Files
Open_Ultra_SAST/plane/runs/validation-46.yaml
T

564 lines
15 KiB
YAML
Raw Normal View History

# generated by `ousast plane workspaces benchmarks/independent/population-v2.toml --validation-set ~/ousast-results/independent-v2-sets/validation.json --candidates ~/ousast-results/independent-v2-modelsinks --triage ~/ousast-results/independent-v2-batched-check/scans --repos ~/.cache/openultrasast/independent --run validation-46 --runner-image ~/.cache/ousast/ax-src/runner-image`; do not edit, regenerate.
# 15 cases, 46 candidates (43 kept by the recorded triage); usd ceiling 2.02.
apiVersion: openultrasast.io/v1alpha1
kind: Run
metadata:
name: validation-46
spec:
tasks:
- name: budibase-mongo-template-nosqli-facts
task: repo-facts-budibase-mongo-template-nosqli
outputs:
- facts.json
budget: &id001
usd: 0
calls: 0
serialize: repo-facts
- name: budibase-mongo-template-nosqli-va
task: verify-a-budibase-mongo-template-nosqli
inputs:
facts: budibase-mongo-template-nosqli-facts/facts.json
outputs:
- units.jsonl
- summary.json
budget:
usd: 0.06
calls: 16
- name: budibase-mongo-template-nosqli-vb
task: verify-b-budibase-mongo-template-nosqli
inputs:
facts: budibase-mongo-template-nosqli-facts/facts.json
outputs:
- units.jsonl
- summary.json
budget:
usd: 0.06
calls: 16
- name: budibase-mongo-template-nosqli-agree
task: agree-budibase-mongo-template-nosqli
inputs:
pass_a: budibase-mongo-template-nosqli-va/units.jsonl
pass_b: budibase-mongo-template-nosqli-vb/units.jsonl
outputs:
- agreed.json
- disputed.json
budget: *id001
- name: cct-studio-server-cmdi-facts
task: repo-facts-cct-studio-server-cmdi
outputs:
- facts.json
budget: &id002
usd: 0
calls: 0
serialize: repo-facts
- name: cct-studio-server-cmdi-va
task: verify-a-cct-studio-server-cmdi
inputs:
facts: cct-studio-server-cmdi-facts/facts.json
outputs:
- units.jsonl
- summary.json
budget:
usd: 0.04
calls: 8
- name: cct-studio-server-cmdi-vb
task: verify-b-cct-studio-server-cmdi
inputs:
facts: cct-studio-server-cmdi-facts/facts.json
outputs:
- units.jsonl
- summary.json
budget:
usd: 0.04
calls: 8
- name: cct-studio-server-cmdi-agree
task: agree-cct-studio-server-cmdi
inputs:
pass_a: cct-studio-server-cmdi-va/units.jsonl
pass_b: cct-studio-server-cmdi-vb/units.jsonl
outputs:
- agreed.json
- disputed.json
budget: *id002
- name: contextforge-prompt-ssti-facts
task: repo-facts-contextforge-prompt-ssti
outputs:
- facts.json
budget: &id003
usd: 0
calls: 0
serialize: repo-facts
- name: contextforge-prompt-ssti-va
task: verify-a-contextforge-prompt-ssti
inputs:
facts: contextforge-prompt-ssti-facts/facts.json
outputs:
- units.jsonl
- summary.json
budget:
usd: 0.04
calls: 8
- name: contextforge-prompt-ssti-vb
task: verify-b-contextforge-prompt-ssti
inputs:
facts: contextforge-prompt-ssti-facts/facts.json
outputs:
- units.jsonl
- summary.json
budget:
usd: 0.04
calls: 8
- name: contextforge-prompt-ssti-agree
task: agree-contextforge-prompt-ssti
inputs:
pass_a: contextforge-prompt-ssti-va/units.jsonl
pass_b: contextforge-prompt-ssti-vb/units.jsonl
outputs:
- agreed.json
- disputed.json
budget: *id003
- name: froxlor-admin-ip-second-order-sqli-facts
task: repo-facts-froxlor-admin-ip-second-order-sqli
outputs:
- facts.json
budget: &id004
usd: 0
calls: 0
serialize: repo-facts
- name: froxlor-admin-ip-second-order-sqli-va
task: verify-a-froxlor-admin-ip-second-order-sqli
inputs:
facts: froxlor-admin-ip-second-order-sqli-facts/facts.json
outputs:
- units.jsonl
- summary.json
budget:
usd: 0.07
calls: 16
- name: froxlor-admin-ip-second-order-sqli-vb
task: verify-b-froxlor-admin-ip-second-order-sqli
inputs:
facts: froxlor-admin-ip-second-order-sqli-facts/facts.json
outputs:
- units.jsonl
- summary.json
budget:
usd: 0.07
calls: 16
- name: froxlor-admin-ip-second-order-sqli-agree
task: agree-froxlor-admin-ip-second-order-sqli
inputs:
pass_a: froxlor-admin-ip-second-order-sqli-va/units.jsonl
pass_b: froxlor-admin-ip-second-order-sqli-vb/units.jsonl
outputs:
- agreed.json
- disputed.json
budget: *id004
- name: fuxa-tdengine-escape-sqli-facts
task: repo-facts-fuxa-tdengine-escape-sqli
outputs:
- facts.json
budget: &id005
usd: 0
calls: 0
serialize: repo-facts
- name: fuxa-tdengine-escape-sqli-va
task: verify-a-fuxa-tdengine-escape-sqli
inputs:
facts: fuxa-tdengine-escape-sqli-facts/facts.json
outputs:
- units.jsonl
- summary.json
budget:
usd: 0.12
calls: 8
- name: fuxa-tdengine-escape-sqli-vb
task: verify-b-fuxa-tdengine-escape-sqli
inputs:
facts: fuxa-tdengine-escape-sqli-facts/facts.json
outputs:
- units.jsonl
- summary.json
budget:
usd: 0.12
calls: 8
- name: fuxa-tdengine-escape-sqli-agree
task: agree-fuxa-tdengine-escape-sqli
inputs:
pass_a: fuxa-tdengine-escape-sqli-va/units.jsonl
pass_b: fuxa-tdengine-escape-sqli-vb/units.jsonl
outputs:
- agreed.json
- disputed.json
budget: *id005
- name: lmdeploy-media-url-ssrf-facts
task: repo-facts-lmdeploy-media-url-ssrf
outputs:
- facts.json
budget: &id006
usd: 0
calls: 0
serialize: repo-facts
- name: lmdeploy-media-url-ssrf-va
task: verify-a-lmdeploy-media-url-ssrf
inputs:
facts: lmdeploy-media-url-ssrf-facts/facts.json
outputs:
- units.jsonl
- summary.json
budget:
usd: 0.02
calls: 8
- name: lmdeploy-media-url-ssrf-vb
task: verify-b-lmdeploy-media-url-ssrf
inputs:
facts: lmdeploy-media-url-ssrf-facts/facts.json
outputs:
- units.jsonl
- summary.json
budget:
usd: 0.02
calls: 8
- name: lmdeploy-media-url-ssrf-agree
task: agree-lmdeploy-media-url-ssrf
inputs:
pass_a: lmdeploy-media-url-ssrf-va/units.jsonl
pass_b: lmdeploy-media-url-ssrf-vb/units.jsonl
outputs:
- agreed.json
- disputed.json
budget: *id006
- name: medplum-external-auth-open-redirect-facts
task: repo-facts-medplum-external-auth-open-redirect
outputs:
- facts.json
budget: &id007
usd: 0
calls: 0
serialize: repo-facts
- name: medplum-external-auth-open-redirect-va
task: verify-a-medplum-external-auth-open-redirect
inputs:
facts: medplum-external-auth-open-redirect-facts/facts.json
outputs:
- units.jsonl
- summary.json
budget:
usd: 0.02
calls: 8
- name: medplum-external-auth-open-redirect-vb
task: verify-b-medplum-external-auth-open-redirect
inputs:
facts: medplum-external-auth-open-redirect-facts/facts.json
outputs:
- units.jsonl
- summary.json
budget:
usd: 0.02
calls: 8
- name: medplum-external-auth-open-redirect-agree
task: agree-medplum-external-auth-open-redirect
inputs:
pass_a: medplum-external-auth-open-redirect-va/units.jsonl
pass_b: medplum-external-auth-open-redirect-vb/units.jsonl
outputs:
- agreed.json
- disputed.json
budget: *id007
- name: pgadmin-maintenance-sqli-facts
task: repo-facts-pgadmin-maintenance-sqli
outputs:
- facts.json
budget: &id008
usd: 0
calls: 0
serialize: repo-facts
- name: pgadmin-maintenance-sqli-va
task: verify-a-pgadmin-maintenance-sqli
inputs:
facts: pgadmin-maintenance-sqli-facts/facts.json
outputs:
- units.jsonl
- summary.json
budget:
usd: 0.45
calls: 96
- name: pgadmin-maintenance-sqli-vb
task: verify-b-pgadmin-maintenance-sqli
inputs:
facts: pgadmin-maintenance-sqli-facts/facts.json
outputs:
- units.jsonl
- summary.json
budget:
usd: 0.45
calls: 96
- name: pgadmin-maintenance-sqli-agree
task: agree-pgadmin-maintenance-sqli
inputs:
pass_a: pgadmin-maintenance-sqli-va/units.jsonl
pass_b: pgadmin-maintenance-sqli-vb/units.jsonl
outputs:
- agreed.json
- disputed.json
budget: *id008
- name: qwed-sympy-eval-facts
task: repo-facts-qwed-sympy-eval
outputs:
- facts.json
budget: &id009
usd: 0
calls: 0
serialize: repo-facts
- name: qwed-sympy-eval-va
task: verify-a-qwed-sympy-eval
inputs:
facts: qwed-sympy-eval-facts/facts.json
outputs:
- units.jsonl
- summary.json
budget:
usd: 0.02
calls: 8
- name: qwed-sympy-eval-vb
task: verify-b-qwed-sympy-eval
inputs:
facts: qwed-sympy-eval-facts/facts.json
outputs:
- units.jsonl
- summary.json
budget:
usd: 0.02
calls: 8
- name: qwed-sympy-eval-agree
task: agree-qwed-sympy-eval
inputs:
pass_a: qwed-sympy-eval-va/units.jsonl
pass_b: qwed-sympy-eval-vb/units.jsonl
outputs:
- agreed.json
- disputed.json
budget: *id009
- name: reactpress-findall-column-sqli-facts
task: repo-facts-reactpress-findall-column-sqli
outputs:
- facts.json
budget: &id010
usd: 0
calls: 0
serialize: repo-facts
- name: reactpress-findall-column-sqli-va
task: verify-a-reactpress-findall-column-sqli
inputs:
facts: reactpress-findall-column-sqli-facts/facts.json
outputs:
- units.jsonl
- summary.json
budget:
usd: 0.02
calls: 8
- name: reactpress-findall-column-sqli-vb
task: verify-b-reactpress-findall-column-sqli
inputs:
facts: reactpress-findall-column-sqli-facts/facts.json
outputs:
- units.jsonl
- summary.json
budget:
usd: 0.02
calls: 8
- name: reactpress-findall-column-sqli-agree
task: agree-reactpress-findall-column-sqli
inputs:
pass_a: reactpress-findall-column-sqli-va/units.jsonl
pass_b: reactpress-findall-column-sqli-vb/units.jsonl
outputs:
- agreed.json
- disputed.json
budget: *id010
- name: unleash-addon-webhook-ssrf-facts
task: repo-facts-unleash-addon-webhook-ssrf
outputs:
- facts.json
budget: &id011
usd: 0
calls: 0
serialize: repo-facts
- name: unleash-addon-webhook-ssrf-va
task: verify-a-unleash-addon-webhook-ssrf
inputs:
facts: unleash-addon-webhook-ssrf-facts/facts.json
outputs:
- units.jsonl
- summary.json
budget:
usd: 0.03
calls: 8
- name: unleash-addon-webhook-ssrf-vb
task: verify-b-unleash-addon-webhook-ssrf
inputs:
facts: unleash-addon-webhook-ssrf-facts/facts.json
outputs:
- units.jsonl
- summary.json
budget:
usd: 0.03
calls: 8
- name: unleash-addon-webhook-ssrf-agree
task: agree-unleash-addon-webhook-ssrf
inputs:
pass_a: unleash-addon-webhook-ssrf-va/units.jsonl
pass_b: unleash-addon-webhook-ssrf-vb/units.jsonl
outputs:
- agreed.json
- disputed.json
budget: *id011
- name: winml-cli-serve-cors-facts
task: repo-facts-winml-cli-serve-cors
outputs:
- facts.json
budget: &id012
usd: 0
calls: 0
serialize: repo-facts
- name: winml-cli-serve-cors-va
task: verify-a-winml-cli-serve-cors
inputs:
facts: winml-cli-serve-cors-facts/facts.json
outputs:
- units.jsonl
- summary.json
budget:
usd: 0.02
calls: 8
- name: winml-cli-serve-cors-vb
task: verify-b-winml-cli-serve-cors
inputs:
facts: winml-cli-serve-cors-facts/facts.json
outputs:
- units.jsonl
- summary.json
budget:
usd: 0.02
calls: 8
- name: winml-cli-serve-cors-agree
task: agree-winml-cli-serve-cors
inputs:
pass_a: winml-cli-serve-cors-va/units.jsonl
pass_b: winml-cli-serve-cors-vb/units.jsonl
outputs:
- agreed.json
- disputed.json
budget: *id012
- name: wp-directorist-avatar-ssrf-facts
task: repo-facts-wp-directorist-avatar-ssrf
outputs:
- facts.json
budget: &id013
usd: 0
calls: 0
serialize: repo-facts
- name: wp-directorist-avatar-ssrf-va
task: verify-a-wp-directorist-avatar-ssrf
inputs:
facts: wp-directorist-avatar-ssrf-facts/facts.json
outputs:
- units.jsonl
- summary.json
budget:
usd: 0.03
calls: 8
- name: wp-directorist-avatar-ssrf-vb
task: verify-b-wp-directorist-avatar-ssrf
inputs:
facts: wp-directorist-avatar-ssrf-facts/facts.json
outputs:
- units.jsonl
- summary.json
budget:
usd: 0.03
calls: 8
- name: wp-directorist-avatar-ssrf-agree
task: agree-wp-directorist-avatar-ssrf
inputs:
pass_a: wp-directorist-avatar-ssrf-va/units.jsonl
pass_b: wp-directorist-avatar-ssrf-vb/units.jsonl
outputs:
- agreed.json
- disputed.json
budget: *id013
- name: wp-gopay-log-filter-sqli-facts
task: repo-facts-wp-gopay-log-filter-sqli
outputs:
- facts.json
budget: &id014
usd: 0
calls: 0
serialize: repo-facts
- name: wp-gopay-log-filter-sqli-va
task: verify-a-wp-gopay-log-filter-sqli
inputs:
facts: wp-gopay-log-filter-sqli-facts/facts.json
outputs:
- units.jsonl
- summary.json
budget:
usd: 0.02
calls: 8
- name: wp-gopay-log-filter-sqli-vb
task: verify-b-wp-gopay-log-filter-sqli
inputs:
facts: wp-gopay-log-filter-sqli-facts/facts.json
outputs:
- units.jsonl
- summary.json
budget:
usd: 0.02
calls: 8
- name: wp-gopay-log-filter-sqli-agree
task: agree-wp-gopay-log-filter-sqli
inputs:
pass_a: wp-gopay-log-filter-sqli-va/units.jsonl
pass_b: wp-gopay-log-filter-sqli-vb/units.jsonl
outputs:
- agreed.json
- disputed.json
budget: *id014
- name: wp-groundhogg-confirm-redirect-facts
task: repo-facts-wp-groundhogg-confirm-redirect
outputs:
- facts.json
budget: &id015
usd: 0
calls: 0
serialize: repo-facts
- name: wp-groundhogg-confirm-redirect-va
task: verify-a-wp-groundhogg-confirm-redirect
inputs:
facts: wp-groundhogg-confirm-redirect-facts/facts.json
outputs:
- units.jsonl
- summary.json
budget:
usd: 0.05
calls: 8
- name: wp-groundhogg-confirm-redirect-vb
task: verify-b-wp-groundhogg-confirm-redirect
inputs:
facts: wp-groundhogg-confirm-redirect-facts/facts.json
outputs:
- units.jsonl
- summary.json
budget:
usd: 0.05
calls: 8
- name: wp-groundhogg-confirm-redirect-agree
task: agree-wp-groundhogg-confirm-redirect
inputs:
pass_a: wp-groundhogg-confirm-redirect-va/units.jsonl
pass_b: wp-groundhogg-confirm-redirect-vb/units.jsonl
outputs:
- agreed.json
- disputed.json
budget: *id015