cc/sonnet

20260818T012743Z-f4e44a

runningdev / untrustedhigh effort · prompt=baseline · matrix · — · — · 2026-08-18T01:27:44.811834+00:00

dev / untrusted. Candidate code ran on the host through the development-only native backend, with the evaluator inside the same trust boundary. Useful evidence, not a publishable benchmark score.How to read this →

Verdict

No grade was produced · running

A run with status running is not converted into a score. The manifest and digests below are preserved exactly as recorded so the failure stays auditable.

Apparatus

Agent, variant, timing, usage and content digests, as recorded.

Agent

Selector: cc/sonnet
Runtime: claude_code · 2.1.233 (Claude Code)
Model requested: sonnet
Model resolved: —
Effort: high
Tool profile: baseline

Variant & network

Variant: prompt=baseline · matrix
Network policy: deny
Enforced: false

The declared network policy was not enforced by a sandbox in this run.

Timing & usage

Started: 2026-08-18 01:27:44 UTC
Finished: —
Elapsed: —
Input tokens: —
Output tokens: —
Cost: —

Content digests

evaluator_environment: 55f70d30b5f1a5601563cf20f63ec372ace54a40b2893fd72300a76ff04713b8
grader
oracle
runtime_config: b8f42d18225a47e0c28bb0cb095f136b31668defc61e9d59325e2f3bc1fedce4
submission
task: 0733100788cf2f2f084fff23ed8d9862cf151e153bca06e6addd16f4c3302c9e

Digests pin the exact task package, submission tree, oracle, grader and evaluator environment used. They are how a result is re-verified without trusting this page.

Trace summary

Derived counts only — raw logs are not published.

trace summarizedclaude-code-jsonl-v1

Derived trace statistics
events	57
tool calls	19
commands	4
file edits	0
wall clock	—
input tokens	14
output tokens	4.1k

A full trace viewer is planned. v1 publishes only the derived summary above — raw events.jsonl and stderr.log stay in the run directory and are never uploaded, since they can contain host paths, environment details and unredacted tool output.