod does not construct an Anthropic request or proxy its response. Claude Code owns authentication, request fields, streaming, tools, model fallback, and its native usage ledger. No local gateway process or port is part of the runtime. The Ollama classifier calls an Ollama server that you run; the Mod starts none.
|
|
32
32
|
|
|
33
|
-
The active classifier receives only the current prompt and up to six recent user or assistant text messages. Each text is capped at 1,200 characters. System messages, tool inputs, and tool results are not included. Only the active classifier is asked: Jev sends that text to typesafe.ai, Clef and Clef
|
|
33
|
+
The active classifier receives only the current prompt and up to six recent user or assistant text messages. Each text is capped at 1,200 characters. System messages, tool inputs, and tool results are not included. Only the active classifier is asked: Jev sends that text to typesafe.ai, Clef, Clef Flash and Clef Omni to Cloudflare Workers AI, OpenAI to api.openai.com. Ollama runs on this machine, so the text stays here. The Jev, Cloudflare, and OpenAI keys are sensitive plugin options; the Cloudflare account ID is a plain one that fills the endpoint. Ollama takes no key. The Mod does not receive Claude's Anthropic credentials.
|
|
34
34
|
|
|
35
|
-
Each classifier names its wire protocol, `api`, and `classifier-apis.mjs` holds one adapter per protocol. `system-one` (Jev, Clef, Clef Flash) sends the typed `state` and `questions` request and reads `answers.route`; Cloudflare's REST API wraps that answer in `{ result, success, errors }`, which the parser unwraps. `openai-decisions` (OpenAI) sends the state as `input` text and reads a `choice` answer plus a `predicate` for the continuation. `ollama` asks for one letter with thinking off and reads the log-probability of each letter from the top 20 candidates. A letter's probability is its share of the letters found there; a letter missing from the list counts as zero. The Ollama adapter asks no continuation question, so `continuation` is null there. Every adapter returns the same advice shape, so the routing policy is unchanged.
|
|
35
|
+
Each classifier names its wire protocol, `api`, and `classifier-apis.mjs` holds one adapter per protocol. `system-one` (Jev, Clef, Clef Flash, Clef Omni) sends the typed `state` and `questions` request and reads `answers.route`; Cloudflare's REST API wraps that answer in `{ result, success, errors }`, which the parser unwraps. `openai-decisions` (OpenAI) sends the state as `input` text and reads a `choice` answer plus a `predicate` for the continuation. `ollama` asks for one letter with thinking off and reads the log-probability of each letter from the top 20 candidates. A letter's probability is its share of the letters found there; a letter missing from the list counts as zero. The Ollama adapter asks no continuation question, so `continuation` is null there. Every adapter returns the same advice shape, so the routing policy is unchanged.
|
|
36
36
|
|
|
37
37
|
The advice also carries `activity`: `{ choice, probabilities }` over the seven activities and `uncertain`, or null. Each protocol asks for it differently:
|
|
38
38
|
|
|
@@ -72,11 +72,11 @@ Node.js is not part of routing. It is used for development, the optional configu
|
|
|
72
72
|
|
|
73
73
|
## Decision and context
|
|
74
74
|
|
|
75
|
-
|
|
75
|
+
A logical turn is classified at its first step, and again every `reviewEveryRequests` requests in the background (see Review inside a turn below). Other tool continuations reuse the decision and add no votes. Before each step, the controller checks whether the selected model still fits after new tool output.
|
|
76
76
|
|
|
77
77
|
Repeated tool-error escalation needs matching non-empty error output and an intervening edit attempt within a 40-message lookback. A bounded fingerprint covers the full output, not only the runner header. It removes ANSI colors, normalizes line endings, and normalizes source positions in stack frames and TAP `location` fields. TAP durations and Jest `✕ ... (N ms)` timings are also normalized. Other test names, case, whitespace, and numeric error values are kept. Unrecognized runner noise remains part of the signature: changed output can miss a repeat rather than trigger a false escalation.
|
|
78
78
|
|
|
79
|
-
An upgrade or downgrade needs a streak of supporting votes (`upgradeVotes`, `downgradeVotes`), because staying keeps a warm cache. Before the first measured reply of a history (session start, clear, rewind, committed compaction), no cache from it exists yet, so one vote is enough. The mass bars, the cash gate, escalation
|
|
79
|
+
An upgrade or downgrade needs a streak of supporting votes (`upgradeVotes`, `downgradeVotes`), because staying keeps a warm cache. A downgrade lands on the cheapest tier from the choice up to just below the incumbent whose mass at or below it, with its streak, clears that tier's bar: a split answer (micro 0.63, low 0.3) on `high` moves to `low` rather than waiting for `micro` alone, and a cold candidate can give way to a warm one in between, such as the same model at a lower effort. Before the first measured reply of a history (session start, clear, rewind, committed compaction), no cache from it exists yet, so one vote is enough. The mass bars, the cash gate, escalation and holds apply as usual. On the same task (`sameTaskOf`: a review inside a turn, or a prompt with advice `continuation` at or above `continuationMass`) an upgrade proceeds as usual, but a downgrade that has its votes stays as `downgrade-held` until `facts.downAllowed`: at least `minDwellRequests` requests since the last model change (`loop.switchedAt`), and the last verification command since then (`verifiedSince`: a Bash test, build, lint or type-check run) passed.
|
|
80
80
|
|
|
81
81
|
A move to a smaller context window needs a main-response measurement from the current history and Claude's local context estimate. The router uses the larger available reading and reserves 20% of the model window. If it cannot establish fit, it keeps the current model with `context-unknown`. It does not infer token count from character count.
|
|
82
82
|
|
|
@@ -84,16 +84,20 @@ An explicit context-window failure marks that model ineligible until the next hi
|
|
|
84
84
|
|
|
85
85
|
## Activity routing
|
|
86
86
|
|
|
87
|
-
The router decides on a
|
|
87
|
+
The router decides on a _cell_: a tier and an activity. The route is `{ ...routes[tier], ...activities[activity][tier] }`; the (model, effort) pair, not the label, is what Claude Code receives and what owns a cache. Two cells that resolve to the same route never switch.
|
|
88
88
|
|
|
89
|
-
`chooseRoute`
|
|
89
|
+
`chooseRoute` decides at a turn's first step, and again on each review answered inside the turn:
|
|
90
90
|
|
|
91
91
|
1. **Pin.** A pinned tier wins, with no activity: the pin's tier route.
|
|
92
|
-
2. **Activity.** Only in `on`; `off` uses none, and ignores the activity a turn in `on` ran with before the mode changed. With no advice, or advice without an activity answer, the incumbent's activity stays. On
|
|
92
|
+
2. **Activity.** Only in `on`; `off` uses none, and ignores the activity a turn in `on` ran with before the mode changed. With no advice, or advice without an activity answer, the incumbent's activity stays. On the same task (`sameTaskOf`; Ollama gives no continuation answer, so only reviews count for it) a label whose route at the incumbent tier is stronger applies at once, and a weaker one only with at least `downgradeMass` probability once `facts.downAllowed` holds; an unclear label keeps the running activity. Otherwise the activity with probability at least `activityMass` applies; below that, or for `uncertain`, there is none and the tier's base route is used.
|
|
93
93
|
3. **Tier.** The tier policy runs as before, with every candidate tier priced at its route for the activity and the incumbent at the route it runs. Repeated tool errors move to the lowest tier above whose route is stronger than the running one, or sends a request that cannot be ordered against it. With no activity that is one tier up, as in 1.5. If no tier above helps, nothing escalates and the failure signature is kept.
|
|
94
94
|
4. **Lateral move.** If the tier stays and the route would change, `lateralMove` decides. Routes order by configured output price, then effort; a session effort counts as the effort Claude Code sends. Two routes that send the same request do not switch. A pair that cannot be ordered (an unpriced model, a numeric effort, two models at one price) counts as a move up, so the base route is always reachable. A stronger route needs the activity's probability to clear the upgrade bar for its switching tax (a base route with no activity counts as probability 1). A cheaper route needs its downgrade tax to be paid back within `downgradeHorizonTurns`, except before the first measured reply of a history, when there is no cache to protect. The cash gate applies to both. The result is `activity-up`, `activity-down`, `activity-pending` (stay on the running route and its activity), `cash-gate`, or `hold`. While an escalation hold lasts, only a stronger route may be taken. One vote is enough: an activity is a fact about the turn, not a noisy estimate. If the tier changes, the tier gates have already priced the new route, and no lateral rule applies.
|
|
95
95
|
5. **Fit.** Context fit and model availability resolve through `route(tier, activity)`. A fallback tries the last cell, the native model's cell, then each tier with the activity. The last and the native cell count only when they carry no activity or the turn's; otherwise, or when nothing fits, the native model runs at the session effort.
|
|
96
96
|
|
|
97
|
+
**Observed move.** On each later step of the turn, `observedMove` reads the turn's tool calls since its prompt, or since the transcript snapshot of the last review applied inside the turn (`loop.observedFrom`), since the review already judged the earlier ones, (`observedActivity`: `code` after an edit to a non-doc file, `docs` after edits to doc files only, else `ops`, `read` or `talk`). When that activity's route at the decision's tier is stronger than the running one, the remaining steps take it, with reason `activity-up` and `lastActivity` set, so the next turn starts there. The tier is fitted to the context first, so the gates price the route that will run. It goes through `lateralMove` with probability 1: the tools are the evidence, so the switching tax never holds it back (the upgrade bar tops out near 0.9), and only the cash gate can. The turn's reply label (`turnActivities.label`) follows the move; the classifier's answer stays as recorded. It never moves down, and skips pinned turns and `off`. `canMoveUp` skips the transcript read when no activity's route at the tier is stronger, so a step on the strongest route costs nothing. A move to a smaller window waits for a measured context, as in `routeTurn`. Context fit and model availability then apply as for any continuation.
|
|
98
|
+
|
|
99
|
+
**Review inside a turn.** Every `reviewEveryRequests` requests after the last classifier answer (`loop.reviewedAt`, counted in `loop.requests`), `reviewMidTurn` reads the transcript and asks the active classifier in the background, with the turn's prompt and recent dialogue and `facts.midTurn` set. The step does not wait. The first step after the answer arrives runs `chooseRoute` on it, before the observed move, with only the advice taken from the review: the checkpoint, context and cache facts are read at that step. Tier and activity move by the same rules as a continuation prompt: up on votes, down only past `downAllowed`. A pinned turn, a failed read or ask, an answer without advice, and an answer about a replaced history (another `generation`) change nothing; turn completion and session end abort a review in flight. `observeResponse` counts each request and records the request and message count at which the served (model, effort) last changed (`switchedAt`), which the dwell and the verification window start from. A failed transcript read keeps the running route. A verification is a Bash command whose status is the check's: a checker (`pytest`, `tsc`, `eslint`, …) or a launcher whose subcommand names a check (`npm test`, `npm run lint`, `go test`, `make lint`, `uv run pytest`), alone or in an `&&` list. A pipeline, `;`, `||` or background job does not count, since its status can hide a failed check; neither do a `--version` or `--help` run, an option before the subcommand (`npm --prefix test install`), package names, paths and file names, nor a command Claude Code sent to the background, by request or after a timeout. The first response a loop serves also sets `switchedAt`, so checks from before the router chose a route do not count. When compaction or a rewind shortens the history, `prepareLoop` moves `switchedAt.messageCount` and `observedFrom` back to its end and keeps the request-based dwell.
|
|
100
|
+
|
|
97
101
|
At session start the incumbent is the session's model. `cellForModel` looks for it in the base routes, baseline first, and only with `activityRouting: on` in the activity overrides too, so a session that starts on an override route keeps its cell until the classifier says otherwise. The start mode uses the same lookup: with `on`, a start model that only an override uses (Sonnet 5.5 on the defaults) starts with routing on.
|
|
98
102
|
|
|
99
103
|
## State and controls
|
|
@@ -131,7 +135,7 @@ Every spawn in `on` gets a record keyed by `agentId`, from `next(e)`'s answer. I
|
|
|
131
135
|
|
|
132
136
|
## Classifier deadline and failure handling
|
|
133
137
|
|
|
134
|
-
Each advice attempt has a total deadline, including any transient retry: 1,500 ms for Jev, 3,000 ms for Clef, Clef Flash and OpenAI, and 5,000 ms for Ollama by default. Ollama's first call after its model unloads can exceed that while the model loads. A timeout does not retry. `Retry-After` is honored only when it fits the remaining budget. Three launched failures pause the classifier for 60 seconds. Missing-key, missing-account, busy, and paused skips do not count as failures. The Ollama activity request is a second call inside the same deadline and the same in-flight slot. It is sent only when the route answer parsed and at least 300 ms remain, gets one attempt and no retry, and a timeout or error there returns the route advice with `activity: null`. It counts as no failure and never opens a pause. Each classifier has its own failure count, pause, and in-flight slot. A turn that started under one classifier finishes with it, even if the pane switches meanwhile. Switching the classifier, or starting a session with another one, starts the new one from a clean count.
|
|
138
|
+
Each advice attempt has a total deadline, including any transient retry: 1,500 ms for Jev, 3,000 ms for Clef, Clef Flash, Clef Omni and OpenAI, and 5,000 ms for Ollama by default. Ollama's first call after its model unloads can exceed that while the model loads. A timeout does not retry. `Retry-After` is honored only when it fits the remaining budget. Three launched failures pause the classifier for 60 seconds. Missing-key, missing-account, busy, and paused skips do not count as failures. The Ollama activity request is a second call inside the same deadline and the same in-flight slot. It is sent only when the route answer parsed and at least 300 ms remain, gets one attempt and no retry, and a timeout or error there returns the route advice with `activity: null`. It counts as no failure and never opens a pause. Each classifier has its own failure count, pause, and in-flight slot. A turn that started under one classifier finishes with it, even if the pane switches meanwhile. Switching the classifier, or starting a session with another one, starts the new one from a clean count.
|
|
135
139
|
|
|
136
140
|
The pinned host API does not expose a per-request abort option for `$.http.fetch`. The Mod returns to the incumbent at its own deadline and blocks another advice request while that wire call remains pending. Claude Code's observed host timeout is about 30 seconds. Recorded reload and unload checks closed pending sockets after 24.7 and 28.5 seconds. Replacement closed one after 31.6 seconds. These runs do not show immediate per-call cancellation. A plugin network-policy refusal degrades routing and is never bypassed with a subprocess.
|
|
137
141
|
|
package/docs/configuration.md
CHANGED
|
@@ -7,7 +7,7 @@ The router works with its built-in defaults. Configure the key of the active cla
|
|
|
7
7
|
| Item | Where it lives | How to change it |
|
|
8
8
|
| ---------------------------------- | ----------------------------------------------- | ----------------------------------------------------------------------------------------------- |
|
|
9
9
|
| Jev API key | Sensitive plugin option `typesafe_api_key` | Run `/plugin configure router`, or use **Set up** or **Credentials → Edit** in the router pane. |
|
|
10
|
-
| Cloudflare API token | Sensitive plugin option `cloudflare_api_token` | The same dialog. Used by Clef and Clef
|
|
10
|
+
| Cloudflare API token | Sensitive plugin option `cloudflare_api_token` | The same dialog. Used by Clef, Clef Flash and Clef Omni. |
|
|
11
11
|
| Cloudflare account ID | Plugin option `cloudflare_account_id` | The same dialog, or the config menu row. Not a secret. |
|
|
12
12
|
| OpenAI API key | Sensitive plugin option `openai_api_key` | The same dialog. Used only by the OpenAI classifier. Ollama needs no key. |
|
|
13
13
|
| Router settings and classifier | `router.json` in the active Claude Code profile | Edit the file directly, or use the pane's Routes, Policy and Classifier tabs. |
|
|
@@ -70,7 +70,7 @@ The classifier also names the activity of a turn: what the turn produces. `activ
|
|
|
70
70
|
|
|
71
71
|
A `router.json` that still sets `"shadow"`, a mode removed in 1.10.1, is read as not setting it: the default, `on`, runs, and the next save from the pane drops the value.
|
|
72
72
|
|
|
73
|
-
The seven activities are `code`, `debug`, `explore`, `plan`, `review`, `ops`, and `docs`. The classifier may also answer `uncertain`. In `on`, an activity applies only when its probability is at least `policy.activityMass`. Below that, or on `uncertain`, a new task goes back to its base route if the move passes the cache checks
|
|
73
|
+
The seven activities are `code`, `debug`, `explore`, `plan`, `review`, `ops`, and `docs`. The classifier may also answer `uncertain`. In `on`, an activity applies only when its probability is at least `policy.activityMass`. Below that, or on `uncertain`, a new task goes back to its base route if the move passes the cache checks. On the running task the activity can move either way during the turn; see [Changes during a task](user-guide.md#changes-during-a-task). When the classifier gives no usable activity answer, the running activity stays and the tier answer still applies. When it gives no answer at all, the running route stays. In `off`, the base route runs.
|
|
74
74
|
|
|
75
75
|
`activities` maps an activity to a tier to a route override. An override may set `model`, `effort`, or both. A field it leaves out comes from the built-in override for that cell, or from the tier's own route, the **base route**, where there is none:
|
|
76
76
|
|
|
@@ -125,13 +125,14 @@ Switch the mode with `/router activities off|on`, or with the **Activity routing
|
|
|
125
125
|
|
|
126
126
|
## Classifiers
|
|
127
127
|
|
|
128
|
-
`classifier` names the active entry of `classifiers`. Each entry names its wire protocol in `api`: `system-one` for Jev, Clef and Clef
|
|
128
|
+
`classifier` names the active entry of `classifiers`. Each entry names its wire protocol in `api`: `system-one` for Jev, Clef, Clef Flash and Clef Omni, `openai-decisions` for OpenAI, and `ollama` for a local Ollama server. Every protocol is converted to the same advice, so the routing policy does not change with the classifier. Unless `activityRouting` is `off`, each one is also asked for the turn's [activity](#activity-routing); the answer is optional, and a missing or malformed one keeps the activity already running. Cloudflare wraps the System One answer in `{ result, success, errors }`; Router reads both forms.
|
|
129
129
|
|
|
130
130
|
| `classifier` | Endpoint | Request `model` | Key option | `timeoutMs` |
|
|
131
131
|
| ------------ | -------------------------------------------------------------------------------------------------------- | --------------- | ---------------------- | ----------: |
|
|
132
132
|
| `jev` | `https://api.typesafe.ai/v1/systemone` | `jev-1.13.0` | `typesafe_api_key` | `1500` |
|
|
133
133
|
| `clef` | `https://api.cloudflare.com/client/v4/accounts/{cloudflare_account_id}/ai/run/@cf/cloudflare/clef` | `clef` | `cloudflare_api_token` | `3000` |
|
|
134
134
|
| `clef-flash` | `https://api.cloudflare.com/client/v4/accounts/{cloudflare_account_id}/ai/run/@cf/cloudflare/clef-flash` | `clef-flash` | `cloudflare_api_token` | `3000` |
|
|
135
|
+
| `clef-omni` | `https://api.cloudflare.com/client/v4/accounts/{cloudflare_account_id}/ai/run/@cf/cloudflare/clef-omni` | `clef-omni` | `cloudflare_api_token` | `3000` |
|
|
135
136
|
| `openai` | `https://api.openai.com/v1/decisions` | `gpt-6-luna` | `openai_api_key` | `3000` |
|
|
136
137
|
| `ollama` | `http://127.0.0.1:11434/api/chat` | `qwen3.5:9b` | none | `5000` |
|
|
137
138
|
|
|
@@ -267,14 +268,16 @@ Policy defaults:
|
|
|
267
268
|
| `upgradePivotUsd` | `0.5` | Cost scale used by the upgrade bar. Must be positive. |
|
|
268
269
|
| `jumpConfidence` | `0.95` | Support for a two-tier jump without waiting for votes. |
|
|
269
270
|
| `downgradeVotes` | `2` | Consecutive supporting votes before a downgrade; one before a history's first measured reply. |
|
|
270
|
-
| `downgradeMass` | `0.9` | Minimum probability mass required for a downgrade.
|
|
271
|
+
| `downgradeMass` | `0.9` | Minimum probability mass at or below the target tier required for a downgrade. |
|
|
271
272
|
| `downgradeSlope` | `0.08` | Maximum increase to the downgrade bar from estimated switching cost. |
|
|
272
273
|
| `downgradePivotUsd` | `0.5` | Cost scale used by the downgrade bar. Must be positive. |
|
|
273
274
|
| `downgradeHorizonTurns` | `5` | Later turns included in a downgrade payback estimate. |
|
|
274
|
-
| `continuationMass` | `0.7` | Advice probability
|
|
275
|
-
| `escalationHoldTurns` | `2` |
|
|
275
|
+
| `continuationMass` | `0.7` | Advice probability at which a new prompt counts as the running task, not a new one. |
|
|
276
|
+
| `escalationHoldTurns` | `2` | Decisions held after a repeated tool error escalates the route; each turn and in-turn review counts. |
|
|
276
277
|
| `cashCapUsd` | `2` | Maximum estimated cold cache write for a `credits` model. It does not cap output or session spend. |
|
|
277
278
|
| `activityMass` | `0.6` | Minimum probability for the classifier's activity to apply. Not tuned on any classifier. |
|
|
279
|
+
| `reviewEveryRequests` | `4` | Requests between classifier reviews inside a turn; `0` turns them off. Not tuned. |
|
|
280
|
+
| `minDwellRequests` | `8` | Requests after a model change before a cheaper route on the same task. Not tuned. |
|
|
278
281
|
|
|
279
282
|
The pane writes a subset of these settings. The **Routes** and **Policy** tabs set `routes`, `activities`, `activityRouting`, `baselineTier`, `policy.downgradeVotes`, `policy.downgradeHorizonTurns`, `policy.cashCapUsd`, and `policy.activityMass` in one save: it writes only values that differ from the defaults and removes the rest. `/router activities <mode>` writes `activityRouting` at once, with **Undo**. The **Classifier** tab sets `classifier` and the active classifier's `timeoutMs` at once. **Undo** reverts the settings the last pane write changed. Each save validates the whole file, keeps other keys, and applies from the next turn. A failed check names the setting and leaves the file unchanged. The pane refuses to write through a symlink. Model aliases and the other settings require a direct edit.
|
|
280
283
|
|
package/docs/evaluation.md
CHANGED
|
@@ -106,7 +106,7 @@ Five probe results are checked in, run on 2026-10-09 and 2026-10-10 (UTC) with t
|
|
|
106
106
|
| Clef Flash | 69 of 70 | 470 ms | 874 ms | [`activity-probe-clef-flash.json`](../experiments/mod-router/results/activity-probe-clef-flash.json) |
|
|
107
107
|
| Ollama | 66 of 70 | 2986 ms | 3483 ms | [`activity-probe-ollama.json`](../experiments/mod-router/results/activity-probe-ollama.json) |
|
|
108
108
|
|
|
109
|
-
Ollama's latency covers both requests and stays inside its 5000 ms deadline; on slower hardware the activity step is skipped when less than 300 ms remain. They measure the label on synthetic prompts, not routing quality or savings. The activity question costs little time: 30 alternating pairs of the same prompts, with and without it, gave Jev p50 259 ms both ways and p95 300 ms without, 306 ms with; OpenAI p50 163 ms and 162 ms, p95 296 ms and 349 ms. OpenAI's first request of a run took about 2.3 s of its 3000 ms deadline. The Classifier tab shows each result under its classifier, such as `activity probe 70/70 · p95 299 ms · Oct 10` (UTC date), when the classifier runs the probed model. The script regenerates `lib/probe-results.mjs` from the result files after each run; `npm run probe:results` does it alone, and a test fails when the two disagree. A run without a single activity answer (an activity or `uncertain`) or without a p95 latency, such as one with a revoked key or a model whose answers all lack the activity, exits with an error and writes neither file. `policy.activityMass` was not tuned on any classifier; run the probe before you trust it with a new one.
|
|
109
|
+
Ollama's latency covers both requests and stays inside its 5000 ms deadline; on slower hardware the activity step is skipped when less than 300 ms remain. They measure the label on synthetic prompts, not routing quality or savings. The activity question costs little time: 30 alternating pairs of the same prompts, with and without it, gave Jev p50 259 ms both ways and p95 300 ms without, 306 ms with; OpenAI p50 163 ms and 162 ms, p95 296 ms and 349 ms. OpenAI's first request of a run took about 2.3 s of its 3000 ms deadline. The Classifier tab shows each result under its classifier, such as `activity probe 70/70 · p95 299 ms · Oct 10` (UTC date), when the classifier runs the probed model. The script regenerates `lib/probe-results.mjs` from the result files after each run; `npm run probe:results` does it alone, and a test fails when the two disagree. A run without a single activity answer (an activity or `uncertain`) or without a p95 latency, such as one with a revoked key or a model whose answers all lack the activity, exits with an error and writes neither file. `policy.activityMass` was not tuned on any classifier; run the probe before you trust it with a new one. Clef Omni has no probe result yet, so its Classifier tab row shows none.
|
|
110
110
|
|
|
111
111
|
## Routing vs your model
|
|
112
112
|
|
package/docs/native-router.md
CHANGED
|
@@ -4,9 +4,9 @@ This guide describes the Mod shipped in the plugin. Claude Code 2.1.289 is the m
|
|
|
4
4
|
|
|
5
5
|
## Turn behavior
|
|
6
6
|
|
|
7
|
-
At the first main step of a logical turn, the Mod reads recent message text and asks the active classifier (Jev, Clef, Clef Flash, OpenAI, or Ollama) for a tier, and for the turn's activity unless `activityRouting` is `off`. It applies local policy and saves the route. Tool continuations reuse that route. Before each continuation, context fit is checked again. A large tool result can move the step to a model with a larger window.
|
|
7
|
+
At the first main step of a logical turn, the Mod reads recent message text and asks the active classifier (Jev, Clef, Clef Flash, Clef Omni, OpenAI, or Ollama) for a tier, and for the turn's activity unless `activityRouting` is `off`. It applies local policy and saves the route. Tool continuations reuse that route, except that a turn whose tool calls edit code moves up to the `code` route, and every `policy.reviewEveryRequests` requests a background classifier review may move it either way ([Changes during a task](user-guide.md#changes-during-a-task)). Before each continuation, context fit is checked again. A large tool result can move the step to a model with a larger window.
|
|
8
8
|
|
|
9
|
-
The classifier receives the current prompt and at most six preceding user or assistant text messages. Each item is limited to 1,200 characters. It does not receive system messages, tool inputs, or tool results. Claude Code's Anthropic credentials never enter the classifier request. Jev, Clef, Clef Flash, and OpenAI receive the text on their services; Ollama receives it on this machine. [Configuration](configuration.md#classifiers) lists each host.
|
|
9
|
+
The classifier receives the current prompt and at most six preceding user or assistant text messages. Each item is limited to 1,200 characters. It does not receive system messages, tool inputs, or tool results. Claude Code's Anthropic credentials never enter the classifier request. Jev, Clef, Clef Flash, Clef Omni, and OpenAI receive the text on their services; Ollama receives it on this machine. [Configuration](configuration.md#classifiers) lists each host.
|
|
10
10
|
|
|
11
11
|
The default tiers are defined in [Configuration](configuration.md#built-in-defaults). Policy uses vote hysteresis, repeated tool errors, context fit, model availability, and switching-cost estimates. The [architecture](architecture.md#request-flow) explains the event boundary and failure handling.
|
|
12
12
|
|
|
@@ -1,12 +1,12 @@
|
|
|
1
|
-
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 672
|
|
1
|
+
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 672 610" role="img" aria-labelledby="title desc">
|
|
2
2
|
<title id="title">Router pane, Classifier tab</title>
|
|
3
3
|
<desc id="desc">One row per classifier to compare: where the prompt goes, its activity probe, its p95 wait and whether its credentials are complete, with Set up for a missing Jev API key; the active classifier’s deadline against its wait, the host that receives the prompt, health and credentials; the agreement with tools across sessions; and Undo after a switch.</desc>
|
|
4
4
|
<style>
|
|
5
5
|
text { font-family: ui-monospace, SFMono-Regular, Menlo, Consolas, 'Liberation Mono', monospace; font-size: 13px; }
|
|
6
6
|
.cap { font-family: system-ui, -apple-system, 'Segoe UI', sans-serif; font-size: 12px; fill: #9a988f; }
|
|
7
7
|
</style>
|
|
8
|
-
<rect width="672" height="
|
|
9
|
-
<rect x="10" y="10" width="652" height="
|
|
8
|
+
<rect width="672" height="610" rx="8" fill="#1c1c1c"/>
|
|
9
|
+
<rect x="10" y="10" width="652" height="590" rx="6" fill="none" stroke="#3f5f8f"/>
|
|
10
10
|
<text x="24.0" y="34" fill="#5fafd7" font-weight="700" xml:space="preserve">ROUTING</text>
|
|
11
11
|
<rect x="94.2" y="20" width="62.4" height="18" rx="4" fill="#87afff"/>
|
|
12
12
|
<text x="94.2" y="34" fill="#0d1016" font-weight="700" xml:space="preserve">[ ◉ on ]</text>
|
|
@@ -49,45 +49,51 @@
|
|
|
49
49
|
<text x="382.8" y="167" fill="#d0d0d0" xml:space="preserve">874 ms</text>
|
|
50
50
|
<text x="460.8" y="167" fill="#87d787" xml:space="preserve">● ready</text>
|
|
51
51
|
<rect x="24.0" y="172" width="124.8" height="18" rx="4" fill="#2e3340"/>
|
|
52
|
-
<text x="24.0" y="186" fill="#d0d0d0" xml:space="preserve">[ ○
|
|
53
|
-
<text x="156.6" y="186" fill="#7c7c7c" xml:space="preserve">api.
|
|
54
|
-
<text x="320.4" y="186" fill="#
|
|
55
|
-
<text x="382.8" y="186" fill="#
|
|
56
|
-
<text x="460.8" y="186" fill="#
|
|
57
|
-
<rect x="570.0" y="172" width="78.0" height="18" rx="4" fill="#2e3340"/>
|
|
58
|
-
<text x="570.0" y="186" fill="#d0d0d0" xml:space="preserve">[ Set up ]</text>
|
|
52
|
+
<text x="24.0" y="186" fill="#d0d0d0" xml:space="preserve">[ ○ Clef Omni ]</text>
|
|
53
|
+
<text x="156.6" y="186" fill="#7c7c7c" xml:space="preserve">api.cloudflare.com</text>
|
|
54
|
+
<text x="320.4" y="186" fill="#7c7c7c" xml:space="preserve">—</text>
|
|
55
|
+
<text x="382.8" y="186" fill="#7c7c7c" xml:space="preserve">—</text>
|
|
56
|
+
<text x="460.8" y="186" fill="#87d787" xml:space="preserve">● ready</text>
|
|
59
57
|
<rect x="24.0" y="191" width="124.8" height="18" rx="4" fill="#2e3340"/>
|
|
60
|
-
<text x="24.0" y="205" fill="#d0d0d0" xml:space="preserve">[ ○
|
|
61
|
-
<text x="156.6" y="205" fill="#
|
|
62
|
-
<text x="320.4" y="205" fill="#d0d0d0" xml:space="preserve">
|
|
63
|
-
<text x="382.8" y="205" fill="#d0d0d0" xml:space="preserve">
|
|
64
|
-
<text x="460.8" y="205" fill="#
|
|
65
|
-
<
|
|
66
|
-
<text x="
|
|
67
|
-
<
|
|
68
|
-
<text x="24.0" y="
|
|
69
|
-
<text x="
|
|
70
|
-
<text x="
|
|
71
|
-
<text x="
|
|
72
|
-
<text x="
|
|
73
|
-
<text x="
|
|
74
|
-
<text x="24.0" y="
|
|
75
|
-
<text x="
|
|
76
|
-
<text x="
|
|
77
|
-
<text x="
|
|
78
|
-
<
|
|
79
|
-
<text x="
|
|
80
|
-
<text x="
|
|
81
|
-
<text x="
|
|
82
|
-
<text x="24.0" y="
|
|
83
|
-
<text x="
|
|
84
|
-
<text x="
|
|
85
|
-
<text x="24.0" y="
|
|
86
|
-
<
|
|
87
|
-
<text x="
|
|
88
|
-
<text x="
|
|
89
|
-
<
|
|
90
|
-
<text x="
|
|
91
|
-
<
|
|
92
|
-
<text x="24.0" y="
|
|
58
|
+
<text x="24.0" y="205" fill="#d0d0d0" xml:space="preserve">[ ○ OpenAI ]</text>
|
|
59
|
+
<text x="156.6" y="205" fill="#7c7c7c" xml:space="preserve">api.openai.com</text>
|
|
60
|
+
<text x="320.4" y="205" fill="#d0d0d0" xml:space="preserve">70/70</text>
|
|
61
|
+
<text x="382.8" y="205" fill="#d0d0d0" xml:space="preserve">286 ms</text>
|
|
62
|
+
<text x="460.8" y="205" fill="#d7af5f" xml:space="preserve">○ no API key </text>
|
|
63
|
+
<rect x="570.0" y="191" width="78.0" height="18" rx="4" fill="#2e3340"/>
|
|
64
|
+
<text x="570.0" y="205" fill="#d0d0d0" xml:space="preserve">[ Set up ]</text>
|
|
65
|
+
<rect x="24.0" y="210" width="124.8" height="18" rx="4" fill="#2e3340"/>
|
|
66
|
+
<text x="24.0" y="224" fill="#d0d0d0" xml:space="preserve">[ ○ Ollama ]</text>
|
|
67
|
+
<text x="156.6" y="224" fill="#87d787" xml:space="preserve">this machine</text>
|
|
68
|
+
<text x="320.4" y="224" fill="#d0d0d0" xml:space="preserve">66/70</text>
|
|
69
|
+
<text x="382.8" y="224" fill="#d0d0d0" xml:space="preserve">3483 ms</text>
|
|
70
|
+
<text x="460.8" y="224" fill="#87d787" xml:space="preserve">● ready</text>
|
|
71
|
+
<text x="24.0" y="243" fill="#7c7c7c" xml:space="preserve"> probe: test prompts with a known activity it answered right, and its p95 wait</text>
|
|
72
|
+
<text x="24.0" y="281" fill="#5fafd7" font-weight="700" xml:space="preserve">CLEF FLASH</text>
|
|
73
|
+
<text x="102.0" y="281" fill="#7c7c7c" xml:space="preserve"> · active</text>
|
|
74
|
+
<text x="24.0" y="300" fill="#d0d0d0" xml:space="preserve">Deadline</text>
|
|
75
|
+
<text x="117.6" y="300" fill="#d0d0d0" xml:space="preserve">3000 ms ▾</text>
|
|
76
|
+
<text x="211.2" y="300" fill="#7c7c7c" xml:space="preserve">p95 874 ms in its probe</text>
|
|
77
|
+
<text x="24.0" y="319" fill="#7c7c7c" xml:space="preserve"> past the deadline the turn keeps its model</text>
|
|
78
|
+
<text x="24.0" y="338" fill="#d0d0d0" xml:space="preserve">Sends</text>
|
|
79
|
+
<text x="117.6" y="338" fill="#d0d0d0" xml:space="preserve">prompt + 6 recent turns → api.cloudflare.com</text>
|
|
80
|
+
<text x="24.0" y="357" fill="#d0d0d0" xml:space="preserve">Health</text>
|
|
81
|
+
<text x="117.6" y="357" fill="#d0d0d0" xml:space="preserve">not paused</text>
|
|
82
|
+
<text x="195.6" y="357" fill="#d0d0d0" xml:space="preserve"> · 0 recent failures</text>
|
|
83
|
+
<text x="24.0" y="376" fill="#d0d0d0" xml:space="preserve">Credentials</text>
|
|
84
|
+
<rect x="117.6" y="362" width="62.4" height="18" rx="4" fill="#2e3340"/>
|
|
85
|
+
<text x="117.6" y="376" fill="#d0d0d0" xml:space="preserve">[ Edit ]</text>
|
|
86
|
+
<text x="180.0" y="376" fill="#7c7c7c" xml:space="preserve"> opens /plugin configure</text>
|
|
87
|
+
<text x="24.0" y="414" fill="#5fafd7" font-weight="700" xml:space="preserve">ACROSS SESSIONS · since the last reset</text>
|
|
88
|
+
<text x="24.0" y="433" fill="#d0d0d0" xml:space="preserve">Labelled 83 turns</text>
|
|
89
|
+
<text x="24.0" y="452" fill="#d0d0d0" xml:space="preserve">Agreement classifier vs tools: 78 of 83 turns (94%)</text>
|
|
90
|
+
<text x="24.0" y="471" fill="#d0d0d0" xml:space="preserve">Code/ops/explore classifier vs tools: 78 of 83 turns (94%)</text>
|
|
91
|
+
<text x="24.0" y="490" fill="#d0d0d0" xml:space="preserve">Mismatch code → read 4 · ops → code 1</text>
|
|
92
|
+
<text x="24.0" y="509" fill="#d0d0d0" xml:space="preserve">Activity moves 9 taken · 4 refused</text>
|
|
93
|
+
<text x="24.0" y="528" fill="#d0d0d0" xml:space="preserve">Escalations after a cheaper activity move: 0 in 7 moves</text>
|
|
94
|
+
<text x="24.0" y="566" fill="#87d787" xml:space="preserve">Saved: classifier Clef → Clef Flash. Applies from the next turn. </text>
|
|
95
|
+
<rect x="538.8" y="552" width="62.4" height="18" rx="4" fill="#2e3340"/>
|
|
96
|
+
<text x="538.8" y="566" fill="#d0d0d0" xml:space="preserve">[ Undo ]</text>
|
|
97
|
+
<rect x="24.0" y="571" width="70.2" height="18" rx="4" fill="#2e3340"/>
|
|
98
|
+
<text x="24.0" y="585" fill="#d0d0d0" xml:space="preserve">[ Close ]</text>
|
|
93
99
|
</svg>
|