Emergent
v1
requirements unknown
2026-04-08
Gate passed

Escape the Dungeon · Emergent

escape-the-dungeon-emergent/v1

Composite score

0.303

Human review

0.10

Human review note

Main menu renders and recognizes saved game state (improvement over bare v1). But game crashes with 'Styled text error: unclosed tags START' when entering gameplay. Cannot play. Score 0.10: better than blank screen, menu works, but game is broken.

Reviewed by human · 2026-04-08

Dimension scores

Where the composite came from

Each dimension is scored 0.0 – 1.0 and combined using the weights in evals/escape-the-dungeon-emergent/gad.json. Human review dominates on purpose — process metrics alone can't rescue a broken run.

DimensionScoreBar
Human review0.100
Requirement coverage0.208
Implementation quality0.100
Iteration evidence0.140
Time efficiency0.971

Composite formula

How 0.303 was calculated

The composite score is a weighted sum of the dimensions above. Weights come from evals/escape-the-dungeon-emergent/gad.json. Contribution = score × weight; dimensions sorted by contribution so you can see what actually moved the needle.

DimensionWeightScoreContribution
skill_reuse0.150.7500.1125(36%)
workflow_quality0.100.6000.0600(19%)
time_efficiency0.050.9710.0486(15%)
requirement_coverage0.200.2080.0416(13%)
human_review0.300.1000.0300(10%)
implementation_quality0.150.1000.0150(5%)
iteration_evidence0.050.1400.0070(2%)
Weighted sum1.000.3146

Note: The weighted sum above (0.3146) doesn't exactly match the stored composite (0.3030). The difference is usually the v3 low-score cap (composite < 0.20 → 0.40, composite < 0.10 → 0.25) or a run with an older scoring pass.

Skill accuracy breakdown

Did the agent invoke the right skills at the right moments?

Skill accuracy data isn't relevant for this run (no expected trigger set).

Gate report

Requirement coverage

Total criteria
12
Fully met
2
Partially met
1
Not met
9

Reviewer notes on gates

Main menu renders, recognized saved game. But entering game gives error: 'Styled text error: unclosed tags START'. Cannot progress past main menu.

Process metrics

How the agent actually worked

Primary runtime
older runs may not carry runtime attribution yet
Agent lanes
0
0 root · 0 subagent · source missing
Observed depth
0 traced event(s) with agent lineage
Wall clock
14m
7 phases · 0 tasks
Started
Run start captured in TRACE timing metadata
Ended
Missing end time usually means the run was scaffolded but never finalized
Tool uses
79
67,375 tokens
Commits
2
0 with task id · 2 batch
Client debug · NEXT_PUBLIC_CLIENT_DEBUG=1
0 lines

No events yet. Window errors, unhandled rejections, and React render errors appear here. Set NEXT_PUBLIC_CLIENT_DEBUG_CONSOLE=1 to mirror console.error / console.warn.