From 13b750a85c24a108271ac22e5265f455bd5404e8 Mon Sep 17 00:00:00 2001 From: philipph-askui Date: Thu, 20 Aug 2026 15:57:06 -0700 Subject: [PATCH 01/15] docs(extending): add In-Flight Guidance page + capture step New Extending page documenting the in-run reviewer (enable, cadence, reviewer model, guidance criteria) and a matching desktop-utils-inflight capture step so the screenshot stays in sync with the app UI. Merge after the app PR. Co-Authored-By: Claude Opus 4.8 (1M context) --- content/docs/extending/in-flight-guidance.mdx | 55 +++++++++++++++++++ content/docs/extending/meta.json | 2 +- .../tests/capture_desktop_ui.md | 5 ++ 3 files changed, 61 insertions(+), 1 deletion(-) create mode 100644 content/docs/extending/in-flight-guidance.mdx diff --git a/content/docs/extending/in-flight-guidance.mdx b/content/docs/extending/in-flight-guidance.mdx new file mode 100644 index 0000000..4aff037 --- /dev/null +++ b/content/docs/extending/in-flight-guidance.mdx @@ -0,0 +1,55 @@ +--- +icon: Compass +title: In-Flight Guidance +description: An independent reviewer watches each run every few steps and can nudge the agent back on course — useful for keeping weaker or local models on track. +--- + +In-Flight Guidance puts a second, independent reviewer over the shoulder of the +agent **while a test runs**. Every N steps it looks at the run so far — with a +fresh, clean context — and answers one question: *is this run still on track +toward its objective?* When it isn't, the reviewer sends the agent one short +corrective message to nudge it back on course. Manage it under +**Extending → In-Flight Guidance**. + +It exists for weaker or local (BYOM) models, which tend to drift off-task, loop, +or fail to stop and mark a test failed when something has clearly gone wrong. A +cheap, periodic outside check — optionally from a **stronger** model — makes those +models usably reliable without giving up on running them. + +The In-Flight Guidance tab on the Extending page with the enable toggle, cadence, reviewer model and criteria + +## Settings + +- **Enable** — off by default; set per project. +- **Review every N steps** — how often the reviewer runs (a typical value is + 20–50). Lower is a tighter safety net but costs more reviewer calls; higher is + cheaper. +- **Reviewer model** — *Same as the test run*, or a specific model. Choose a + stronger model to supervise a cheaper or local executor. +- **Guidance criteria** (optional) — free text telling the reviewer what "on + track" means for this project, e.g. *"The run must reach the checkout page; stop + and fail if a login error keeps appearing."* + +Settings are stored in your project at `utils/inflight.json`. + +## During a run + +1. Every N steps the reviewer receives the recent (truncated) trace plus the test + being run, in a clean context. +2. A healthy run gets **no** message — the reviewer has no bias toward intervening. +3. If the run is off track, the agent receives one message (prefixed + `[In-Flight Guidance]`) and honors it next: refocusing, breaking out of a loop, + or stopping and marking the test **failed** when it clearly cannot succeed. + +Every reviewer check — its verdict, rationale and any injected message — is +recorded in the run's report, so a run's outcome stays fully auditable. + + +In-Flight Guidance only sends messages to the running agent. It never edits your +tests, rules, or any project file — improving your suite between runs is what +Dreaming does. The reviewer runs on its own model call over a truncated trace, so +it does not interfere with the agent's prompt caching. + diff --git a/content/docs/extending/meta.json b/content/docs/extending/meta.json index b88a44d..05f1c84 100644 --- a/content/docs/extending/meta.json +++ b/content/docs/extending/meta.json @@ -1 +1 @@ -{ "title": "Extending", "icon": "Puzzle", "pages": ["tools", "custom-tools", "mcp", "secrets", "report-format", "model-providers"] } +{ "title": "Extending", "icon": "Puzzle", "pages": ["tools", "custom-tools", "mcp", "secrets", "report-format", "in-flight-guidance", "model-providers"] } diff --git a/screenshots-capture/tests/capture_desktop_ui.md b/screenshots-capture/tests/capture_desktop_ui.md index 5114476..371ef95 100644 --- a/screenshots-capture/tests/capture_desktop_ui.md +++ b/screenshots-capture/tests/capture_desktop_ui.md @@ -107,6 +107,11 @@ heading) and wait a moment so no tooltip or hover highlight is visible. `save_screenshot[desktop-utils-format]`. Do NOT edit the file or click Create. +15d. Switch to the **In-Flight Guidance** tab. Wait until the "Enable In-Flight + Guidance" card is shown (the enable checkbox with its description), then call + `save_screenshot[desktop-utils-inflight]`. Do NOT toggle it on or change any + field. + 16. With the demo project open, click the git branch chip in the top bar so the source-control flyout opens (changes list, commit field, sync row). Call `save_screenshot[desktop-git-chip]`, then press Escape — do NOT From a2e1e3024f0f30b59c73aa6dcd2c8a79cf8bd19a Mon Sep 17 00:00:00 2001 From: philipph-askui Date: Thu, 20 Aug 2026 17:22:00 -0700 Subject: [PATCH 02/15] docs(extending): In-Flight Guidance is experimental; cadence is a slider, no criteria Match the revised UI: mark the feature experimental, describe the discrete cadence slider (5/20/50/100), and drop the removed Guidance Criteria field. Co-Authored-By: Claude Opus 4.8 (1M context) --- content/docs/extending/in-flight-guidance.mdx | 13 +++++-------- 1 file changed, 5 insertions(+), 8 deletions(-) diff --git a/content/docs/extending/in-flight-guidance.mdx b/content/docs/extending/in-flight-guidance.mdx index 4aff037..3a6f6a9 100644 --- a/content/docs/extending/in-flight-guidance.mdx +++ b/content/docs/extending/in-flight-guidance.mdx @@ -18,20 +18,17 @@ models usably reliable without giving up on running them. The In-Flight Guidance tab on the Extending page with the enable toggle, cadence, reviewer model and criteria ## Settings -- **Enable** — off by default; set per project. -- **Review every N steps** — how often the reviewer runs (a typical value is - 20–50). Lower is a tighter safety net but costs more reviewer calls; higher is - cheaper. +- **Enable** — off by default; set per project. In-Flight Guidance is + **experimental**. +- **Review every N steps** — a slider with discrete stops (5 / 20 / 50 / 100). + Lower is a tighter safety net but costs more reviewer calls; higher is cheaper. - **Reviewer model** — *Same as the test run*, or a specific model. Choose a stronger model to supervise a cheaper or local executor. -- **Guidance criteria** (optional) — free text telling the reviewer what "on - track" means for this project, e.g. *"The run must reach the checkout page; stop - and fail if a login error keeps appearing."* Settings are stored in your project at `utils/inflight.json`. From aaf605a5f920882f38af50f8e3490ea4e35d5c58 Mon Sep 17 00:00:00 2001 From: philipph-askui Date: Thu, 20 Aug 2026 17:33:34 -0700 Subject: [PATCH 03/15] docs(extending): note the BYOM reviewer option for In-Flight Guidance The reviewer model can be your own provider even when the test runs on AskUI. Co-Authored-By: Claude Opus 4.8 (1M context) --- content/docs/extending/in-flight-guidance.mdx | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) diff --git a/content/docs/extending/in-flight-guidance.mdx b/content/docs/extending/in-flight-guidance.mdx index 3a6f6a9..3dd3860 100644 --- a/content/docs/extending/in-flight-guidance.mdx +++ b/content/docs/extending/in-flight-guidance.mdx @@ -27,8 +27,10 @@ models usably reliable without giving up on running them. **experimental**. - **Review every N steps** — a slider with discrete stops (5 / 20 / 50 / 100). Lower is a tighter safety net but costs more reviewer calls; higher is cheaper. -- **Reviewer model** — *Same as the test run*, or a specific model. Choose a - stronger model to supervise a cheaper or local executor. +- **Reviewer model** — *Same as the test run*, a specific hosted model, or **Bring + your own model** (your own provider + key). Choose a stronger model to supervise a + cheaper or local executor — you can point the reviewer at your own provider even + when your test itself runs on AskUI. Settings are stored in your project at `utils/inflight.json`. From d442b059d4185c9f6b24126f91bceeea667ae7ca Mon Sep 17 00:00:00 2001 From: philipph-askui Date: Thu, 20 Aug 2026 21:47:11 -0700 Subject: [PATCH 04/15] docs(extending): note the licence-mode BYOM-only reviewer Co-Authored-By: Claude Opus 4.8 (1M context) --- content/docs/extending/in-flight-guidance.mdx | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/content/docs/extending/in-flight-guidance.mdx b/content/docs/extending/in-flight-guidance.mdx index 3dd3860..cf932a8 100644 --- a/content/docs/extending/in-flight-guidance.mdx +++ b/content/docs/extending/in-flight-guidance.mdx @@ -30,7 +30,8 @@ models usably reliable without giving up on running them. - **Reviewer model** — *Same as the test run*, a specific hosted model, or **Bring your own model** (your own provider + key). Choose a stronger model to supervise a cheaper or local executor — you can point the reviewer at your own provider even - when your test itself runs on AskUI. + when your test itself runs on AskUI. In local-licence mode (no AskUI hosted + models), the reviewer simply runs on your own provider. Settings are stored in your project at `utils/inflight.json`. From ee306a7b6086afa5361cfed541297a54a6964bea Mon Sep 17 00:00:00 2001 From: philipph-askui Date: Thu, 20 Aug 2026 21:52:42 -0700 Subject: [PATCH 05/15] docs(extending): licence-mode reviewer form is pre-filled from the run model Co-Authored-By: Claude Opus 4.8 (1M context) --- content/docs/extending/in-flight-guidance.mdx | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/content/docs/extending/in-flight-guidance.mdx b/content/docs/extending/in-flight-guidance.mdx index cf932a8..a5e97c6 100644 --- a/content/docs/extending/in-flight-guidance.mdx +++ b/content/docs/extending/in-flight-guidance.mdx @@ -31,7 +31,8 @@ models usably reliable without giving up on running them. your own model** (your own provider + key). Choose a stronger model to supervise a cheaper or local executor — you can point the reviewer at your own provider even when your test itself runs on AskUI. In local-licence mode (no AskUI hosted - models), the reviewer simply runs on your own provider. + models), the reviewer runs on your own provider — the form comes pre-filled with + your test-run model's settings, so you only change what differs. Settings are stored in your project at `utils/inflight.json`. From 4d7f4386b63a28d818cae3ad8450f5587d354a8b Mon Sep 17 00:00:00 2001 From: philipph-askui Date: Sun, 23 Aug 2026 20:52:31 -0400 Subject: [PATCH 06/15] docs(extending): In-Flight Guidance reviewer tokens/cost shown in the report Co-Authored-By: Claude Opus 4.8 (1M context) --- content/docs/extending/in-flight-guidance.mdx | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-) diff --git a/content/docs/extending/in-flight-guidance.mdx b/content/docs/extending/in-flight-guidance.mdx index a5e97c6..819e13f 100644 --- a/content/docs/extending/in-flight-guidance.mdx +++ b/content/docs/extending/in-flight-guidance.mdx @@ -45,8 +45,11 @@ Settings are stored in your project at `utils/inflight.json`. `[In-Flight Guidance]`) and honors it next: refocusing, breaking out of a loop, or stopping and marking the test **failed** when it clearly cannot succeed. -Every reviewer check — its verdict, rationale and any injected message — is -recorded in the run's report, so a run's outcome stays fully auditable. +Every reviewer check — its verdict, rationale, any injected message and that check's +token count — is recorded in the run's report, so a run's outcome stays fully +auditable. The run summary also shows a separate **In-Flight Guidance** line (total +tokens, cost and number of checks), priced at the reviewer model's own rate and kept +apart from your execution cost. In-Flight Guidance only sends messages to the running agent. It never edits your From 79c8e91535bd4b529859cbead410ddfaaed6a287 Mon Sep 17 00:00:00 2001 From: philipph-askui Date: Sun, 23 Aug 2026 21:05:50 -0400 Subject: [PATCH 07/15] docs(extending): where to enter In-Flight Guidance reviewer model rates Co-Authored-By: Claude Opus 4.8 (1M context) --- content/docs/extending/in-flight-guidance.mdx | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/content/docs/extending/in-flight-guidance.mdx b/content/docs/extending/in-flight-guidance.mdx index 819e13f..2a0104c 100644 --- a/content/docs/extending/in-flight-guidance.mdx +++ b/content/docs/extending/in-flight-guidance.mdx @@ -33,6 +33,10 @@ models usably reliable without giving up on running them. when your test itself runs on AskUI. In local-licence mode (no AskUI hosted models), the reviewer runs on your own provider — the form comes pre-filled with your test-run model's settings, so you only change what differs. +- **Reviewer model rates** — when the reviewer runs on a custom (BYOM) model, a + token-price block appears below it (input / output / cache $/Mtok) so the report can + show the guidance cost. Hosted reviewer models are priced automatically; the run + model's rate stays on **Settings → Account → Model rates**. Settings are stored in your project at `utils/inflight.json`. From 4cd2240620b0ca38db43310d8e9d377eb8ef95cd Mon Sep 17 00:00:00 2001 From: philipph-askui Date: Mon, 24 Aug 2026 08:38:43 -0400 Subject: [PATCH 08/15] docs(extending): reviewer now sees the latest screenshots to catch loops Co-Authored-By: Claude Opus 4.8 (1M context) --- content/docs/extending/in-flight-guidance.mdx | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) diff --git a/content/docs/extending/in-flight-guidance.mdx b/content/docs/extending/in-flight-guidance.mdx index 2a0104c..2b60931 100644 --- a/content/docs/extending/in-flight-guidance.mdx +++ b/content/docs/extending/in-flight-guidance.mdx @@ -43,7 +43,9 @@ Settings are stored in your project at `utils/inflight.json`. ## During a run 1. Every N steps the reviewer receives the recent (truncated) trace plus the test - being run, in a clean context. + being run — including the latest few screenshots as real images — in a clean + context, so it can see when the screen has stopped changing (a stuck loop). Use a + vision-capable reviewer model. 2. A healthy run gets **no** message — the reviewer has no bias toward intervening. 3. If the run is off track, the agent receives one message (prefixed `[In-Flight Guidance]`) and honors it next: refocusing, breaking out of a loop, From 0f88719b9db1670e184464ff3c9b9fe6b854c7ba Mon Sep 17 00:00:00 2001 From: philipph-askui Date: Mon, 24 Aug 2026 10:33:41 -0400 Subject: [PATCH 09/15] docs(extending): In-Flight Guidance cost is per-test; top-line cost is all-in Co-Authored-By: Claude Opus 4.8 (1M context) --- content/docs/extending/in-flight-guidance.mdx | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/content/docs/extending/in-flight-guidance.mdx b/content/docs/extending/in-flight-guidance.mdx index 2b60931..850d3db 100644 --- a/content/docs/extending/in-flight-guidance.mdx +++ b/content/docs/extending/in-flight-guidance.mdx @@ -53,9 +53,9 @@ Settings are stored in your project at `utils/inflight.json`. Every reviewer check — its verdict, rationale, any injected message and that check's token count — is recorded in the run's report, so a run's outcome stays fully -auditable. The run summary also shows a separate **In-Flight Guidance** line (total -tokens, cost and number of checks), priced at the reviewer model's own rate and kept -apart from your execution cost. +auditable. Each **test section** shows its own In-Flight Guidance roll-up (tokens, cost +and number of checks) priced at the reviewer model's own rate, and the report's +top-line **Cost** is all-in — test execution plus In-Flight Guidance. In-Flight Guidance only sends messages to the running agent. It never edits your From 00aa988ffdb2ed5d5a2fc5d51b9a24982cf90586 Mon Sep 17 00:00:00 2001 From: philipph-askui Date: Mon, 24 Aug 2026 11:53:02 -0400 Subject: [PATCH 10/15] docs(extending): cadence slider now includes 10 (5 / 10 / 20 / 50 / 100) Co-Authored-By: Claude Opus 4.8 (1M context) --- content/docs/extending/in-flight-guidance.mdx | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/content/docs/extending/in-flight-guidance.mdx b/content/docs/extending/in-flight-guidance.mdx index 850d3db..24ff423 100644 --- a/content/docs/extending/in-flight-guidance.mdx +++ b/content/docs/extending/in-flight-guidance.mdx @@ -25,7 +25,7 @@ models usably reliable without giving up on running them. - **Enable** — off by default; set per project. In-Flight Guidance is **experimental**. -- **Review every N steps** — a slider with discrete stops (5 / 20 / 50 / 100). +- **Review every N steps** — a slider with discrete stops (5 / 10 / 20 / 50 / 100). Lower is a tighter safety net but costs more reviewer calls; higher is cheaper. - **Reviewer model** — *Same as the test run*, a specific hosted model, or **Bring your own model** (your own provider + key). Choose a stronger model to supervise a From ce1df99df2adc986f175e5004f8416a0082da776 Mon Sep 17 00:00:00 2001 From: philipph-askui Date: Mon, 24 Aug 2026 12:03:30 -0400 Subject: [PATCH 11/15] docs(extending): reviewer rates now live in the provider form's Model rates section Co-Authored-By: Claude Opus 4.8 (1M context) --- content/docs/extending/in-flight-guidance.mdx | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/content/docs/extending/in-flight-guidance.mdx b/content/docs/extending/in-flight-guidance.mdx index 24ff423..ee2c23a 100644 --- a/content/docs/extending/in-flight-guidance.mdx +++ b/content/docs/extending/in-flight-guidance.mdx @@ -33,10 +33,10 @@ models usably reliable without giving up on running them. when your test itself runs on AskUI. In local-licence mode (no AskUI hosted models), the reviewer runs on your own provider — the form comes pre-filled with your test-run model's settings, so you only change what differs. -- **Reviewer model rates** — when the reviewer runs on a custom (BYOM) model, a - token-price block appears below it (input / output / cache $/Mtok) so the report can - show the guidance cost. Hosted reviewer models are priced automatically; the run - model's rate stays on **Settings → Account → Model rates**. +- **Reviewer model rates** — when the reviewer runs on a custom (BYOM) model, its + provider form includes a **Model rates** section (input / output / cache $/Mtok) so + the report can show the guidance cost. Hosted reviewer models are priced + automatically. Settings are stored in your project at `utils/inflight.json`. From 31a76000849c4d208c5290f26099df0a24f595c9 Mon Sep 17 00:00:00 2001 From: philipph-askui Date: Mon, 24 Aug 2026 12:19:01 -0400 Subject: [PATCH 12/15] docs(extending): per-test Cost all-in; guidance collapsible; reviewer model in provenance Co-Authored-By: Claude Opus 4.8 (1M context) --- content/docs/extending/in-flight-guidance.mdx | 8 +++++--- 1 file changed, 5 insertions(+), 3 deletions(-) diff --git a/content/docs/extending/in-flight-guidance.mdx b/content/docs/extending/in-flight-guidance.mdx index ee2c23a..80617d6 100644 --- a/content/docs/extending/in-flight-guidance.mdx +++ b/content/docs/extending/in-flight-guidance.mdx @@ -53,9 +53,11 @@ Settings are stored in your project at `utils/inflight.json`. Every reviewer check — its verdict, rationale, any injected message and that check's token count — is recorded in the run's report, so a run's outcome stays fully -auditable. Each **test section** shows its own In-Flight Guidance roll-up (tokens, cost -and number of checks) priced at the reviewer model's own rate, and the report's -top-line **Cost** is all-in — test execution plus In-Flight Guidance. +auditable. A test's **Cost** is its total (its own execution plus its In-Flight +Guidance, priced at the reviewer model's own rate), and the top-line **Cost** is the +sum of those. The guidance share is broken out quietly as a small collapsible under +each test's metrics, and the reviewer model is shown once in the run's provenance +strip. In-Flight Guidance only sends messages to the running agent. It never edits your From 57f28682d0470c8af36fc5d772a4476847347eb3 Mon Sep 17 00:00:00 2001 From: philipph-askui Date: Tue, 25 Aug 2026 08:45:28 -0400 Subject: [PATCH 13/15] docs(extending): licence-mode 'Same as for my test runs' checkbox for the reviewer Co-Authored-By: Claude Opus 4.8 (1M context) --- content/docs/extending/in-flight-guidance.mdx | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/content/docs/extending/in-flight-guidance.mdx b/content/docs/extending/in-flight-guidance.mdx index 80617d6..d35383c 100644 --- a/content/docs/extending/in-flight-guidance.mdx +++ b/content/docs/extending/in-flight-guidance.mdx @@ -31,8 +31,9 @@ models usably reliable without giving up on running them. your own model** (your own provider + key). Choose a stronger model to supervise a cheaper or local executor — you can point the reviewer at your own provider even when your test itself runs on AskUI. In local-licence mode (no AskUI hosted - models), the reviewer runs on your own provider — the form comes pre-filled with - your test-run model's settings, so you only change what differs. + models), a **"Same as for my test runs"** checkbox (on by default) reuses your run + model for the reviewer; untick it to configure a separate reviewer provider (its + form is pre-filled from your test-run model, so you only change what differs). - **Reviewer model rates** — when the reviewer runs on a custom (BYOM) model, its provider form includes a **Model rates** section (input / output / cache $/Mtok) so the report can show the guidance cost. Hosted reviewer models are priced From 43b06182eacf41fdd8c2392814ba75e6345d670f Mon Sep 17 00:00:00 2001 From: philipph-askui Date: Tue, 25 Aug 2026 09:52:35 -0400 Subject: [PATCH 14/15] In-Flight Guidance: document weak-reviewer robustness + timeout safety --- content/docs/extending/in-flight-guidance.mdx | 13 +++++++++++++ 1 file changed, 13 insertions(+) diff --git a/content/docs/extending/in-flight-guidance.mdx b/content/docs/extending/in-flight-guidance.mdx index d35383c..cc54451 100644 --- a/content/docs/extending/in-flight-guidance.mdx +++ b/content/docs/extending/in-flight-guidance.mdx @@ -52,6 +52,19 @@ Settings are stored in your project at `utils/inflight.json`. `[In-Flight Guidance]`) and honors it next: refocusing, breaking out of a loop, or stopping and marking the test **failed** when it clearly cannot succeed. +The reviewer only has to *classify* — it never has to phrase the fix — so it works even +on a small or local model. Alongside the on/off judgment it returns one decisive flag, +*cannot succeed*, for a run that can no longer reach its goal (stuck with no approach +left, the same screen recurring, an impossible objective). When it's set, the **app** +issues a fixed stop-and-fail directive using the reviewer's rationale as the reason — so +a weak reviewer that can only notice a doomed run still yields a clean FAIL instead of an +endless loop. The reviewer is also never allowed to tell the agent to "try again", the +one nudge that keeps a loop going. + +The reviewer can neither crash nor stall a run: reviewer errors are recorded and the run +continues, and because the check runs inline each reviewer call is **time-bounded** — a +slow or hung reviewer is skipped for that step, not waited on. + Every reviewer check — its verdict, rationale, any injected message and that check's token count — is recorded in the run's report, so a run's outcome stays fully auditable. A test's **Cost** is its total (its own execution plus its In-Flight From a408181f9a8eb473d40b16c8d0be7dd17f234bdb Mon Sep 17 00:00:00 2001 From: philipph-askui Date: Tue, 25 Aug 2026 10:08:55 -0400 Subject: [PATCH 15/15] In-Flight Guidance: note that checks also appear live in the conversation log --- content/docs/extending/in-flight-guidance.mdx | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/content/docs/extending/in-flight-guidance.mdx b/content/docs/extending/in-flight-guidance.mdx index cc54451..c218d3c 100644 --- a/content/docs/extending/in-flight-guidance.mdx +++ b/content/docs/extending/in-flight-guidance.mdx @@ -65,6 +65,10 @@ The reviewer can neither crash nor stall a run: reviewer errors are recorded and continues, and because the check runs inline each reviewer call is **time-bounded** — a slow or hung reviewer is skipped for that step, not waited on. +Each reviewer check also appears **inline in the live Conversation Log** while the run +executes — its own "In-Flight Guidance" notice, quiet when on track and showing the sent +message when it intervenes — so you can watch the Guide work, not just read it afterwards. + Every reviewer check — its verdict, rationale, any injected message and that check's token count — is recorded in the run's report, so a run's outcome stays fully auditable. A test's **Cost** is its total (its own execution plus its In-Flight