fix(observability): the coworker page after the live check (T3-16)

(a) The history table has fixed columns and wrapping headers, so it stays
inside its card; a version kept without its files is still numbered (v1),
so its row is no longer the long ⓕ fingerprint that wrapped. (b) "too few
sessions to compare" is centred across Typical tokens and Approvals asked.
(d) The coworker's allowed models read "Can use: …" and the history's
"Models used in the last 30 days: …", which now go through the canonical
rule (only the openai: prefix is dropped, as canonical_model on
issue/ope-247). (e) The Default mode and Workspace lines are gone.
(c) and (f) were already fixed in 3942ed9e and 96aae337; the backend on 8765
was started before them and needs a restart to show them.
This commit is contained in:
Devika Verma
2026-11-05 16:43:28 +05:30
parent e609534ded
commit cffbce91fd
9 changed files with 60 additions and 49 deletions
+13 -2
View File
@@ -93,6 +93,13 @@ def heaviest(store: Any, *, start: int, end: int, limit: int = 20,
ASKED = ("unsure", "context_changed", "could_not_run")
def canonical_model(model: str) -> str:
"""One name per model (T3-16, T3-4): OpenAI is the default provider, so `openai:gpt-5.6-sol`
and `gpt-5.6-sol` are the same model. Only the `openai:` prefix is dropped, the same rule
as `canonical_model` on issue/ope-247-openai-model-prefix; every other provider keeps its."""
return model[len("openai:"):] if isinstance(model, str) and model.startswith("openai:") else model
def approver(store: Any, *, start: int, end: int, persona_id: Optional[str] = None) -> dict[str, Any]:
"""`GET /v1/observability/approver`: the Auto-approve strip (§7.3). Live checks by result
(asked you includes could-not-run: the card falls back to asking), overridden tool
@@ -177,7 +184,8 @@ def persona_doing(store: Any, persona_id: str, *, start: int, end: int, current_
Usage page. Fewer than `min_runs` runs: no medians, "too few runs to compare"."""
turns = store.usage_turns(start, end, where={"persona_id": persona_id})
kept_from = store.usage_kept_from()
models_of_turn = {t["turn_id"]: t.get("own_models", t["models"]) for t in turns}
models_of_turn = {t["turn_id"]: list(dict.fromkeys(canonical_model(m) for m in t.get("own_models", t["models"])))
for t in turns}
runs: dict[tuple, dict[str, Any]] = {}
for t in turns:
k = (t["persona_fp"], t["session_id"])
@@ -220,7 +228,10 @@ def persona_doing(store: Any, persona_id: str, *, start: int, end: int, current_
changes = {}
for prev, cur in zip(ordered, ordered[1:]):
changes[cur["fp"]] = {"from": prev["fp"], **diff_versions(prev, cur, store)}
number = {v["fp"]: i + 1 for i, v in enumerate(ordered)}
# a version whose files were not kept still gets its number, by when it was first seen here
fps = [v["fp"] for v in ordered]
fps += [r["fp"] for r in sorted(out, key=lambda r: r["first_seen"] or 0) if r["fp"] and r["fp"] not in fps]
number = {fp: i + 1 for i, fp in enumerate(fps)}
for row in out:
row["changed"] = changes.get(row["fp"])
# T3-15: "v1 · current": versions numbered oldest first, by first sighting
@@ -82,6 +82,7 @@ describe("PersonaDoing", () => {
const tooFew = screen.getByTestId("persona-doing-too-few");
expect(tooFew.textContent).toBe("too few sessions to compare");
expect(tooFew.getAttribute("colspan")).toBe("2");
expect(tooFew.className).toContain("text-center"); // T3-16 (b): across both columns, not under the second
expect(b.textContent).toContain("or earlier");
expect(b.textContent).toContain("2 / 1");
// a later version can be compared
@@ -101,7 +102,7 @@ describe("PersonaDoing", () => {
expect(screen.getByTestId("persona-doing-changed").textContent).toContain("First version seen on this machine.");
expect(screen.getByTestId("persona-doing-changed").textContent).not.toContain("unchanged");
const line = screen.getByTestId("persona-doing-models");
expect(line.textContent).toBe("Models used: gpt-5.6-sol and 2 more");
expect(line.textContent).toBe("Models used in the last 30 days: gpt-5.6-sol and 2 more");
expect(line.getAttribute("title")).toBe("gpt-5.6-sol, model-a, model-b");
});
+7 -3
View File
@@ -41,7 +41,11 @@ export function PersonaDoing({ personaId, machineId, secClass, scrollIntoView }:
{doing.rows.length === 0 ? (
<div className="text-meta text-faint">{t("p3.doing.none")}</div>
) : (
<table className="p3-t">
<table className="p3-t p3-doing">
<colgroup>
<col style={{ width: "15%" }} /><col style={{ width: "19%" }} /><col style={{ width: "11%" }} />
<col style={{ width: "15%" }} /><col style={{ width: "14%" }} /><col style={{ width: "12%" }} /><col style={{ width: "14%" }} />
</colgroup>
<thead>
<tr>
<th>{t("p3.doing.version")}</th>
@@ -120,14 +124,14 @@ function DoingTr({ r, t }: { r: DoingRow; t: T }) {
{r.current && <span className="text-ok"> · {t("p3.doing.current")}</span>}
</span>
</td>
<td className="whitespace-nowrap">{inUseSpan(r.first_seen, r.last_seen, r.before_kept, t)}</td>
<td>{inUseSpan(r.first_seen, r.last_seen, r.before_kept, t)}</td>
<td className="n">
{r.runs}
{r.automation_runs > 0 && <span className="text-faint"> {t("p3.doing.automation_runs", { count: r.automation_runs })}</span>}
</td>
{r.too_few ? (
// T3-15 (h): one note across both typical-number columns
<td className="n text-faint" colSpan={2} data-testid="persona-doing-too-few">{t("p3.doing.too_few")}</td>
<td className="text-faint text-center" colSpan={2} data-testid="persona-doing-too-few">{t("p3.doing.too_few")}</td>
) : (
<>
<td className="n">{(r.at_least ? "≥ " : "") + formatTokens(r.median_tokens ?? 0)}</td>
@@ -80,8 +80,12 @@ describe("PersonaView", () => {
expect(screen.getAllByText(/Ready/).length).toBeGreaterThan(0);
expect(screen.getByText("Connect")).toBeTruthy(); // datadog (core, not connected)
expect(screen.getByText("Add")).toBeTruthy(); // filesystem (mcp, not connected)
// defaults footer
// the models it may use, labelled apart from the models its sessions used (T3-16 d);
// Default mode and Workspace are gone (T3-16 e)
expect(screen.getByText("claude-opus-4-8")).toBeTruthy();
expect(screen.getByTestId("persona-can-use").textContent).toBe("Can use: claude-opus-4-8, gpt-5.5");
expect(screen.queryByText(/Default mode/)).toBeNull();
expect(screen.queryByText(/Workspace/)).toBeNull();
});
it("toggling a default connection POSTs /connections and applies the returned defaults", async () => {
+13 -23
View File
@@ -360,29 +360,19 @@ export function PersonaView({
{/* Phase 3 OB-15: how each version of this coworker has been doing */}
<PersonaDoing personaId={personaId} machineId={mid} secClass={SEC_H} scrollIntoView={focus === "doing"} />
{/* defaults footer */}
<section className="flex flex-wrap gap-x-8 gap-y-2 text-ui">
{detail.recommended_models.length > 0 && (
<div>
<span className="text-faint">{t("persona.models_label")}</span> ·{" "}
{detail.recommended_models.map((m, i) => (
<span key={m}>
<span className="font-mono">{m}</span>
{i < detail.recommended_models.length - 1 ? ", " : ""}
</span>
))}
</div>
)}
{detail.default_permission_mode && (
<div>
<span className="text-faint">{t("persona.default_mode_label")}</span> · {detail.default_permission_mode}
</div>
)}
<div>
<span className="text-faint">{t("persona.workspace_label")}</span> ·{" "}
{detail.requires_folder ? t("persona.workspace_picked") : t("persona.workspace_scratch")}
</div>
</section>
{/* the models this coworker may use (T3-16: "Can use", apart from the models its
sessions used, under How it's been doing; Default mode and Workspace removed) */}
{detail.recommended_models.length > 0 && (
<section className="text-ui" data-testid="persona-can-use">
<span className="text-faint">{t("persona.can_use")}</span>{" "}
{detail.recommended_models.map((m, i) => (
<span key={m}>
<span className="font-mono">{m}</span>
{i < detail.recommended_models.length - 1 ? ", " : ""}
</span>
))}
</section>
)}
{/* management — the controls that left the list page (UX-035) */}
<section className="border-t border-line pt-4 flex flex-wrap items-center gap-x-5 gap-y-2 text-ui">
+4 -8
View File
@@ -1532,19 +1532,16 @@
"connections_desc": "Declared by the persona — wire {{name}} into these to unlock its full workflow.",
"connections_full_benefit": "Connections for full benefit",
"core_tag": "core",
"default_mode_label": "Default mode",
"disabled": "Disabled",
"enable_title": "Enable this coworker",
"enabled": "Enabled",
"hero_art": "{{slug}} artwork",
"load_error": "Could not load this coworker.",
"loading": "Loading…",
"models_label": "Models",
"new_sessions_default": "New sessions get by default",
"new_sessions_desc": "When you start a {{name}} session these are enabled automatically. You can still mute any of them per session.",
"on_by_default": "On by default for new sessions",
"persona": "Coworker",
"workspace_label": "Workspace",
"prev_screenshot": "Previous screenshot",
"next_screenshot": "Next screenshot",
"screenshot_alt": "{{name}} screenshot {{n}}",
@@ -1554,8 +1551,6 @@
"col_enable": "Enable",
"advanced": "Advanced",
"tool_calls": "Tool calls",
"workspace_picked": "picked folder",
"workspace_scratch": "scratch",
"show_in_picker": "Show in picker",
"default_for_new": "Default for new sessions",
"make_default": "Make default",
@@ -1566,7 +1561,8 @@
"delete": "Delete",
"keep": "Keep",
"delete_ellipsis": "Delete…",
"delete_failed": "delete failed"
"delete_failed": "delete failed",
"can_use": "Can use:"
},
"personas": {
"add_help": "Load from a local directory or a public GitHub repo. Files are copied into a managed area (a snapshot), so the source can change without affecting the install.",
@@ -4100,8 +4096,8 @@
"ratings": "Ratings 👍 / 👎",
"today": "today",
"or_earlier": "{{date}} or earlier",
"models_used_one": "Models used: {{model}}",
"models_used_more": "Models used: {{model}} and {{count}} more",
"models_used_one": "Models used in the last 30 days: {{model}}",
"models_used_more": "Models used in the last 30 days: {{model}} and {{count}} more",
"nothing_to_compare": "Nothing to compare yet"
},
"wake": {
+4 -8
View File
@@ -1511,19 +1511,16 @@
"connections_desc": "由角色声明 —— 把 {{name}} 接入这些连接以解锁它的完整工作流。",
"connections_full_benefit": "充分发挥的连接",
"core_tag": "核心",
"default_mode_label": "默认模式",
"disabled": "已禁用",
"enable_title": "启用此同事",
"enabled": "已启用",
"hero_art": "{{slug}} 配图",
"load_error": "无法加载此同事。",
"loading": "加载中…",
"models_label": "模型",
"new_sessions_default": "新会话默认启用",
"new_sessions_desc": "当你开始一个 {{name}} 会话时,这些会自动启用。你仍可在每个会话中静音任意一个。",
"on_by_default": "新会话默认开启",
"persona": "同事",
"workspace_label": "工作区",
"prev_screenshot": "上一张截图",
"next_screenshot": "下一张截图",
"screenshot_alt": "{{name}} 截图 {{n}}",
@@ -1533,8 +1530,6 @@
"col_enable": "启用",
"advanced": "高级",
"tool_calls": "工具调用",
"workspace_picked": "选定的文件夹",
"workspace_scratch": "临时目录",
"show_in_picker": "显示在选择器中",
"default_for_new": "新会话默认",
"make_default": "设为默认",
@@ -1545,7 +1540,8 @@
"delete": "删除",
"keep": "保留",
"delete_ellipsis": "删除…",
"delete_failed": "删除失败"
"delete_failed": "删除失败",
"can_use": "可以使用:"
},
"personas": {
"add_help": "从本地目录或公开 GitHub 仓库加载。文件会复制到托管区域(快照),因此源文件变更不会影响已安装版本。",
@@ -4033,8 +4029,8 @@
"ratings": "评价 👍 / 👎",
"today": "今天",
"or_earlier": "{{date}} 或更早",
"models_used_one": "使用的模型:{{model}}",
"models_used_more": "使用的模型:{{model}} 等另外 {{count}} 个",
"models_used_one": "最近 30 天使用的模型:{{model}}",
"models_used_more": "最近 30 天使用的模型:{{model}} 等另外 {{count}} 个",
"nothing_to_compare": "还没有可比较的版本"
},
"wake": {
+4
View File
@@ -2662,6 +2662,10 @@ html[data-theme="dark"] { --bar-model:#4c8dff; --bar-tool:#3fb8ad; --bar-wait:#3
.p3-t .n { text-align:right; font-variant-numeric:tabular-nums; white-space:nowrap; }
/* T3-5: Heaviest turns stays inside its card: fixed columns, text cut with the full text on hover */
.p3-heavy { table-layout:fixed; }
/* T3-16: the coworker history table stays inside its card; headers wrap */
.p3-doing { table-layout:fixed; }
.p3-doing th { white-space:normal; vertical-align:bottom; }
.p3-doing td { overflow-wrap:anywhere; }
.p3-heavy td { overflow:hidden; }
.p3-cut { display:block; overflow:hidden; text-overflow:ellipsis; white-space:nowrap; }
.p3-badges { display:flex; flex-wrap:nowrap; gap:3px; min-width:0; }
@@ -198,13 +198,18 @@ def test_a_versions_models_are_its_own_work_only(tmp_path):
t["stamp"].update({"persona_id": "cowork", "persona_fp": "fp1"})
store.write_turn(t)
usage = {"reported": True, "input_fresh": 10, "cache_read": 0, "cache_write": 0, "output": 1}
for n, (purpose, model) in enumerate([("main", "model-x"), ("helper", "model-h"), ("auto_approver", "reviewer-model"),
for n, (purpose, model) in enumerate([("main", "model-x"), ("helper", "model-h"), ("main", "openai:model-h"),
("main", "together:model-h"), ("auto_approver", "reviewer-model"),
("title", "title-model"), ("summary", "summary-model")], start=1):
store.write_step(_step("T1", "s1", n, RECENT + n, "model", {"purpose": purpose, "model": model, "usage": usage}))
store.write_issues("T1", {"signals_version": signals.version(), "warn": 0, "info": 0, "issues": [], "fine": []}, seen=True)
store.flush()
got = usage_api.persona_doing(store, "cowork", start=RECENT - 1, end=RECENT + 60_000)
[row] = got["rows"]
assert row["models"] == ["model-h", "model-x"]
assert row["model_line"] == {"kind": "first", "models": ["model-h", "model-x"]}
# T3-16 (d): openai:model-h is model-h (only the openai: prefix goes); together: keeps its own
assert row["models"] == ["model-h", "model-x", "together:model-h"]
assert row["model_line"] == {"kind": "first", "models": ["model-h", "model-x", "together:model-h"]}
assert row["version_n"] == 1 # T3-16: numbered even when its files were not kept
assert usage_api.canonical_model("openai:gpt-5.6-sol") == "gpt-5.6-sol"
assert usage_api.canonical_model("openrouter:openai/gpt-5") == "openrouter:openai/gpt-5"
store.close()