Pages
Control Expand Row — Two Sections

Source: qdash /controls — c64: the control row's expand card as two stacked UxTableListExpandSection cards (the /pages/inventory-agent-expand-v2 idiom). Top section reads (name · active findings · framework coverage), bottom section edits (condition · decisions · actions). Three variations, each auto-opening the egress control for comparison.

V1 — Header + workbench

big-name header with stat wells · edit surface in three columns
control
category
severity
matches
Non-allowlisted egress
egress: allowlisted = false
Network egress
medium
83 →
Sensitive file access
file: severity in (high, medium)
Data access
high
0
Sandbox disabled
run: sandbox_mode contains "danger"
Agent autonomy
medium
0

V2 — Findings-forward + sentence flow

top section shows the actual findings · edit surface as labeled sentence rows
control
category
severity
matches
Non-allowlisted egress
egress: allowlisted = false
Network egress
medium
83 →
Sensitive file access
file: severity in (high, medium)
Data access
high
0
Sandbox disabled
run: sandbox_mode contains "danger"
Agent autonomy
medium
0

V3 — Compact split

one-line identity strip · dense two-column edit
control
category
severity
matches
Non-allowlisted egress
egress: allowlisted = false
Network egress
medium
83 →
Sensitive file access
file: severity in (high, medium)
Data access
high
0
Sandbox disabled
run: sandbox_mode contains "danger"
Agent autonomy
medium
0

V4 — V1 header + one-line edit

V1's read section · V3's edit compressed to a single flex row, condition fills · save beneath
control
category
severity
matches
Non-allowlisted egress
egress: allowlisted = false
Network egress
medium
83 →
Sensitive file access
file: severity in (high, medium)
Data access
high
0
Sandbox disabled
run: sandbox_mode contains "danger"
Agent autonomy
medium
0

V5 — Three sections, full option sets

V4's header + condition · decisions/actions in a third section · every GovernSheet option, severity clickable
control
category
severity
matches
Non-allowlisted egress
egress: allowlisted = false
Network egress
medium
83 →
Sensitive file access
file: severity in (high, medium)
Data access
high
0
Sandbox disabled
run: sandbox_mode contains "danger"
Agent autonomy
medium
0

c64 — two-section expand card on the controls table, five variations (egress auto-opens in each). Section 1 = read (name · active findings · framework coverage), section 2 = edit (condition · decisions · actions), composed from UxTableListExpandSection stacks (well tint through the gap, close X on section 1 only) + the sentence family for editable pills. V1: big-name header with divided stat wells, three-column workbench. V2: the findings themselves (real egress snapshot rows) in section 1; section 2 as SentenceRow-labeled sentence flow with pill-toggle decisions. V3: one-line identity strip, dense two-column edit. Findings data = live /api/security snapshot (2026-08-10); decisions vary per control signal as in the real vocab. V4: V1's read header over V3's edit compressed to one flex row — condition (flex-1, no mono summary line, UxAddBtn for AND clauses) · decisions · actions — with the save button on its own rule-separated line beneath, and wider spacing (space-y-8) between the two sections. V5: three sections based on V4 — read header · condition + clickable severity · decisions/remediation/when-it-happens/satisfies with the GovernSheet's FULL option sets (7 decisions with hints, 5 actions incl. require-approval and route, all 10 packs editable) — save beneath section 3. Condition adding is wired in V4 + V5: "Add a Condition" appends an clause with an and/or join pill (signal-appropriate field, sentence pills, per-clause ✕ remove, capped at 4 like the GovernSheet); the edit model is a clause array shared across variations, so V1–V3 render clause one only. Note: "or" is aspirational — the backend rule model is all-clauses (AND) today, so an or-join is a real capability ask, not just UI.

Qpoint Brand Style Guide