{% if name_chip %}{{ name_chip }}{% endif %}{{ display_name }} {# The old standalone "kind" + "phase" pills were visual noise next to the title (uniform across a computational investigation; the phase just duplicated the gate ladder below). Kind is kept as a data attribute for any consumer; the lifecycle stage is now shown — and highlighted "you are here" — inside the six-gate status row below (gate-current). #}
{# Two-axis study markers — Progress (● Completed) × Verdict (◆ Supported) — lined up inline with the title (filled by _renderStudyAxisMarkers in study-detail.js) so they don't take a separate row above the content. #}Q {{ _q }}
{% endif %}
{# Figures download moved to the study tab bar's ↓ Export menu (top-right);
the per-tab "↓ visualizations" button + "Figures" heading were removed. #}
{# ─── Embedded HTML visualizations (e.g. pre-execution preview pages
produced by a runner script). study.embed_visualizations is a list of
{name, url, description} entries; each renders as an inline iframe
alongside the native gallery / auto-rendered charts, so a biologist
gets immediate payoff without waiting for runs.db / Visualization-
class plumbing.
Fable §4.5 (Task V2): same `.figure-card` shell as the other two
figure sources — no bordered/backgrounded header bar. The name +
"Open in new tab" affordance move into the caption row footer,
alongside a muted `embed` source chip. `embed-viz-card` stays as a
second class for `_figuresHasEmbeds()` / existing hooks to key on.
Task V3: `v.run_id` is populated only for the auto-rendered
studies//viz/*.html source (discover_viz_html_files attaches
the run whose data the file represents); hand-authored
reports/figures/ entries and study.yaml-declared embeds have no
run_id key, so `{% if v.run_id %}` omits the link rather than
fabricate one. The anchor is server-rendered here (unlike the
async native/chart sources) and wired via the SAME delegated
`_wireFigureRunLinks` listener, called once at load in
study-detail.js since these cards exist in the DOM from page load.
#}
{% if study.embed_visualizations %}
{% for v in study.embed_visualizations %}
{% if v.mode == 'launch' or v.module %}
{# A deeper interactive Explore module (e.g. the Corpus Explorer): too rich to
cram into an inline iframe, so link out to it full-size. These are the same
modules the Explore tab lists; this is the study→module connection. #}
{% else %}
{% endif %}
{% endfor %}
{% endif %}
{# Shared empty state for the Figures section (Fable A #3). The native
gallery and latest-run charts each resolve async and embeds render
synchronously above; JS shows this line only once all three have
reported and none produced content — never while any one of them
still holds actual figures. #}
Loading figures…
Explore module
{{ v.name }}
{% if v.description %}
Open {{ v.name }}
{{ v.description }}
{% endif %}Model composition
{# Download the runnable notebook for this study's investigation (includes this study, with its composite + exposed parameters). Lives in the Model tab — where the composition/params are — rather than the header. #}Loading composite card(s)…
Runnable models — the composite(s) this study simulates
{% for b in study.baseline %}
{{ b.name or 'baseline' }}
{% if b.composite %}
{# Replace this entry's composite ref — the only way to point a study at a
real composite once created: the pre-unification "+ Add baseline" form
(_submitBaselineAdd) was removed from this template, leaving its JS
handler and the /api/study-baseline-add /-remove endpoints orphaned. #}
{{ b.composite }}
{% else %}no composite declared{% endif %}
{# Populated from the live composite registry by
_populateBaselineCompositeSelects() (study-detail.js) once
DataSource.loadComposites() resolves — this initial single option
(the currently-declared ref, resolved or not) is just the
pre-JS/first-paint state so the field is never empty. #}
{# Fold in study.conditions.baseline's params when it names the SAME
(non-empty) composite as this entry — only the FIRST matching
study.baseline[] entry gets the fold (review fix: two entries
sharing a composite must not both re-render the same extras), so
nothing renders twice (Fable #14). A param present in both with a
DIFFERENT value is never silently dropped — it's rendered as a
visible divergence instead (review fix: same-key-different-value
was disappearing from the DOM entirely). #}
{% if _cond_baseline and _cond_baseline.composite and not _ns.cond_baseline_folded and _cond_baseline.composite == b.composite %}
{% set _ns.cond_baseline_folded = true %}
{% set _extra_params = {} %}
{% set _diverging_params = {} %}
{% for k, v in (_cond_baseline.params or {}).items() %}
{% if not (b.params and k in b.params) %}
{% set _ = _extra_params.update({k: v}) %}
{% elif b.params[k] != v %}
{% set _ = _diverging_params.update({k: v}) %}
{% endif %}
{% endfor %}
{% if _extra_params or _diverging_params %}
{% for k, v in _extra_params.items() %}
{{ k }}=
{% endif %}
{% endif %}
{# Config that runs — resolved params (composite defaults ⊕ this study's
overrides), fetched on demand from /api/composite-resolve. #}
{{ v }}
{% endfor %}
{% for k, v in _diverging_params.items() %}
{{ k }}={{ v }}
⚠ differs — baseline: {{ b.params[k] }}
{% endfor %}
Loading configuration…
{% if study.robustness %}seeds={{ study.robustness.seeds|length if study.robustness.seeds else (study.robustness.n_replicates or 1) }}{{ ' (sweep)' if study.robustness.parameter_sweep else '' }}{% endif %}
{% if _r and _r.n_steps %}steps={{ _r.n_steps }}{% endif %}
{% if _r and _r.emitter %}emitter={{ _r.emitter }}{% endif %}
{% if _r and _r.status %}status={{ _r.status }}{% endif %}
{% endif %}
baseline
{% if _cond_baseline.composite %}
{% if _cond_baseline.params %}
{{ _cond_baseline.composite }}
{% else %}no composite declared{% endif %}
{% for k, v in _cond_baseline.params.items() %}
{{ k }}=
{% endif %}
{{ v }}
{% endfor %}
Loading configuration…
{% if study.robustness %}seeds={{ study.robustness.seeds|length if study.robustness.seeds else (study.robustness.n_replicates or 1) }}{{ ' (sweep)' if study.robustness.parameter_sweep else '' }}{% endif %}
{% if _r and _r.n_steps %}steps={{ _r.n_steps }}{% endif %}
{% if _r and _r.emitter %}emitter={{ _r.emitter }}{% endif %}
{% if _r and _r.status %}status={{ _r.status }}{% endif %}
{% endif %}
Conditions — what we set up to test it
The experimental knobs: how variants perturb the baseline, and which model settings still need a human-set value.
{% if _cond.variants %}Variants ({{ _cond.variants | length }})
| Variant | Composite / base | Parameter overrides | Notes |
|---|---|---|---|
{{ v.name or '?' }} |
{% if v.composite %}{{ v.composite }}
{% elif v.base_composite %}{{ v.base_composite }}
{% else %}(inherits baseline){% endif %}
|
{% if _ovr %}
{% for k, val in _ovr.items() %}
{{ k }}=
{% endfor %}
{% else %}
(no overrides)
{% endif %}
{{ val }} |
{{ v.description or v.notes or '' }} |
{% set _n_required = _model_settings | selectattr('gate','equalto','required-before-run') | list | length %}
{% endif %}
Model settings ({{ _model_settings | length }})
{% if _n_required %}{{ _n_required }} required{% endif %}
Edit a value in the Current column and click Save. The next pbg_runner invocation for this study reads the updated value from study.yaml.
| Name | Type | Default | Current | Range | Gate | |
|---|---|---|---|---|---|---|
{{ e.name }} |
{{ e.type or '' }} | {{ e.default }} |
{% if e.options %} {# A declared finite value set — select from it rather than typing a value that has to exactly match one of e.options from memory. #} {% else %} {% endif %} {# Fable #14 — reuse the SAME required-before-run gate signal the header badge above is computed from, inline per-row. #} {% if e.gate == 'required-before-run' and (e.current is none or e.current == '') %} ⚠ needs a value {% endif %} |
{% if e.range %}[{{ e.range[0] }}, {{ e.range[1] }}]
{% elif e.options %}{{ e.options | join(' | ') }}
{% else %}—{% endif %}
|
{% if e.gate == 'required-before-run' %}required {% else %}optional{% endif %} |
Model change
{% if study.model_change.base_model %}Base model: {{ study.model_change.base_model }}
New processes:
-
{% for p in study.model_change.new_processes %}
{{ p.name }}({{ p.kind or 'Process' }}{% if p.requirement_id %} · req: {{ p.requirement_id }}{% endif %}) {% endfor %}
New state variables:
-
{% for v in study.model_change.new_state_variables %}
{{ v.id or v }}{% if v.kind %} ({{ v.kind }}{% if v.units %}, {{ v.units }}{% endif %}){% endif %} {% endfor %}
New parameters:
-
{% for p in study.model_change.new_parameters %}
{{ p.name }}{% if p.default is defined %} = {{ p.default }}{% endif %}{% if p.units %} ({{ p.units }}){% endif %} {% endfor %}
New listeners:
-
{% for l in study.model_change.new_listeners %}
{{ l.name }}{% if l.emits %} →{{ l.emits | join(', ') }}{% endif %} {% endfor %}
Modified processes:
-
{% for m in study.model_change.modified_processes %}
-
{% if m is mapping %}
{% if m.why %}
{{ m.name }}{% if m.status %}{{ m.status }}{% endif %} {% if m.requirement_id %}req: {{ m.requirement_id }}{% endif %}{{ m.why }}{% endif %} {% else %} {{ m }} {% endif %}
{% endfor %}
{{ study.model_change.notes }}
{% endif %}Implementation requirements ({{ study.implementation_requirements | length }} items)
Concrete pieces of work that gate this study from Design → Build. Each generates a task.
{% for r in study.implementation_requirements %} {% if r._prose or r is string %} {# Prose-form requirement (authored as `implementation_requirements: |`). #}{{ r.description if r is mapping else r }}
{% else %}
{{ r.id }} — {{ r.title }} {% if r.effort %}{{ r.effort }}{% endif %} {% if r.kind %}[{{ r.kind }}]{% endif %}
{% if r.description %}{{ r.description }}
{% endif %} {% if r.steps %}Steps:
-
{% for s in r.steps %}
- {{ s }} {% endfor %}
Unblocks:
{% for u in r.unblocks %}{{ u }}{% if not loop.last %} · {% endif %}{% endfor %}
Deferred until: {{ r.defer_until }}
Readouts
{# Emitter summary — always visible, one line (kind · interval · scope →
output), auto-filled from the composite's resolved default emitter. #}
{# Points at Simulations' raw-data downloads rather than duplicating them;
_loadReadoutsDownloadPointer only shows it when there's data to download. #}
{% if study.expected_behavior %}
{% endif %}
Loading emitter…
Loading readouts from the composite emit plan…
How observables tie to Expected Behavior
Observables can be referenced by name in an
expected_behavior[i].measure block (e.g. kind: bulk_count, id:
"MONOMER0-160[c]"). When the dashboard validates a study, it cross-checks
that every measure resolves to a declared observable, or to a
built-in helper.
{# Launch actions relocated here from the study header (they were crowding the
top-right corner where the docked investigation graph parks, and they belong
with the runs they produce). All element ids preserved so study-detail.js
wires them unchanged; still live-backend only (hidden in the snapshot). #}
{# item 110: dispatch ANY registered process-bigraph composite_id (e.g. pbg-native's
v2ecoli.composites.lineage_ray_batch, item 101/109) to the remote compute backend --
independent of this study's own pinned baseline composite. Remote-pinned only, same
gate as "Run current spec" (see _dispatchRemoteComposite in study-detail.js). #}
Loading simulations…
{# Compact run summary (N runs · completed · latest run + time · n_steps ·
emitter · size) — auto-filled by _loadResultsSummary from the runs list. #}
{# Gate results — each acceptance test's graded scalar readout (or an
unresolved stub). These are run-derived RESULTS, so they live here, not in
the Visualizations tab (which holds only rendered figures). Filled by
_loadGateResults from the study-charts payload's `gate_results`. #}
Gate results — each acceptance test's value, graded from run data
Latest run preview — downsampled sparkline + first/last/min/max per store
Loading results preview…
Raw simulation data
Loading runs…
{# Analyses to run at dispatch time — which post-run analyses this study's
dispatch should trigger, translated into v2ecoli's analysis_options
(lib.study_run_post.build_analysis_options). Populated by
_loadStudyAnalyses() (study-detail.js) from the live
/api/visualization-classes registry, preserving any name already declared
in this study's own spec even if the current registry doesn't have it.
Saved via _saveStudyAnalyses() → /api/study-set-analyses. Lives here on the
Analyses tab, alongside the result files these analyses produce. #}
Analyses to run — which of the workspace's
Analyses to run — which of the workspace's ANALYSIS_REGISTRY entries to compute on dispatch
Loading analyses…
Derived result files (.csv/.tsv) computed from the runs —
distinct from Results, which holds the raw emitted arrays.
Analysis result files
Loading result files…
Loading readiness…
{# C5 moved literature_anchors here verbatim from the dissolved Overview
"Plan & provenance" group. C6 restyles it as an "Expectations" band: each
anchor is a stated expectation this study is measured against, and where
it carries a status (`status_in_v2ecoli`) that status is routed through
the SAME G3 outcome_label/_class/_glyph filters the acceptance-criteria
band below uses — met / conditional-pass / not met / not assessable —
instead of a raw free-text string, so it reads in the page's one shared
vocabulary. Read-only (editing anchors is out of scope, §7). Placed
immediately before the acceptance-criteria band: a coherent "what we
expected, then what the predefined bar is" grouping. Guarded so absent
renders nothing (§2 R2). #}
{% if study.literature_anchors %}
{% set _exp_outcome_colors = {
'met': {'bg': '#d1fae5', 'fg': '#065f46'},
'conditional': {'bg': '#fef3c7', 'fg': '#92400e'},
'not-met': {'bg': '#fee2e2', 'fg': '#991b1b'},
'not-assessable': {'bg': '#f1f5f9', 'fg': '#475569'},
} %}
{% endif %}
{# Fable G4 (§10.1 band 1 — "Acceptance criteria"): the predefined bar,
stated BEFORE the outcome. Source is `spine_acceptance` — already
attached to the spec by study_spec.load_study_detail_spec via
study_enrichment.study_acceptance_criterion (a pure read of the owning
investigation's PERSISTED executive.computed_acceptance, filtered to
this study's criteria; never recomputed here). Absent/empty means the
study has no owning investigation or no persisted acceptance — render
NOTHING (§2 R2: absent != empty; no fabricated criteria, no empty box).
Each criterion's `result` (discovery_engine.investigation_status's
passing/failing/passing-with-caveats/in-progress vocabulary) is mapped
through the SAME shared outcome_label/_class/_glyph filters G3 defined,
so this band reads in the identical four-value vocabulary as the rest
of the page. `behavior` names the behavior_tests entry the criterion
rests on — where it also appears in this study's Behavioral tests list
below, link to it (`#bt-`) as the evidence the outcome rests
on. #}
{% set _sa = study.spine_acceptance %}
{% if _sa and _sa.criteria %}
{% set _outcome_colors = {
'met': {'bg': '#d1fae5', 'fg': '#065f46', 'border': '#10b981'},
'conditional': {'bg': '#fef3c7', 'fg': '#92400e', 'border': '#f59e0b'},
'not-met': {'bg': '#fee2e2', 'fg': '#991b1b', 'border': '#ef4444'},
'not-assessable': {'bg': '#f1f5f9', 'fg': '#475569', 'border': '#94a3b8'},
} %}
Acceptance criteria
— the predefined bar, from
{% endif %}
{# Gate/audit score line — "N/M gates passed", filled client-side by
_renderTestsGateSummary from spec.outcome_rollup so it can't drift from
the per-gate list below. The per-gate detail is NOT repeated here — it
renders once, in the "Behavioral tests" list (#tests-list) below
(Fable §4.6: one score line + one list, not the gate set twice). #}
{# Study-spine reorg (spec §3.6): the rigor scorecard (Fable G5) and L0-L5
reproducibility audit (Fable G6) check groups MOVED out of Tests into
Assurance › Audit (`panel-audit`, below) — Tests grades the study's own
Test bar; Audit judges whether that bar is trustworthy. #}
{# REQUIREMENT (study-spine reorg, plan Task 3): the Tests panel must show
the COMPLETE set of this study's report cards — every card under
`viz/report_card/`, not just the ones wired to a `behavior_tests` /
`expected_behavior` entry below. `render_report_cards_section` scans the
directory directly (the same source `report_card_urls` feeds), so a
card with no matching Test entry still renders here. Server-rendered
(best-effort — a study with no report cards emits "" and this section
drops, §2 R2 absent != empty). #}
{% if report_cards_html %}
{% endif %}
{% set _bt = study.behavior_tests or study.expected_behavior or study.tests %}
{# Latest test outcomes per test — single-sourced from study_spec._latest_outcomes
(spec.latest_outcomes) so the row pills, the Tests-tab JS summary, and the
Conclusions rollup can't drift. Falls back to an empty map for older specs. #}
{% set _outcomes = study.latest_outcomes or {} %}
{% if _bt %}
Expectations — from literature, mapped to the model observable that measures each
-
{% for anchor in study.literature_anchors %}
{% set _ocls = anchor.status_in_v2ecoli|outcome_class if anchor.status_in_v2ecoli else None %}
-
{{ anchor.expectation }} {% if anchor.status_in_v2ecoli %} {% set _oc = _exp_outcome_colors[_ocls] %} {{ anchor.status_in_v2ecoli|outcome_glyph }} {{ anchor.status_in_v2ecoli|outcome_label }} {% endif %}Model observable:{% if anchor.source %}
{{ anchor.model_observable }}Source: {{ anchor.source }}{% endif %}
{% endfor %}
Acceptance criteria
— the predefined bar, from {{ _sa.investigation }}'s persisted acceptance roll-up
-
{% for c in _sa.criteria %}
{% set _ocls = c.result|outcome_class %}
{% set _oc = _outcome_colors[_ocls] %}
-
{{ c.result|outcome_glyph }} {{ c.result|outcome_label }}
{{ c.behavior }}{% if c.behavior %} View test ↓ {% endif %}
{% endfor %}
Loading gate summary…
{{ report_cards_html | safe }}
{% endif %}
{# C6: each report_card-kind row (below, in Behavioral tests) now carries
its own inline expander rendering the FULL card in place — so this top
panel would double-render every card if it also emitted the per-card
stack. It is narrowed to ONLY the cross-card plotly comparison overview
(no per-row equivalent exists for that), and — per absent != empty
(§2 R2) — the whole section is server-gated on comparison_plotly_url so
a study with no comparison plot renders no empty box here at all. #}
{% if study.comparison_plotly_url %}
Report cards — cross-card comparison; each card's full scorecard is inline at its row in Behavioral tests below
Loading comparison…
Behavioral tests
— no test results yet —
auto_discover:
true
data_source: latest_run
-
{% for b in _bt %}
{% set _cls = b.classification or 'unclassified' %}
{% set _border = '#10b981' if _cls == 'primary' else ('#3b82f6' if _cls == 'supporting' else ('#f59e0b' if _cls == 'diagnostic' else ('#94a3b8' if _cls == 'regression' else '#cbd5e1'))) %}
{% set _out = _outcomes.get(b.name) %}
{% set _res = _out.result if _out else None %}
{% set _pill_bg = '#d1fae5' if _res == 'PASS' else ('#fee2e2' if _res == 'FAIL' else ('#fef3c7' if _res == 'SKIP' else ('#fde68a' if _res == 'PARTIAL' else '#f1f5f9'))) %}
{% set _pill_fg = '#065f46' if _res == 'PASS' else ('#991b1b' if _res == 'FAIL' else ('#92400e' if _res == 'SKIP' else ('#92400e' if _res == 'PARTIAL' else '#475569'))) %}
{% set _pill_text = ('✓ PASS' if _res == 'PASS' else ('✗ FAIL' if _res == 'FAIL' else ('⏭ SKIP' if _res == 'SKIP' else ('◐ partial' if _res == 'PARTIAL' else '⏳ pending')))) %}
-
{% if (b.kind or 'behavioral') == 'report_card' %}
{# C6: the card expands INLINE at this row via a native — reveals the same rich content _renderRichReportCard(card) renders on the Tests tab, mounted lazily on first expand (JS binds a `toggle` listener; avoids paying for hidden Plotly/markup for cards nobody opens). The verdict pill stays on the summary (recoloured by the existing _fillReportCardModules); there is no more "View report card ↑" scroll-away link — the top panel is now plotly-comparison-only, so there is nothing up there to jump to for this card any more. #} {% else %} {# --- existing behavioral rendering, UNCHANGED, kept verbatim below --- #}{{ _pill_text }} {{ _cls }} {{ b.name or '(unnamed)' }} {% if b.requires_simulation %} requires:
{{ b.requires_simulation }}{% endif %}{{ b.description or b.en or '' }}{% if _out %}Latest result: {# `_out` is usually a mapping of observable -> value, but some studies record a plain string/scalar. Guard `.items()` so a string result renders as-is instead of raising UndefinedError and aborting the whole per-study shell render (which silently 404s the snapshot). #} {% if _out is mapping %} {% for k, v in _out.items() if k != 'result' %}{% endif %} {% if b.measure or b.pass_if or b.given or b.expect %}{{ k }}: {{ v }}{% endfor %} {% else %}{{ _out }}{% endif %}{% endif %} {% if b.calibration_anchor %}Assertion
{# `measure` and (in the else branch) `expect` are rendered unconditionally here — only `given`/`pass_if`/`requires`/`cites` are behind their own `{% if %}`. When one of those two is simply absent, Jinja's Undefined sentinel isn't JSON-serializable and a bare `| tojson` raised `TypeError: Object of type Undefined is not JSON serializable` (a real, reproducible 500 for any behavior test that declares `pass_if` but no `measure`). `default(none)` converts the sentinel to a real `None` -> `null` first, matching the JSON-dump style of the rest of this block. #}{% if b.given %}given: {{ b.given|tojson }} {% endif %}measure: {{ b.measure|default(none)|tojson }} {% if b.pass_if %}pass_if: {{ b.pass_if|tojson }}{% else %}expect: {{ b.expect|default(none)|tojson }}{% endif %}{% if b.requires %} requires: {{ b.requires|tojson }}{% endif %}{% if b.cites %} cites: {{ b.cites|tojson }}{% endif %}{% endif %} {% if b.notes %}⚠ Calibration anchor
{{ b.calibration_anchor | tojson(indent=2) }}Note: {{ b.notes }}
{% endif %} {% endif %}
{% endfor %}
- No
expected_behavior:entries declared in study.yaml yet. Add some to see them appear here as test rows. Click "Run tests" to discover and execute additionalstudies/<slug>/tests/test_*.pyfiles alongside the auto-generated assertions.
Loading sufficiency checks…
Loading quality checks…
Loading reproducibility checks…
Loading sourcing…
Plan & lifecycle
{% for g in gate_ladder %}
{{ g.number }} · {{ g.name }}
{% if g.authored_value %}{{ g.authored_value }}{% endif %}
{% if g.current %}you are here{% endif %}
{% endfor %}
{# Terminal: when every gate is done the study is released → implementation. #}
Implementation{% if _all_done %} complete{% endif %}
Verdict & conclusion
{# G7: honest attribution for the frozen verdict. No actor field is persisted with the conclusion card (viz/report_card/conclusion.verdict.json is {schema, overall, tracks, insight} — checked, not guessed) so this always renders the literal "unattributed" token today. `none` is passed explicitly — no actor field is read here, per the "no new schema fields" constraint. #}{{ none | actor_glyph }} {{ none | attribution_text }}
{# G8: frozen-record indicator (Fable §11.4). conclusion_card_frozen is ONLY set (study_spec.load_study_detail_spec) when viz/report_card/conclusion.verdict.json actually exists and parses — the file's presence IS the freeze signal, so this block is absent (never an empty placeholder) for a study whose verdict has never been persisted. `when` is the file's own mtime (fmt_ts, same convention as every run timestamp elsewhere on this page); `digest` is computed on the fly over the exact frozen payload via conclusion_digest — never stored, never fabricated, changes iff the payload changes. #} {% set _frozen = study.conclusion_card_frozen %} {% if _frozen %}
🔒 frozen {{ _frozen.when | fmt_ts }} UTC ·
{{ _frozen.payload | conclusion_digest }}
⚖️ Verdicts — three-track outcome (regression / empirical / explanatory)
Computed gate verdict
The coded gate evaluator's result from the behavior-test outcomes — the evaluation the conclusion rests on.
{{ _cgv.result or 'pending' }}
{% if _cgv.evaluated_by %}evaluated by {{ _cgv.evaluated_by }}{% endif %}
{% if _cgv.counts %}{{ _cgv.counts.pass }}/{{ _cgv.counts.total }} passed{% if _cgv.counts.fail %} · {{ _cgv.counts.fail }} failed{% endif %}{% if _cgv.counts.skip %} · {{ _cgv.counts.skip }} skipped{% endif %}{% endif %}
{% if _cgv.diverges_from_authored %}
⚠ gate drift — code:
{% endif %}
{{ _cgv.result }} · authored: {{ study.gate_status or '—' }}
Conclusion logic — gate decision
{% if _cl.if_primary_tests_pass %}
✓ If primary tests pass:
{% if _cl.if_primary_tests_pass.implementation_status %}
{% endif %}
{% if _cl.if_primary_tests_fail %}
Implementation: {{ _cl.if_primary_tests_pass.implementation_status }}
{% endif %}
{% if _cl.if_primary_tests_pass.biological_validation %}
Empirical validation: {{ _cl.if_primary_tests_pass.biological_validation }}
{% endif %}
{% if _cl.if_primary_tests_pass.pipeline_unblocks %}
Pipeline unblocks:
{% endif %}
- {% for u in _cl.if_primary_tests_pass.pipeline_unblocks %}
- {{ u }} {% endfor %}
✗ If primary tests fail:
{% if _cl.if_primary_tests_fail.diagnose %}
{% endif %}
Diagnose:
{% endif %}
{% if _cl.if_primary_tests_fail.block_downstream %}
- {% for d in _cl.if_primary_tests_fail.diagnose %}
- {{ d }} {% endfor %}
Block downstream: {{ _cl.if_primary_tests_fail.block_downstream }}
{% endif %}
Conclusion
{{ study.conclusion | markdown }}
Plan & assumptions
{% if study.pipeline_gate %}Pipeline gate
Each study is a model-building gate: it answers "Can we safely build the next layer?"
{% if study.pipeline_gate.prerequisites %}Prerequisites:
{% for p in study.pipeline_gate.prerequisites %}
{% else %}
{{ p }}{% if not loop.last %} · {% endif %}{% endfor %}
Prerequisites: none (root study)
{% endif %}
{% if study.pipeline_gate.enables %}
Enables:
{% for p in study.pipeline_gate.enables %}
{% endif %}
{% if study.pipeline_gate.proceed_condition %}
{{ p }}{% if not loop.last %} · {% endif %}{% endfor %}
Proceed when: {{ study.pipeline_gate.proceed_condition }}
{% endif %}
Key assumptions
-
{% for a in study.key_assumptions %}
- {{ a }} {% endfor %}
Limitations
What this study explicitly does NOT validate (prevents overclaim).
-
{% for l in study.limitations %}
- {{ l }} {% endfor %}
Evidence
{% if _is_v3 %} {# ── Latest run outcomes ────────────────────────────────────────── #} {% set _latest = (study.runs or [])[-1] if study.runs else None %} {% if _latest %}Latest run outcomes
{{ _latest.simulation or 'baseline' }} · seed {{ _latest.seed or '?' }} ·
{{ _latest.duration_s or '?' }}s simulated ·
{{ _latest.rows or '?' }} history rows ·
started {{ _latest.started_at or '?' }}
{# Spine C1b: outcomes are CANONICAL on the Tests tab (thread B — per-test
measured value · evaluator · band). Defer to it instead of re-dumping a
second, cruder raw k:v table here. #}
{% set _roll = study.outcome_rollup or {} %}
{% if _roll.total %}
{{ _roll.total }} test outcome{{ '' if _roll.total == 1 else 's' }} — {{ _roll.PASS }} PASS · {{ _roll.FAIL }} FAIL{% if _roll.PARTIAL %} · {{ _roll.PARTIAL }} partial{% endif %}. Per-test outcomes (measured value · evaluator · band) are canonical on the Tests (in Audit) →.
{% endif %}
Synthesised read-only from canonical fields — edit findings,
limitations, and discovery_implications to change
this (and the generated report) together.
Claims (from findings)
{% if study.findings %}-
{% for f in study.findings %}{% if f.statement or f.summary %}
- {{ f.statement or f.summary }}{% if f.tier %} {{ f.tier }}{% endif %} {% endif %}{% endfor %}
No findings recorded yet.
{% endif %}Evidence (from findings)
{% set _has_ev = study.findings and (study.findings | selectattr('evidence') | list) %} {% if _has_ev %}-
{% for f in study.findings %}{% if f.evidence %}
- {% if f.evidence is string %}{{ f.evidence }}{% elif f.evidence is mapping %}{% for k, v in f.evidence.items() %}{{ k }}: {{ v }}{% if not loop.last %}; {% endif %}{% endfor %}{% else %}{% for e in f.evidence %}{{ e }}{% if not loop.last %}; {% endif %}{% endfor %}{% endif %} {% endif %}{% endfor %}
No evidence recorded on findings yet.
{% endif %}Follow-ups & decisions
{# ── Open debts — pre-run expert review ───────────────────────────────── C5: relocated verbatim from the dissolved Overview "Plan & provenance" group. These are open biological questions the planning is contingent on — grouped here under the Decision act's follow-ups/decisions area as "Open debts" (leads the group; they gate the decision). Guarded so absent renders nothing. The G7 attribution line + all {% if %} guards are preserved verbatim from the original card. #} {% if study.expert_decisions_needed is sequence and study.expert_decisions_needed is not string and study.expert_decisions_needed %}Open debts (pre-run expert review)
Open biological questions the planning is contingent on. A "wrong" answer here means a primary test threshold needs to change before the simulation is run, not after. Each question lists alternatives + the impact of the wrong choice + what downstream work it blocks.
{% for q in study.expert_decisions_needed %}
{% endfor %}
{{ q.id }}
{{ q.status or 'open' }}
{# G7: honest attribution — asked_to is the only actor field this
card carries (no timestamp field exists for it). Never blank:
the literal "unattributed" token renders when asked_to is unset,
with the human/agent glyph on whatever actor IS recorded. #}
asked to: {{ q.asked_to | actor_glyph }} {{ q.asked_to if q.asked_to else 'unattributed' }}
Q. {{ q.question }}
{% if q.alternatives %}
Alternatives:
{% endif %}
{% if q.impact_if_wrong %}
- {% for a in q.alternatives %}
- {{ a }} {% endfor %}
Impact if wrong: {{ q.impact_if_wrong }}
{% endif %}
{% if q.blocks %}
What this blocks ({{ q.blocks|length }} items)
{% endif %}
{% if q.requested_response %}
- {% for b in q.blocks %}
- {{ b }} {% endfor %}
Requested response format
{% endif %}
{{ q.requested_response }}
Discovery implications
Where this study's results leave the mechanism model — and what to investigate next.
{% if _di.resolved_uncertainties or _di.remaining_uncertainties %}
{% if _di.resolved_uncertainties %}
{% endif %}
{% if _di.remaining_uncertainties %}
{% endif %}
{% endif %}
{% if _di.alternate_hypotheses %}
✓ Resolved uncertainties
- {% for u in _di.resolved_uncertainties %}
- {{ u }} {% endfor %}
● Remaining uncertainties
- {% for u in _di.remaining_uncertainties %}
- {{ u }} {% endfor %}
Alternate hypotheses ({{ _di.alternate_hypotheses | length }})
{% for h in _di.alternate_hypotheses %}{{ h.statement or '(untitled hypothesis)' }}
{% if h.why_plausible %}{{ h.why_plausible }}
{% endif %}
{% if h.mechanism_elements_affected %}
Mechanism elements:
{% for e in h.mechanism_elements_affected %}
{% endif %}
{% if h.discriminating_observables %}
{{ e }}{% if not loop.last %}, {% endif %}{% endfor %}
Discriminating observables:
{% for d in h.discriminating_observables %}
{% endif %}
{{ d }}{% if not loop.last %}, {% endif %}{% endfor %}
Mechanism update proposals ({{ _di.mechanism_update_proposals | length }})
{% for m in _di.mechanism_update_proposals %}{{ m.mechanism_node_or_edge or '(unspecified)' }}
{{ m.update_type or 'revise' }}
{% if m.confidence_change %}Δconfidence: {{ m.confidence_change }}{% endif %}
{% if m.requires_expert_approval %}needs expert approval{% endif %}
{{ m.rationale }}
{% endif %}
Follow-up study proposals (discovered) ({{ _di.followup_study_proposals | length }})
Click ➕ Add to investigation to spawn a new study node in the investigation graph.
{% for p in _di.followup_study_proposals if (p.id or '') not in _fup_ids %}
{% set _gain = (p.expected_information_gain or '')|lower %}
{% endfor %}
{{ p.title or '(untitled proposal)' }}
{% if p.study_type %}{{ p.study_type }}{% endif %}
{% if p.source_trigger %}{{ p.source_trigger }}{% endif %}
{% if _gain %}gain: {{ _gain }}{% endif %}
{% if p.priority %}priority: {{ p.priority }}{% endif %}
{% if p.proposed_experiment %}{{ p.proposed_experiment }}
{% endif %}
{% if p.target_mechanism_elements %}
Targets:
{% for t in p.target_mechanism_elements %}
{% endif %}
{{ t }}{% if not loop.last %}, {% endif %}{% endfor %}
Follow-up studies — pick one to seed (authored)
Click Seed new study → on any non-existing follow-up to scaffold a child study.yaml.
The new study inherits this one as a pipeline_gate prerequisite and starts at phase: Design.
{% for f in study.follow_up_studies %}
{% endif %}
{% endfor %}
{{ f.kind or 'other' }}
{% if f.effort %}{{ f.effort }}{% endif %}
{% if f.status %}{{ f.status }}{% endif %}
{{ f.title }}
{% if f.why %}Why: {{ f.why }}
{% endif %}
{% if f.hypothesized_mechanism %}
Hypothesized mechanism: {{ f.hypothesized_mechanism }}
{% endif %}
{% if f.unblocks %}
Unblocks:
{% endif %}
{% if f.acceptance %}
- {% for u in f.unblocks %}
- {{ u }} {% endfor %}
Acceptance:
{% endif %}
{% if f.kind not in ('existing',) %}
- {% for a in f.acceptance %}
- {{ a }} {% endfor %}
avoids an empty
collapsed shell for v3 studies. #}
{% if not _is_v3 %}
{% endif %}
Limitations & provenance
{# ── Read-only synthesis: Next steps (C3) ──────────────────────────────── Next steps ← discovery_implications.followup_study_proposals (or follow_up_studies). Edit those source fields and this synthesis — and the generated report — update together. C5: the Limitations column that used to sit here was REMOVED to reconcile a duplicate — limitations now render once, in the richer moved "Plan & assumptions" block above (Group 1b), for v3 and non-v3 studies alike. #} {% set _di = study.discovery_implications or {} %} {% set _nexts = _di.followup_study_proposals or study.follow_up_studies or [] %}Next steps (from discovery implications)
{% if _nexts %}-
{% for n in _nexts %}
{% if n is string %}
- {{ n }} {% else %}
- {{ n.title or n.name or 'proposal' }}{% if n.why or n.motivation %} — {{ n.why or n.motivation }}{% endif %} {% endif %} {% endfor %}
No follow-up proposals recorded yet.
{% endif %}