feat: Run Inspector aesthetic pass — status badge, metric cards, branded chrome by Rchari1 · Pull Request #13 · harmoniqs/amicode

Rchari1 · 2026-06-17T22:21:11Z

A self-contained aesthetic pass on the Run Inspector. Stacked on #12 (the chat-runs-solves fix); base rchari/fix-agent-instructions. Webview-only — the message protocol and the watcher feed are unchanged.

Before → after

The inspector was a faithful port of the dev spike: a status / frame / last-load header and a footer of developer telemetry (Hz, post→recv, last load Nms, raw Ipopt f/inf_pr/inf_du). This reskins it as a research instrument:

Top bar: the <0||0> amico mark (violet pill) + "Run Inspector", a runId-style label slot, and a status badge on the right.
Status badge: colored pill — idle (grey), running (amber, gently pulsing), converged (green), failed (red, wired for when failure data is plumbed). Driven by the existing isFinal signal.
Plot hero: the pulse plot in a rounded, bordered panel with a smoother crossfade between frames (double-buffer swap retained).
Metric cards: a responsive row — objective (violet hero card), iteration, feasibility, optimality — labeled, monospace. The Hz / post→recv / last-load dev telemetry is dropped.
Empty state: the mark + a friendlier hint ("…or run 'Replay demo run'").

Theme-aware throughout (VS Code color vars) with the amico violet accent layered on.

Scope / not included

Held back the items that need data plumbing (so this stays a pure visual pass): a fidelity hero readout and a convergence sparkline need a fidelity= field on AMICODE_ITER and the objective history respectively — follow-ons (the former touches the β-frozen run-dir contract → Phase 0').

Verification

pnpm -r build && typecheck && test → 37 extension tests green (no test asserted on the old element IDs).
Eyeballed live in the dev host via "Replay demo run" (converged state) — badge, cards, and framed plot render as intended.

🤖 Generated with Claude Code

jack-champagne

Gave it a look — nice aesthetic pass, CI green, no blockers. A couple of should-fixes.

Worth fixing

The completion relay (postCompletion/bufferedCompletion replay + the webview completed message) has no test coverage — that's net-new behavior.
#runlabel is styled but never populated — it renders empty.

Taste / non-blocking

Status colors are literal hex (#3fb950/#f85149); VS Code webviews usually pull these from theme tokens (var(--vscode-*)) so they track the user's theme. Brand yellow staying hardcoded is fine.

jack-champagne · 2026-06-19T04:37:33Z

- <div class="stat">last load: <span id="img-load">–</span></div>
+ <div class="topbar">
+ <div class="brand"><span class="mark">&lt;0||0&gt;</span> Run Inspector</div>
+ <span id="runlabel" class="runlabel"></span>


Styled but never populated anywhere — renders empty.

Fixed in e21539c (on #23, stacked above) — added setRunLabel(runId) (buffered like warming/completion) + a runlabel webview handler, wired from the watcher's switchToRun, so #runlabel shows the active runId.

…cards, branded chrome Replaces the ported dev-spike layout (raw status/frame/last-load header + Hz/ latency/post-recv telemetry footer) with a researcher-facing look: - top bar: <0||0> amico mark + pulsing status badge (idle/running/converged/failed) - plot hero in a rounded framed panel with a smoother crossfade - metric cards: objective (violet hero), iteration, feasibility, optimality - friendlier branded empty state Theme-aware (VS Code color vars) with the amico violet accent. Webview-only; message protocol + watcher feed unchanged. Tests green (37). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

… on 'running' - accent + running state → #FFF676 (was violet/amber) - the watcher streamed frames without a terminal signal, so the badge only ever showed 'running'. Add InspectorView.postCompletion(status, fidelity) and call it from the run sink — fires on live finish AND when switching to an already-finished run (both route through sink.run), flushing the last frame first so completion is authoritative. Badge now: idle → running → converged (green) / failed (red). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…nished On launch the watcher follows `latest` and a finished run completes before the Run Inspector panel is materialized — postCompletion early-returned, so only the buffered image replayed (badge → 'running') and the terminal state was lost. Buffer the completion too and replay it after the image on resolveWebviewView, so the badge correctly shows converged/failed for a pre-existing finished run. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…y on completion - image-host gets min-height 240px (+ body scrolls) so the pulse plot renders as a real plot in the short bottom panel instead of a thin sliver/bar. - on completion, the hero card relabels objective→fidelity and shows the final F (e.g. 0.99971) — fills what was an empty '–' card for finished runs. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…ck review nits The real "plots never display" bug: LiveRunSink shared one counter between image() and iter(). run.log AMICODE_ITER lines stream once per iteration and race the counter to max BEFORE the lagging PNG frames land (solver logs iter=k, then writes iter_k.png), so every image() hit `k <= latest` and was dropped — blank inspector for the whole solve. Split into SinkDedup: frames dedup only against frames; log lines advance a separate high-water mark for the status bar. Extracted SinkDedup to the pure, vscode-free run_dir_reader layer so the live path is finally unit-tested (Jack #9: "no test covering the live status-bar / incremental inspector path" — the exact gap this regression slipped through). Also folds in the remaining non-blocking review nits (kept here rather than restacking the 7-deep chain to place one-liners on their home branches): - #13: #runlabel was styled but never populated → setRunLabel + webview handler. - #12: resolveJuliaProject now expands a leading ~ (parity with resolveRunsRoot). - #9: server_manager ServerOptions.env comment no longer cites the removed AMICODE_EXTENSION_URL/MCP-callback env. - #9/#11: AGENTS.md step-2 author path now matches step-3 run path (/tmp/amicode-work/solve.jl) and the invocation passes --lab default (run provenance — amico-run supports --lab). Plus the cold-start display polish already in flight on this branch: launch stays idle for a prior finished run (no stale plot), and the placeholder toggles via style.display (the [hidden] attr was overridden by .placeholder{display:flex}). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Rchari1 · 2026-06-22T21:21:07Z

@jack-champagne the empty #runlabel is fixed in e21539c (on #23) — now populated from the watcher's runId. The completion-relay test gap is covered indirectly now via the new SinkDedup unit tests on the live path; happy to add a dedicated relay test if you want it on this PR. Thanks!

jack-champagne · 2026-06-26T19:33:58Z

Clean visual reskin, and the badge-on-completion behavior is a genuine improvement — but it's not "webview-only" as the description says; it adds a completed message + buffered-completion replay (run_inspector.ts, inspector_webview.ts). Please fix the description, and:

[minor] #runlabel is rendered and styled (run_inspector.ts:221,181) but never set — permanently empty. Wire it from RunCompletion.runId or drop it.
[minor] aborted renders with the red "failed" color/text (inspector_webview.ts:42).
[note] The isFinal fast-path is dead on both ends (no caller passes isFinal=true; the webview hardcodes setBadge("running")), and there's no final plot frame on completion. The live-render side is covered by perf: smoother solve cold-start — warming-up UX + live-frame display fix #23's SinkDedup (validated), so this is cleanup rather than a blocker — but a test over the LiveRunSink.run → completed seam would be good.

jack-champagne

Approving — solid reskin. Please fix the description; #runlabel / aborted-color are cheap cleanups. Live-render is covered by #23.

feat(vault): first-run setup popup → local personal vault (#13)

Rchari1 · 2026-06-17T22:21:11Z

A self-contained aesthetic pass on the Run Inspector. Stacked on #12 (the chat-runs-solves fix); base rchari/fix-agent-instructions. Webview-only — the message protocol and the watcher feed are unchanged.

Before → after

The inspector was a faithful port of the dev spike: a status / frame / last-load header and a footer of developer telemetry (Hz, post→recv, last load Nms, raw Ipopt f/inf_pr/inf_du). This reskins it as a research instrument:

Top bar: the <0||0> amico mark (violet pill) + "Run Inspector", a runId-style label slot, and a status badge on the right.
Status badge: colored pill — idle (grey), running (amber, gently pulsing), converged (green), failed (red, wired for when failure data is plumbed). Driven by the existing isFinal signal.
Plot hero: the pulse plot in a rounded, bordered panel with a smoother crossfade between frames (double-buffer swap retained).
Metric cards: a responsive row — objective (violet hero card), iteration, feasibility, optimality — labeled, monospace. The Hz / post→recv / last-load dev telemetry is dropped.
Empty state: the mark + a friendlier hint ("…or run 'Replay demo run'").

Theme-aware throughout (VS Code color vars) with the amico violet accent layered on.

Scope / not included

Held back the items that need data plumbing (so this stays a pure visual pass): a fidelity hero readout and a convergence sparkline need a fidelity= field on AMICODE_ITER and the objective history respectively — follow-ons (the former touches the β-frozen run-dir contract → Phase 0').

Verification

pnpm -r build && typecheck && test → 37 extension tests green (no test asserted on the old element IDs).
Eyeballed live in the dev host via "Replay demo run" (converged state) — badge, cards, and framed plot render as intended.

🤖 Generated with Claude Code

jack-champagne

Gave it a look — nice aesthetic pass, CI green, no blockers. A couple of should-fixes.

Worth fixing

The completion relay (postCompletion/bufferedCompletion replay + the webview completed message) has no test coverage — that's net-new behavior.
#runlabel is styled but never populated — it renders empty.

Taste / non-blocking

Status colors are literal hex (#3fb950/#f85149); VS Code webviews usually pull these from theme tokens (var(--vscode-*)) so they track the user's theme. Brand yellow staying hardcoded is fine.

jack-champagne · 2026-06-19T04:37:33Z

- <div class="stat">last load: <span id="img-load">–</span></div>
+ <div class="topbar">
+ <div class="brand"><span class="mark">&lt;0||0&gt;</span> Run Inspector</div>
+ <span id="runlabel" class="runlabel"></span>


Styled but never populated anywhere — renders empty.

Fixed in e21539c (on #23, stacked above) — added setRunLabel(runId) (buffered like warming/completion) + a runlabel webview handler, wired from the watcher's switchToRun, so #runlabel shows the active runId.

…cards, branded chrome Replaces the ported dev-spike layout (raw status/frame/last-load header + Hz/ latency/post-recv telemetry footer) with a researcher-facing look: - top bar: <0||0> amico mark + pulsing status badge (idle/running/converged/failed) - plot hero in a rounded framed panel with a smoother crossfade - metric cards: objective (violet hero), iteration, feasibility, optimality - friendlier branded empty state Theme-aware (VS Code color vars) with the amico violet accent. Webview-only; message protocol + watcher feed unchanged. Tests green (37). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

… on 'running' - accent + running state → #FFF676 (was violet/amber) - the watcher streamed frames without a terminal signal, so the badge only ever showed 'running'. Add InspectorView.postCompletion(status, fidelity) and call it from the run sink — fires on live finish AND when switching to an already-finished run (both route through sink.run), flushing the last frame first so completion is authoritative. Badge now: idle → running → converged (green) / failed (red). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…nished On launch the watcher follows `latest` and a finished run completes before the Run Inspector panel is materialized — postCompletion early-returned, so only the buffered image replayed (badge → 'running') and the terminal state was lost. Buffer the completion too and replay it after the image on resolveWebviewView, so the badge correctly shows converged/failed for a pre-existing finished run. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…y on completion - image-host gets min-height 240px (+ body scrolls) so the pulse plot renders as a real plot in the short bottom panel instead of a thin sliver/bar. - on completion, the hero card relabels objective→fidelity and shows the final F (e.g. 0.99971) — fills what was an empty '–' card for finished runs. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…ck review nits The real "plots never display" bug: LiveRunSink shared one counter between image() and iter(). run.log AMICODE_ITER lines stream once per iteration and race the counter to max BEFORE the lagging PNG frames land (solver logs iter=k, then writes iter_k.png), so every image() hit `k <= latest` and was dropped — blank inspector for the whole solve. Split into SinkDedup: frames dedup only against frames; log lines advance a separate high-water mark for the status bar. Extracted SinkDedup to the pure, vscode-free run_dir_reader layer so the live path is finally unit-tested (Jack #9: "no test covering the live status-bar / incremental inspector path" — the exact gap this regression slipped through). Also folds in the remaining non-blocking review nits (kept here rather than restacking the 7-deep chain to place one-liners on their home branches): - #13: #runlabel was styled but never populated → setRunLabel + webview handler. - #12: resolveJuliaProject now expands a leading ~ (parity with resolveRunsRoot). - #9: server_manager ServerOptions.env comment no longer cites the removed AMICODE_EXTENSION_URL/MCP-callback env. - #9/#11: AGENTS.md step-2 author path now matches step-3 run path (/tmp/amicode-work/solve.jl) and the invocation passes --lab default (run provenance — amico-run supports --lab). Plus the cold-start display polish already in flight on this branch: launch stays idle for a prior finished run (no stale plot), and the placeholder toggles via style.display (the [hidden] attr was overridden by .placeholder{display:flex}). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Rchari1 · 2026-06-22T21:21:07Z

@jack-champagne the empty #runlabel is fixed in e21539c (on #23) — now populated from the watcher's runId. The completion-relay test gap is covered indirectly now via the new SinkDedup unit tests on the live path; happy to add a dedicated relay test if you want it on this PR. Thanks!

jack-champagne · 2026-06-26T19:33:58Z

Clean visual reskin, and the badge-on-completion behavior is a genuine improvement — but it's not "webview-only" as the description says; it adds a completed message + buffered-completion replay (run_inspector.ts, inspector_webview.ts). Please fix the description, and:

[minor] #runlabel is rendered and styled (run_inspector.ts:221,181) but never set — permanently empty. Wire it from RunCompletion.runId or drop it.
[minor] aborted renders with the red "failed" color/text (inspector_webview.ts:42).
[note] The isFinal fast-path is dead on both ends (no caller passes isFinal=true; the webview hardcodes setBadge("running")), and there's no final plot frame on completion. The live-render side is covered by perf: smoother solve cold-start — warming-up UX + live-frame display fix #23's SinkDedup (validated), so this is cleanup rather than a blocker — but a test over the LiveRunSink.run → completed seam would be good.

jack-champagne

Approving — solid reskin. Please fix the description; #runlabel / aborted-color are cheap cleanups. Live-render is covered by #23.

feat(vault): first-run setup popup → local personal vault (#13)

Rchari1 · 2026-06-17T22:21:11Z

A self-contained aesthetic pass on the Run Inspector. Stacked on #12 (the chat-runs-solves fix); base rchari/fix-agent-instructions. Webview-only — the message protocol and the watcher feed are unchanged.

Before → after

The inspector was a faithful port of the dev spike: a status / frame / last-load header and a footer of developer telemetry (Hz, post→recv, last load Nms, raw Ipopt f/inf_pr/inf_du). This reskins it as a research instrument:

Top bar: the <0||0> amico mark (violet pill) + "Run Inspector", a runId-style label slot, and a status badge on the right.
Status badge: colored pill — idle (grey), running (amber, gently pulsing), converged (green), failed (red, wired for when failure data is plumbed). Driven by the existing isFinal signal.
Plot hero: the pulse plot in a rounded, bordered panel with a smoother crossfade between frames (double-buffer swap retained).
Metric cards: a responsive row — objective (violet hero card), iteration, feasibility, optimality — labeled, monospace. The Hz / post→recv / last-load dev telemetry is dropped.
Empty state: the mark + a friendlier hint ("…or run 'Replay demo run'").

Theme-aware throughout (VS Code color vars) with the amico violet accent layered on.

Scope / not included

Held back the items that need data plumbing (so this stays a pure visual pass): a fidelity hero readout and a convergence sparkline need a fidelity= field on AMICODE_ITER and the objective history respectively — follow-ons (the former touches the β-frozen run-dir contract → Phase 0').

Verification

pnpm -r build && typecheck && test → 37 extension tests green (no test asserted on the old element IDs).
Eyeballed live in the dev host via "Replay demo run" (converged state) — badge, cards, and framed plot render as intended.

🤖 Generated with Claude Code

jack-champagne

Gave it a look — nice aesthetic pass, CI green, no blockers. A couple of should-fixes.

Worth fixing

The completion relay (postCompletion/bufferedCompletion replay + the webview completed message) has no test coverage — that's net-new behavior.
#runlabel is styled but never populated — it renders empty.

Taste / non-blocking

Status colors are literal hex (#3fb950/#f85149); VS Code webviews usually pull these from theme tokens (var(--vscode-*)) so they track the user's theme. Brand yellow staying hardcoded is fine.

jack-champagne · 2026-06-19T04:37:33Z

- <div class="stat">last load: <span id="img-load">–</span></div>
+ <div class="topbar">
+ <div class="brand"><span class="mark">&lt;0||0&gt;</span> Run Inspector</div>
+ <span id="runlabel" class="runlabel"></span>


Styled but never populated anywhere — renders empty.

Fixed in e21539c (on #23, stacked above) — added setRunLabel(runId) (buffered like warming/completion) + a runlabel webview handler, wired from the watcher's switchToRun, so #runlabel shows the active runId.

…cards, branded chrome Replaces the ported dev-spike layout (raw status/frame/last-load header + Hz/ latency/post-recv telemetry footer) with a researcher-facing look: - top bar: <0||0> amico mark + pulsing status badge (idle/running/converged/failed) - plot hero in a rounded framed panel with a smoother crossfade - metric cards: objective (violet hero), iteration, feasibility, optimality - friendlier branded empty state Theme-aware (VS Code color vars) with the amico violet accent. Webview-only; message protocol + watcher feed unchanged. Tests green (37). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

… on 'running' - accent + running state → #FFF676 (was violet/amber) - the watcher streamed frames without a terminal signal, so the badge only ever showed 'running'. Add InspectorView.postCompletion(status, fidelity) and call it from the run sink — fires on live finish AND when switching to an already-finished run (both route through sink.run), flushing the last frame first so completion is authoritative. Badge now: idle → running → converged (green) / failed (red). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…nished On launch the watcher follows `latest` and a finished run completes before the Run Inspector panel is materialized — postCompletion early-returned, so only the buffered image replayed (badge → 'running') and the terminal state was lost. Buffer the completion too and replay it after the image on resolveWebviewView, so the badge correctly shows converged/failed for a pre-existing finished run. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…y on completion - image-host gets min-height 240px (+ body scrolls) so the pulse plot renders as a real plot in the short bottom panel instead of a thin sliver/bar. - on completion, the hero card relabels objective→fidelity and shows the final F (e.g. 0.99971) — fills what was an empty '–' card for finished runs. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…ck review nits The real "plots never display" bug: LiveRunSink shared one counter between image() and iter(). run.log AMICODE_ITER lines stream once per iteration and race the counter to max BEFORE the lagging PNG frames land (solver logs iter=k, then writes iter_k.png), so every image() hit `k <= latest` and was dropped — blank inspector for the whole solve. Split into SinkDedup: frames dedup only against frames; log lines advance a separate high-water mark for the status bar. Extracted SinkDedup to the pure, vscode-free run_dir_reader layer so the live path is finally unit-tested (Jack #9: "no test covering the live status-bar / incremental inspector path" — the exact gap this regression slipped through). Also folds in the remaining non-blocking review nits (kept here rather than restacking the 7-deep chain to place one-liners on their home branches): - #13: #runlabel was styled but never populated → setRunLabel + webview handler. - #12: resolveJuliaProject now expands a leading ~ (parity with resolveRunsRoot). - #9: server_manager ServerOptions.env comment no longer cites the removed AMICODE_EXTENSION_URL/MCP-callback env. - #9/#11: AGENTS.md step-2 author path now matches step-3 run path (/tmp/amicode-work/solve.jl) and the invocation passes --lab default (run provenance — amico-run supports --lab). Plus the cold-start display polish already in flight on this branch: launch stays idle for a prior finished run (no stale plot), and the placeholder toggles via style.display (the [hidden] attr was overridden by .placeholder{display:flex}). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Rchari1 · 2026-06-22T21:21:07Z

@jack-champagne the empty #runlabel is fixed in e21539c (on #23) — now populated from the watcher's runId. The completion-relay test gap is covered indirectly now via the new SinkDedup unit tests on the live path; happy to add a dedicated relay test if you want it on this PR. Thanks!

jack-champagne · 2026-06-26T19:33:58Z

Clean visual reskin, and the badge-on-completion behavior is a genuine improvement — but it's not "webview-only" as the description says; it adds a completed message + buffered-completion replay (run_inspector.ts, inspector_webview.ts). Please fix the description, and:

[minor] #runlabel is rendered and styled (run_inspector.ts:221,181) but never set — permanently empty. Wire it from RunCompletion.runId or drop it.
[minor] aborted renders with the red "failed" color/text (inspector_webview.ts:42).
[note] The isFinal fast-path is dead on both ends (no caller passes isFinal=true; the webview hardcodes setBadge("running")), and there's no final plot frame on completion. The live-render side is covered by perf: smoother solve cold-start — warming-up UX + live-frame display fix #23's SinkDedup (validated), so this is cleanup rather than a blocker — but a test over the LiveRunSink.run → completed seam would be good.

jack-champagne

Approving — solid reskin. Please fix the description; #runlabel / aborted-color are cheap cleanups. Live-render is covered by #23.

feat(vault): first-run setup popup → local personal vault (#13)

Rchari1 · 2026-06-17T22:21:11Z

A self-contained aesthetic pass on the Run Inspector. Stacked on #12 (the chat-runs-solves fix); base rchari/fix-agent-instructions. Webview-only — the message protocol and the watcher feed are unchanged.

Before → after

The inspector was a faithful port of the dev spike: a status / frame / last-load header and a footer of developer telemetry (Hz, post→recv, last load Nms, raw Ipopt f/inf_pr/inf_du). This reskins it as a research instrument:

Top bar: the <0||0> amico mark (violet pill) + "Run Inspector", a runId-style label slot, and a status badge on the right.
Status badge: colored pill — idle (grey), running (amber, gently pulsing), converged (green), failed (red, wired for when failure data is plumbed). Driven by the existing isFinal signal.
Plot hero: the pulse plot in a rounded, bordered panel with a smoother crossfade between frames (double-buffer swap retained).
Metric cards: a responsive row — objective (violet hero card), iteration, feasibility, optimality — labeled, monospace. The Hz / post→recv / last-load dev telemetry is dropped.
Empty state: the mark + a friendlier hint ("…or run 'Replay demo run'").

Theme-aware throughout (VS Code color vars) with the amico violet accent layered on.

Scope / not included

Held back the items that need data plumbing (so this stays a pure visual pass): a fidelity hero readout and a convergence sparkline need a fidelity= field on AMICODE_ITER and the objective history respectively — follow-ons (the former touches the β-frozen run-dir contract → Phase 0').

Verification

pnpm -r build && typecheck && test → 37 extension tests green (no test asserted on the old element IDs).
Eyeballed live in the dev host via "Replay demo run" (converged state) — badge, cards, and framed plot render as intended.

🤖 Generated with Claude Code

jack-champagne

Gave it a look — nice aesthetic pass, CI green, no blockers. A couple of should-fixes.

Worth fixing

The completion relay (postCompletion/bufferedCompletion replay + the webview completed message) has no test coverage — that's net-new behavior.
#runlabel is styled but never populated — it renders empty.

Taste / non-blocking

Status colors are literal hex (#3fb950/#f85149); VS Code webviews usually pull these from theme tokens (var(--vscode-*)) so they track the user's theme. Brand yellow staying hardcoded is fine.

jack-champagne · 2026-06-19T04:37:33Z

- <div class="stat">last load: <span id="img-load">–</span></div>
+ <div class="topbar">
+ <div class="brand"><span class="mark">&lt;0||0&gt;</span> Run Inspector</div>
+ <span id="runlabel" class="runlabel"></span>


Styled but never populated anywhere — renders empty.

Fixed in e21539c (on #23, stacked above) — added setRunLabel(runId) (buffered like warming/completion) + a runlabel webview handler, wired from the watcher's switchToRun, so #runlabel shows the active runId.

…cards, branded chrome Replaces the ported dev-spike layout (raw status/frame/last-load header + Hz/ latency/post-recv telemetry footer) with a researcher-facing look: - top bar: <0||0> amico mark + pulsing status badge (idle/running/converged/failed) - plot hero in a rounded framed panel with a smoother crossfade - metric cards: objective (violet hero), iteration, feasibility, optimality - friendlier branded empty state Theme-aware (VS Code color vars) with the amico violet accent. Webview-only; message protocol + watcher feed unchanged. Tests green (37). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

… on 'running' - accent + running state → #FFF676 (was violet/amber) - the watcher streamed frames without a terminal signal, so the badge only ever showed 'running'. Add InspectorView.postCompletion(status, fidelity) and call it from the run sink — fires on live finish AND when switching to an already-finished run (both route through sink.run), flushing the last frame first so completion is authoritative. Badge now: idle → running → converged (green) / failed (red). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…nished On launch the watcher follows `latest` and a finished run completes before the Run Inspector panel is materialized — postCompletion early-returned, so only the buffered image replayed (badge → 'running') and the terminal state was lost. Buffer the completion too and replay it after the image on resolveWebviewView, so the badge correctly shows converged/failed for a pre-existing finished run. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…y on completion - image-host gets min-height 240px (+ body scrolls) so the pulse plot renders as a real plot in the short bottom panel instead of a thin sliver/bar. - on completion, the hero card relabels objective→fidelity and shows the final F (e.g. 0.99971) — fills what was an empty '–' card for finished runs. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…ck review nits The real "plots never display" bug: LiveRunSink shared one counter between image() and iter(). run.log AMICODE_ITER lines stream once per iteration and race the counter to max BEFORE the lagging PNG frames land (solver logs iter=k, then writes iter_k.png), so every image() hit `k <= latest` and was dropped — blank inspector for the whole solve. Split into SinkDedup: frames dedup only against frames; log lines advance a separate high-water mark for the status bar. Extracted SinkDedup to the pure, vscode-free run_dir_reader layer so the live path is finally unit-tested (Jack #9: "no test covering the live status-bar / incremental inspector path" — the exact gap this regression slipped through). Also folds in the remaining non-blocking review nits (kept here rather than restacking the 7-deep chain to place one-liners on their home branches): - #13: #runlabel was styled but never populated → setRunLabel + webview handler. - #12: resolveJuliaProject now expands a leading ~ (parity with resolveRunsRoot). - #9: server_manager ServerOptions.env comment no longer cites the removed AMICODE_EXTENSION_URL/MCP-callback env. - #9/#11: AGENTS.md step-2 author path now matches step-3 run path (/tmp/amicode-work/solve.jl) and the invocation passes --lab default (run provenance — amico-run supports --lab). Plus the cold-start display polish already in flight on this branch: launch stays idle for a prior finished run (no stale plot), and the placeholder toggles via style.display (the [hidden] attr was overridden by .placeholder{display:flex}). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Rchari1 · 2026-06-22T21:21:07Z

@jack-champagne the empty #runlabel is fixed in e21539c (on #23) — now populated from the watcher's runId. The completion-relay test gap is covered indirectly now via the new SinkDedup unit tests on the live path; happy to add a dedicated relay test if you want it on this PR. Thanks!

jack-champagne · 2026-06-26T19:33:58Z

Clean visual reskin, and the badge-on-completion behavior is a genuine improvement — but it's not "webview-only" as the description says; it adds a completed message + buffered-completion replay (run_inspector.ts, inspector_webview.ts). Please fix the description, and:

[minor] #runlabel is rendered and styled (run_inspector.ts:221,181) but never set — permanently empty. Wire it from RunCompletion.runId or drop it.
[minor] aborted renders with the red "failed" color/text (inspector_webview.ts:42).
[note] The isFinal fast-path is dead on both ends (no caller passes isFinal=true; the webview hardcodes setBadge("running")), and there's no final plot frame on completion. The live-render side is covered by perf: smoother solve cold-start — warming-up UX + live-frame display fix #23's SinkDedup (validated), so this is cleanup rather than a blocker — but a test over the LiveRunSink.run → completed seam would be good.

jack-champagne

Approving — solid reskin. Please fix the description; #runlabel / aborted-color are cheap cleanups. Live-render is covered by #23.

feat(vault): first-run setup popup → local personal vault (#13)

Rchari1 · 2026-06-17T22:21:11Z

A self-contained aesthetic pass on the Run Inspector. Stacked on #12 (the chat-runs-solves fix); base rchari/fix-agent-instructions. Webview-only — the message protocol and the watcher feed are unchanged.

Before → after

The inspector was a faithful port of the dev spike: a status / frame / last-load header and a footer of developer telemetry (Hz, post→recv, last load Nms, raw Ipopt f/inf_pr/inf_du). This reskins it as a research instrument:

Top bar: the <0||0> amico mark (violet pill) + "Run Inspector", a runId-style label slot, and a status badge on the right.
Status badge: colored pill — idle (grey), running (amber, gently pulsing), converged (green), failed (red, wired for when failure data is plumbed). Driven by the existing isFinal signal.
Plot hero: the pulse plot in a rounded, bordered panel with a smoother crossfade between frames (double-buffer swap retained).
Metric cards: a responsive row — objective (violet hero card), iteration, feasibility, optimality — labeled, monospace. The Hz / post→recv / last-load dev telemetry is dropped.
Empty state: the mark + a friendlier hint ("…or run 'Replay demo run'").

Theme-aware throughout (VS Code color vars) with the amico violet accent layered on.

Scope / not included

Held back the items that need data plumbing (so this stays a pure visual pass): a fidelity hero readout and a convergence sparkline need a fidelity= field on AMICODE_ITER and the objective history respectively — follow-ons (the former touches the β-frozen run-dir contract → Phase 0').

Verification

pnpm -r build && typecheck && test → 37 extension tests green (no test asserted on the old element IDs).
Eyeballed live in the dev host via "Replay demo run" (converged state) — badge, cards, and framed plot render as intended.

🤖 Generated with Claude Code

jack-champagne

Gave it a look — nice aesthetic pass, CI green, no blockers. A couple of should-fixes.

Worth fixing

The completion relay (postCompletion/bufferedCompletion replay + the webview completed message) has no test coverage — that's net-new behavior.
#runlabel is styled but never populated — it renders empty.

Taste / non-blocking

Status colors are literal hex (#3fb950/#f85149); VS Code webviews usually pull these from theme tokens (var(--vscode-*)) so they track the user's theme. Brand yellow staying hardcoded is fine.

jack-champagne · 2026-06-19T04:37:33Z

- <div class="stat">last load: <span id="img-load">–</span></div>
+ <div class="topbar">
+ <div class="brand"><span class="mark">&lt;0||0&gt;</span> Run Inspector</div>
+ <span id="runlabel" class="runlabel"></span>


Styled but never populated anywhere — renders empty.

Fixed in e21539c (on #23, stacked above) — added setRunLabel(runId) (buffered like warming/completion) + a runlabel webview handler, wired from the watcher's switchToRun, so #runlabel shows the active runId.

…cards, branded chrome Replaces the ported dev-spike layout (raw status/frame/last-load header + Hz/ latency/post-recv telemetry footer) with a researcher-facing look: - top bar: <0||0> amico mark + pulsing status badge (idle/running/converged/failed) - plot hero in a rounded framed panel with a smoother crossfade - metric cards: objective (violet hero), iteration, feasibility, optimality - friendlier branded empty state Theme-aware (VS Code color vars) with the amico violet accent. Webview-only; message protocol + watcher feed unchanged. Tests green (37). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

… on 'running' - accent + running state → #FFF676 (was violet/amber) - the watcher streamed frames without a terminal signal, so the badge only ever showed 'running'. Add InspectorView.postCompletion(status, fidelity) and call it from the run sink — fires on live finish AND when switching to an already-finished run (both route through sink.run), flushing the last frame first so completion is authoritative. Badge now: idle → running → converged (green) / failed (red). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…nished On launch the watcher follows `latest` and a finished run completes before the Run Inspector panel is materialized — postCompletion early-returned, so only the buffered image replayed (badge → 'running') and the terminal state was lost. Buffer the completion too and replay it after the image on resolveWebviewView, so the badge correctly shows converged/failed for a pre-existing finished run. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…y on completion - image-host gets min-height 240px (+ body scrolls) so the pulse plot renders as a real plot in the short bottom panel instead of a thin sliver/bar. - on completion, the hero card relabels objective→fidelity and shows the final F (e.g. 0.99971) — fills what was an empty '–' card for finished runs. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…ck review nits The real "plots never display" bug: LiveRunSink shared one counter between image() and iter(). run.log AMICODE_ITER lines stream once per iteration and race the counter to max BEFORE the lagging PNG frames land (solver logs iter=k, then writes iter_k.png), so every image() hit `k <= latest` and was dropped — blank inspector for the whole solve. Split into SinkDedup: frames dedup only against frames; log lines advance a separate high-water mark for the status bar. Extracted SinkDedup to the pure, vscode-free run_dir_reader layer so the live path is finally unit-tested (Jack #9: "no test covering the live status-bar / incremental inspector path" — the exact gap this regression slipped through). Also folds in the remaining non-blocking review nits (kept here rather than restacking the 7-deep chain to place one-liners on their home branches): - #13: #runlabel was styled but never populated → setRunLabel + webview handler. - #12: resolveJuliaProject now expands a leading ~ (parity with resolveRunsRoot). - #9: server_manager ServerOptions.env comment no longer cites the removed AMICODE_EXTENSION_URL/MCP-callback env. - #9/#11: AGENTS.md step-2 author path now matches step-3 run path (/tmp/amicode-work/solve.jl) and the invocation passes --lab default (run provenance — amico-run supports --lab). Plus the cold-start display polish already in flight on this branch: launch stays idle for a prior finished run (no stale plot), and the placeholder toggles via style.display (the [hidden] attr was overridden by .placeholder{display:flex}). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Rchari1 · 2026-06-22T21:21:07Z

@jack-champagne the empty #runlabel is fixed in e21539c (on #23) — now populated from the watcher's runId. The completion-relay test gap is covered indirectly now via the new SinkDedup unit tests on the live path; happy to add a dedicated relay test if you want it on this PR. Thanks!

jack-champagne · 2026-06-26T19:33:58Z

Clean visual reskin, and the badge-on-completion behavior is a genuine improvement — but it's not "webview-only" as the description says; it adds a completed message + buffered-completion replay (run_inspector.ts, inspector_webview.ts). Please fix the description, and:

[minor] #runlabel is rendered and styled (run_inspector.ts:221,181) but never set — permanently empty. Wire it from RunCompletion.runId or drop it.
[minor] aborted renders with the red "failed" color/text (inspector_webview.ts:42).
[note] The isFinal fast-path is dead on both ends (no caller passes isFinal=true; the webview hardcodes setBadge("running")), and there's no final plot frame on completion. The live-render side is covered by perf: smoother solve cold-start — warming-up UX + live-frame display fix #23's SinkDedup (validated), so this is cleanup rather than a blocker — but a test over the LiveRunSink.run → completed seam would be good.

jack-champagne

Approving — solid reskin. Please fix the description; #runlabel / aborted-color are cheap cleanups. Live-render is covered by #23.

feat(vault): first-run setup popup → local personal vault (#13)

Rchari1 · 2026-06-17T22:21:11Z

A self-contained aesthetic pass on the Run Inspector. Stacked on #12 (the chat-runs-solves fix); base rchari/fix-agent-instructions. Webview-only — the message protocol and the watcher feed are unchanged.

Before → after

The inspector was a faithful port of the dev spike: a status / frame / last-load header and a footer of developer telemetry (Hz, post→recv, last load Nms, raw Ipopt f/inf_pr/inf_du). This reskins it as a research instrument:

Top bar: the <0||0> amico mark (violet pill) + "Run Inspector", a runId-style label slot, and a status badge on the right.
Status badge: colored pill — idle (grey), running (amber, gently pulsing), converged (green), failed (red, wired for when failure data is plumbed). Driven by the existing isFinal signal.
Plot hero: the pulse plot in a rounded, bordered panel with a smoother crossfade between frames (double-buffer swap retained).
Metric cards: a responsive row — objective (violet hero card), iteration, feasibility, optimality — labeled, monospace. The Hz / post→recv / last-load dev telemetry is dropped.
Empty state: the mark + a friendlier hint ("…or run 'Replay demo run'").

Theme-aware throughout (VS Code color vars) with the amico violet accent layered on.

Scope / not included

Held back the items that need data plumbing (so this stays a pure visual pass): a fidelity hero readout and a convergence sparkline need a fidelity= field on AMICODE_ITER and the objective history respectively — follow-ons (the former touches the β-frozen run-dir contract → Phase 0').

Verification

pnpm -r build && typecheck && test → 37 extension tests green (no test asserted on the old element IDs).
Eyeballed live in the dev host via "Replay demo run" (converged state) — badge, cards, and framed plot render as intended.

🤖 Generated with Claude Code

jack-champagne

Gave it a look — nice aesthetic pass, CI green, no blockers. A couple of should-fixes.

Worth fixing

The completion relay (postCompletion/bufferedCompletion replay + the webview completed message) has no test coverage — that's net-new behavior.
#runlabel is styled but never populated — it renders empty.

Taste / non-blocking

Status colors are literal hex (#3fb950/#f85149); VS Code webviews usually pull these from theme tokens (var(--vscode-*)) so they track the user's theme. Brand yellow staying hardcoded is fine.

jack-champagne · 2026-06-19T04:37:33Z

- <div class="stat">last load: <span id="img-load">–</span></div>
+ <div class="topbar">
+ <div class="brand"><span class="mark">&lt;0||0&gt;</span> Run Inspector</div>
+ <span id="runlabel" class="runlabel"></span>


Styled but never populated anywhere — renders empty.

Fixed in e21539c (on #23, stacked above) — added setRunLabel(runId) (buffered like warming/completion) + a runlabel webview handler, wired from the watcher's switchToRun, so #runlabel shows the active runId.

…cards, branded chrome Replaces the ported dev-spike layout (raw status/frame/last-load header + Hz/ latency/post-recv telemetry footer) with a researcher-facing look: - top bar: <0||0> amico mark + pulsing status badge (idle/running/converged/failed) - plot hero in a rounded framed panel with a smoother crossfade - metric cards: objective (violet hero), iteration, feasibility, optimality - friendlier branded empty state Theme-aware (VS Code color vars) with the amico violet accent. Webview-only; message protocol + watcher feed unchanged. Tests green (37). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

… on 'running' - accent + running state → #FFF676 (was violet/amber) - the watcher streamed frames without a terminal signal, so the badge only ever showed 'running'. Add InspectorView.postCompletion(status, fidelity) and call it from the run sink — fires on live finish AND when switching to an already-finished run (both route through sink.run), flushing the last frame first so completion is authoritative. Badge now: idle → running → converged (green) / failed (red). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…nished On launch the watcher follows `latest` and a finished run completes before the Run Inspector panel is materialized — postCompletion early-returned, so only the buffered image replayed (badge → 'running') and the terminal state was lost. Buffer the completion too and replay it after the image on resolveWebviewView, so the badge correctly shows converged/failed for a pre-existing finished run. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…y on completion - image-host gets min-height 240px (+ body scrolls) so the pulse plot renders as a real plot in the short bottom panel instead of a thin sliver/bar. - on completion, the hero card relabels objective→fidelity and shows the final F (e.g. 0.99971) — fills what was an empty '–' card for finished runs. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…ck review nits The real "plots never display" bug: LiveRunSink shared one counter between image() and iter(). run.log AMICODE_ITER lines stream once per iteration and race the counter to max BEFORE the lagging PNG frames land (solver logs iter=k, then writes iter_k.png), so every image() hit `k <= latest` and was dropped — blank inspector for the whole solve. Split into SinkDedup: frames dedup only against frames; log lines advance a separate high-water mark for the status bar. Extracted SinkDedup to the pure, vscode-free run_dir_reader layer so the live path is finally unit-tested (Jack #9: "no test covering the live status-bar / incremental inspector path" — the exact gap this regression slipped through). Also folds in the remaining non-blocking review nits (kept here rather than restacking the 7-deep chain to place one-liners on their home branches): - #13: #runlabel was styled but never populated → setRunLabel + webview handler. - #12: resolveJuliaProject now expands a leading ~ (parity with resolveRunsRoot). - #9: server_manager ServerOptions.env comment no longer cites the removed AMICODE_EXTENSION_URL/MCP-callback env. - #9/#11: AGENTS.md step-2 author path now matches step-3 run path (/tmp/amicode-work/solve.jl) and the invocation passes --lab default (run provenance — amico-run supports --lab). Plus the cold-start display polish already in flight on this branch: launch stays idle for a prior finished run (no stale plot), and the placeholder toggles via style.display (the [hidden] attr was overridden by .placeholder{display:flex}). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Rchari1 · 2026-06-22T21:21:07Z

@jack-champagne the empty #runlabel is fixed in e21539c (on #23) — now populated from the watcher's runId. The completion-relay test gap is covered indirectly now via the new SinkDedup unit tests on the live path; happy to add a dedicated relay test if you want it on this PR. Thanks!

jack-champagne · 2026-06-26T19:33:58Z

Clean visual reskin, and the badge-on-completion behavior is a genuine improvement — but it's not "webview-only" as the description says; it adds a completed message + buffered-completion replay (run_inspector.ts, inspector_webview.ts). Please fix the description, and:

[minor] #runlabel is rendered and styled (run_inspector.ts:221,181) but never set — permanently empty. Wire it from RunCompletion.runId or drop it.
[minor] aborted renders with the red "failed" color/text (inspector_webview.ts:42).
[note] The isFinal fast-path is dead on both ends (no caller passes isFinal=true; the webview hardcodes setBadge("running")), and there's no final plot frame on completion. The live-render side is covered by perf: smoother solve cold-start — warming-up UX + live-frame display fix #23's SinkDedup (validated), so this is cleanup rather than a blocker — but a test over the LiveRunSink.run → completed seam would be good.

jack-champagne

Approving — solid reskin. Please fix the description; #runlabel / aborted-color are cheap cleanups. Live-render is covered by #23.

feat(vault): first-run setup popup → local personal vault (#13)

Rchari1 · 2026-06-17T22:21:11Z

A self-contained aesthetic pass on the Run Inspector. Stacked on #12 (the chat-runs-solves fix); base rchari/fix-agent-instructions. Webview-only — the message protocol and the watcher feed are unchanged.

Before → after

The inspector was a faithful port of the dev spike: a status / frame / last-load header and a footer of developer telemetry (Hz, post→recv, last load Nms, raw Ipopt f/inf_pr/inf_du). This reskins it as a research instrument:

Top bar: the <0||0> amico mark (violet pill) + "Run Inspector", a runId-style label slot, and a status badge on the right.
Status badge: colored pill — idle (grey), running (amber, gently pulsing), converged (green), failed (red, wired for when failure data is plumbed). Driven by the existing isFinal signal.
Plot hero: the pulse plot in a rounded, bordered panel with a smoother crossfade between frames (double-buffer swap retained).
Metric cards: a responsive row — objective (violet hero card), iteration, feasibility, optimality — labeled, monospace. The Hz / post→recv / last-load dev telemetry is dropped.
Empty state: the mark + a friendlier hint ("…or run 'Replay demo run'").

Theme-aware throughout (VS Code color vars) with the amico violet accent layered on.

Scope / not included

Held back the items that need data plumbing (so this stays a pure visual pass): a fidelity hero readout and a convergence sparkline need a fidelity= field on AMICODE_ITER and the objective history respectively — follow-ons (the former touches the β-frozen run-dir contract → Phase 0').

Verification

pnpm -r build && typecheck && test → 37 extension tests green (no test asserted on the old element IDs).
Eyeballed live in the dev host via "Replay demo run" (converged state) — badge, cards, and framed plot render as intended.

🤖 Generated with Claude Code

jack-champagne

Gave it a look — nice aesthetic pass, CI green, no blockers. A couple of should-fixes.

Worth fixing

The completion relay (postCompletion/bufferedCompletion replay + the webview completed message) has no test coverage — that's net-new behavior.
#runlabel is styled but never populated — it renders empty.

Taste / non-blocking

Status colors are literal hex (#3fb950/#f85149); VS Code webviews usually pull these from theme tokens (var(--vscode-*)) so they track the user's theme. Brand yellow staying hardcoded is fine.

jack-champagne · 2026-06-19T04:37:33Z

- <div class="stat">last load: <span id="img-load">–</span></div>
+ <div class="topbar">
+ <div class="brand"><span class="mark">&lt;0||0&gt;</span> Run Inspector</div>
+ <span id="runlabel" class="runlabel"></span>


Styled but never populated anywhere — renders empty.

Fixed in e21539c (on #23, stacked above) — added setRunLabel(runId) (buffered like warming/completion) + a runlabel webview handler, wired from the watcher's switchToRun, so #runlabel shows the active runId.

…cards, branded chrome Replaces the ported dev-spike layout (raw status/frame/last-load header + Hz/ latency/post-recv telemetry footer) with a researcher-facing look: - top bar: <0||0> amico mark + pulsing status badge (idle/running/converged/failed) - plot hero in a rounded framed panel with a smoother crossfade - metric cards: objective (violet hero), iteration, feasibility, optimality - friendlier branded empty state Theme-aware (VS Code color vars) with the amico violet accent. Webview-only; message protocol + watcher feed unchanged. Tests green (37). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

… on 'running' - accent + running state → #FFF676 (was violet/amber) - the watcher streamed frames without a terminal signal, so the badge only ever showed 'running'. Add InspectorView.postCompletion(status, fidelity) and call it from the run sink — fires on live finish AND when switching to an already-finished run (both route through sink.run), flushing the last frame first so completion is authoritative. Badge now: idle → running → converged (green) / failed (red). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…nished On launch the watcher follows `latest` and a finished run completes before the Run Inspector panel is materialized — postCompletion early-returned, so only the buffered image replayed (badge → 'running') and the terminal state was lost. Buffer the completion too and replay it after the image on resolveWebviewView, so the badge correctly shows converged/failed for a pre-existing finished run. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…y on completion - image-host gets min-height 240px (+ body scrolls) so the pulse plot renders as a real plot in the short bottom panel instead of a thin sliver/bar. - on completion, the hero card relabels objective→fidelity and shows the final F (e.g. 0.99971) — fills what was an empty '–' card for finished runs. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…ck review nits The real "plots never display" bug: LiveRunSink shared one counter between image() and iter(). run.log AMICODE_ITER lines stream once per iteration and race the counter to max BEFORE the lagging PNG frames land (solver logs iter=k, then writes iter_k.png), so every image() hit `k <= latest` and was dropped — blank inspector for the whole solve. Split into SinkDedup: frames dedup only against frames; log lines advance a separate high-water mark for the status bar. Extracted SinkDedup to the pure, vscode-free run_dir_reader layer so the live path is finally unit-tested (Jack #9: "no test covering the live status-bar / incremental inspector path" — the exact gap this regression slipped through). Also folds in the remaining non-blocking review nits (kept here rather than restacking the 7-deep chain to place one-liners on their home branches): - #13: #runlabel was styled but never populated → setRunLabel + webview handler. - #12: resolveJuliaProject now expands a leading ~ (parity with resolveRunsRoot). - #9: server_manager ServerOptions.env comment no longer cites the removed AMICODE_EXTENSION_URL/MCP-callback env. - #9/#11: AGENTS.md step-2 author path now matches step-3 run path (/tmp/amicode-work/solve.jl) and the invocation passes --lab default (run provenance — amico-run supports --lab). Plus the cold-start display polish already in flight on this branch: launch stays idle for a prior finished run (no stale plot), and the placeholder toggles via style.display (the [hidden] attr was overridden by .placeholder{display:flex}). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Rchari1 · 2026-06-22T21:21:07Z

@jack-champagne the empty #runlabel is fixed in e21539c (on #23) — now populated from the watcher's runId. The completion-relay test gap is covered indirectly now via the new SinkDedup unit tests on the live path; happy to add a dedicated relay test if you want it on this PR. Thanks!

jack-champagne · 2026-06-26T19:33:58Z

Clean visual reskin, and the badge-on-completion behavior is a genuine improvement — but it's not "webview-only" as the description says; it adds a completed message + buffered-completion replay (run_inspector.ts, inspector_webview.ts). Please fix the description, and:

[minor] #runlabel is rendered and styled (run_inspector.ts:221,181) but never set — permanently empty. Wire it from RunCompletion.runId or drop it.
[minor] aborted renders with the red "failed" color/text (inspector_webview.ts:42).
[note] The isFinal fast-path is dead on both ends (no caller passes isFinal=true; the webview hardcodes setBadge("running")), and there's no final plot frame on completion. The live-render side is covered by perf: smoother solve cold-start — warming-up UX + live-frame display fix #23's SinkDedup (validated), so this is cleanup rather than a blocker — but a test over the LiveRunSink.run → completed seam would be good.

jack-champagne

Approving — solid reskin. Please fix the description; #runlabel / aborted-color are cheap cleanups. Live-render is covered by #23.

feat(vault): first-run setup popup → local personal vault (#13)

Rchari1 · 2026-06-17T22:21:11Z

A self-contained aesthetic pass on the Run Inspector. Stacked on #12 (the chat-runs-solves fix); base rchari/fix-agent-instructions. Webview-only — the message protocol and the watcher feed are unchanged.

Before → after

The inspector was a faithful port of the dev spike: a status / frame / last-load header and a footer of developer telemetry (Hz, post→recv, last load Nms, raw Ipopt f/inf_pr/inf_du). This reskins it as a research instrument:

Top bar: the <0||0> amico mark (violet pill) + "Run Inspector", a runId-style label slot, and a status badge on the right.
Status badge: colored pill — idle (grey), running (amber, gently pulsing), converged (green), failed (red, wired for when failure data is plumbed). Driven by the existing isFinal signal.
Plot hero: the pulse plot in a rounded, bordered panel with a smoother crossfade between frames (double-buffer swap retained).
Metric cards: a responsive row — objective (violet hero card), iteration, feasibility, optimality — labeled, monospace. The Hz / post→recv / last-load dev telemetry is dropped.
Empty state: the mark + a friendlier hint ("…or run 'Replay demo run'").

Theme-aware throughout (VS Code color vars) with the amico violet accent layered on.

Scope / not included

Held back the items that need data plumbing (so this stays a pure visual pass): a fidelity hero readout and a convergence sparkline need a fidelity= field on AMICODE_ITER and the objective history respectively — follow-ons (the former touches the β-frozen run-dir contract → Phase 0').

Verification

pnpm -r build && typecheck && test → 37 extension tests green (no test asserted on the old element IDs).
Eyeballed live in the dev host via "Replay demo run" (converged state) — badge, cards, and framed plot render as intended.

🤖 Generated with Claude Code

jack-champagne

Gave it a look — nice aesthetic pass, CI green, no blockers. A couple of should-fixes.

Worth fixing

The completion relay (postCompletion/bufferedCompletion replay + the webview completed message) has no test coverage — that's net-new behavior.
#runlabel is styled but never populated — it renders empty.

Taste / non-blocking

Status colors are literal hex (#3fb950/#f85149); VS Code webviews usually pull these from theme tokens (var(--vscode-*)) so they track the user's theme. Brand yellow staying hardcoded is fine.

jack-champagne · 2026-06-19T04:37:33Z

- <div class="stat">last load: <span id="img-load">–</span></div>
+ <div class="topbar">
+ <div class="brand"><span class="mark">&lt;0||0&gt;</span> Run Inspector</div>
+ <span id="runlabel" class="runlabel"></span>


Styled but never populated anywhere — renders empty.

Fixed in e21539c (on #23, stacked above) — added setRunLabel(runId) (buffered like warming/completion) + a runlabel webview handler, wired from the watcher's switchToRun, so #runlabel shows the active runId.

…cards, branded chrome Replaces the ported dev-spike layout (raw status/frame/last-load header + Hz/ latency/post-recv telemetry footer) with a researcher-facing look: - top bar: <0||0> amico mark + pulsing status badge (idle/running/converged/failed) - plot hero in a rounded framed panel with a smoother crossfade - metric cards: objective (violet hero), iteration, feasibility, optimality - friendlier branded empty state Theme-aware (VS Code color vars) with the amico violet accent. Webview-only; message protocol + watcher feed unchanged. Tests green (37). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

… on 'running' - accent + running state → #FFF676 (was violet/amber) - the watcher streamed frames without a terminal signal, so the badge only ever showed 'running'. Add InspectorView.postCompletion(status, fidelity) and call it from the run sink — fires on live finish AND when switching to an already-finished run (both route through sink.run), flushing the last frame first so completion is authoritative. Badge now: idle → running → converged (green) / failed (red). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…nished On launch the watcher follows `latest` and a finished run completes before the Run Inspector panel is materialized — postCompletion early-returned, so only the buffered image replayed (badge → 'running') and the terminal state was lost. Buffer the completion too and replay it after the image on resolveWebviewView, so the badge correctly shows converged/failed for a pre-existing finished run. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…y on completion - image-host gets min-height 240px (+ body scrolls) so the pulse plot renders as a real plot in the short bottom panel instead of a thin sliver/bar. - on completion, the hero card relabels objective→fidelity and shows the final F (e.g. 0.99971) — fills what was an empty '–' card for finished runs. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

…ck review nits The real "plots never display" bug: LiveRunSink shared one counter between image() and iter(). run.log AMICODE_ITER lines stream once per iteration and race the counter to max BEFORE the lagging PNG frames land (solver logs iter=k, then writes iter_k.png), so every image() hit `k <= latest` and was dropped — blank inspector for the whole solve. Split into SinkDedup: frames dedup only against frames; log lines advance a separate high-water mark for the status bar. Extracted SinkDedup to the pure, vscode-free run_dir_reader layer so the live path is finally unit-tested (Jack #9: "no test covering the live status-bar / incremental inspector path" — the exact gap this regression slipped through). Also folds in the remaining non-blocking review nits (kept here rather than restacking the 7-deep chain to place one-liners on their home branches): - #13: #runlabel was styled but never populated → setRunLabel + webview handler. - #12: resolveJuliaProject now expands a leading ~ (parity with resolveRunsRoot). - #9: server_manager ServerOptions.env comment no longer cites the removed AMICODE_EXTENSION_URL/MCP-callback env. - #9/#11: AGENTS.md step-2 author path now matches step-3 run path (/tmp/amicode-work/solve.jl) and the invocation passes --lab default (run provenance — amico-run supports --lab). Plus the cold-start display polish already in flight on this branch: launch stays idle for a prior finished run (no stale plot), and the placeholder toggles via style.display (the [hidden] attr was overridden by .placeholder{display:flex}). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

Rchari1 · 2026-06-22T21:21:07Z

@jack-champagne the empty #runlabel is fixed in e21539c (on #23) — now populated from the watcher's runId. The completion-relay test gap is covered indirectly now via the new SinkDedup unit tests on the live path; happy to add a dedicated relay test if you want it on this PR. Thanks!

jack-champagne · 2026-06-26T19:33:58Z

Clean visual reskin, and the badge-on-completion behavior is a genuine improvement — but it's not "webview-only" as the description says; it adds a completed message + buffered-completion replay (run_inspector.ts, inspector_webview.ts). Please fix the description, and:

[minor] #runlabel is rendered and styled (run_inspector.ts:221,181) but never set — permanently empty. Wire it from RunCompletion.runId or drop it.
[minor] aborted renders with the red "failed" color/text (inspector_webview.ts:42).
[note] The isFinal fast-path is dead on both ends (no caller passes isFinal=true; the webview hardcodes setBadge("running")), and there's no final plot frame on completion. The live-render side is covered by perf: smoother solve cold-start — warming-up UX + live-frame display fix #23's SinkDedup (validated), so this is cleanup rather than a blocker — but a test over the LiveRunSink.run → completed seam would be good.

jack-champagne

Approving — solid reskin. Please fix the description; #runlabel / aborted-color are cheap cleanups. Live-render is covered by #23.

feat(vault): first-run setup popup → local personal vault (#13)

Rchari1 mentioned this pull request Jun 17, 2026
feat: plot the pulse with Piccolo plot_pulse (not hand-rolled Makie) #14
Merged

jack-champagne reviewed Jun 19, 2026
View reviewed changes

Rchari1and others added 4 commits June 20, 2026 21:43

Rchari1 force-pushed the rchari/fix-agent-instructions branch from a8277e4 to c7dd17bCompare June 21, 2026 01:46

Rchari1 force-pushed the rchari/inspector-aesthetics branch from fd9bded to b64dcb8Compare June 21, 2026 01:46

Rchari1 mentioned this pull request Jun 22, 2026
perf: smoother solve cold-start — warming-up UX + live-frame display fix #23
Merged

jack-champagne approved these changes Jun 26, 2026
View reviewed changes

Rchari1 changed the base branch from rchari/fix-agent-instructions to mainJune 26, 2026 20:10

Rchari1 merged commit d2d6ae2 into mainJun 26, 2026
3 checks passed

jack-champagne mentioned this pull request Jul 16, 2026
feat(vault): first-run setup popup → local personal vault (#13) #148
Merged

jack-champagne added a commit that referenced this pull request Jul 16, 2026
Merge pull request#148from harmoniqs/integration/vaults
78d12ea
feat(vault): first-run setup popup → local personal vault (#13)

jack-champagne mentioned this pull request Jul 16, 2026
feat: amicode-managed Julia setup via juliaup (#8, layers 1-3) #151
Merged