Karti Tripathi
43 lines
856 B
Plaintext
43 lines
856 B
Plaintext
.venv/
|
|
__pycache__/
|
|
*.pyc
|
|
*.egg-info/
|
|
.env
|
|
.env.local
|
|
|
|
# Raw inspect logs are large and machine-specific. We extract the parts we
|
|
# care about (per-sample outcomes) into results/*.json, which IS committed.
|
|
logs/
|
|
*.eval
|
|
|
|
# Website
|
|
node_modules/
|
|
.next/
|
|
out/
|
|
|
|
# NOTE: data/private/ is intentionally NOT ignored. This repo is private on
|
|
# gitea; the private eval set is the whole point and must be versioned. If
|
|
# this repo is ever made public, data/private/ must be split out FIRST.
|
|
# See docs/DECISIONS.md#d3.
|
|
|
|
# Playwright MCP scratch output
|
|
.playwright-mcp/
|
|
|
|
# Secrets.
|
|
.env
|
|
.env.*
|
|
!.env.example
|
|
*.pem
|
|
*.key
|
|
*_rsa
|
|
*_ed25519
|
|
id_ed25519*
|
|
.ssh/
|
|
*secret*
|
|
*credentials*
|
|
|
|
# Private eval data never belongs in the public repo. Publishing the signal set
|
|
# contaminates it; publishing the canary inverts it. See scripts/publish-bench.sh.
|
|
data/private/
|
|
data/canary/
|