Releases: uw-syfi/TraceLab
Releases · uw-syfi/TraceLab
Release list
SyFI TraceLab v0.0.2
Highlights
- Expanded the sanitized public corpus to 665,453 agent steps across 8,058 sessions and 52 pseudonymous users, including 743,819 normalized tool calls.
- Added the Bash command breakdown: executable popularity for Claude and Codex, raw and lifecycle-aware latency rankings, command-structure skeletons, and privacy-safe executable labels.
- Improved Codex command timing by treating
write_stdincontinuations as part of the originatingexec_commandlifecycle. - Preserved historical rounds during multi-host recollection while filtering empty replay artifacts.
- Refreshed session, generation, tool-call, prefix-cache, and cost analyses, including English and Chinese web narratives.
- Privacy hardening: custom and integration-specific tool names are pseudonymized as stable
custom_tool_Nlabels.
Data assets
syfi_coding_trace.jsonl.gz— compressed sanitized JSONLsyfi_coding_trace.duckdb— query-ready DuckDB database
SHA-256
- JSONL.GZ:
11ce51ec0a25e3d1d95b025bca2f7d1647e47571eb7cc968acd5fc64d4b4fb65 - DuckDB:
a7bab286bc640844560850965ccf47975cf66407154132abaab90f27ec9be744
SyFI TraceLab v0.0.1
First semantic-versioned public release of the SyFI TraceLab dataset and toolkit.
A sanitized, real-world corpus of Claude / Codex coding-agent LLM traces, plus a prebuilt DuckDB
database for immediate analysis.
Dataset
| Rows (LLM rounds) | 357,161 |
| Tool records | 432,510 |
| Pseudonymous users | 43 |
| Providers | claude = 140,338 · codex = 216,823 |
Assets
syfi_coding_trace.jsonl.gz— normalized JSONL tracesyfi_coding_trace.duckdb— prebuilt compact DuckDB database
Verify
echo "9d265eae69a31cae203848bea936f018148eed7ca8bf56050c5abe96da0b4e6b syfi_coding_trace.jsonl.gz" | sha256sum -c -
echo "97715265367cc72376475f5d444c8e1900b88cab1482aa7b9a742894d9f15619 syfi_coding_trace.duckdb" | sha256sum -c -
gzip -t syfi_coding_trace.jsonl.gzSanitized: ids pseudonymized, local paths and tool inputs stripped. Please use responsibly and do
not attempt re-identification.
TraceLab public trace snapshot 2026-06-08
TraceLab Public Trace Snapshot
Sanitized public trace snapshot for the SyFI Trace Atlas public pool.
Assets
syfi_coding_trace.jsonl.gz— compressed normalized JSONL rowssyfi_coding_trace.duckdb— prebuilt DuckDB database for direct querying
Dataset
- Rows:
357,161 - Tool records:
432,510 - Distinct pseudonymous users:
43 - Providers:
claude=140,338,codex=216,823
SHA256
9d265eae69a31cae203848bea936f018148eed7ca8bf56050c5abe96da0b4e6b syfi_coding_trace.jsonl.gz
5d2ef12486cdfc26d770e8432fb45fd17381377be739a4d8e5b8556587721507 syfi_coding_trace.duckdb