MCPcopy Create free account

hub / github.com/NVIDIA/SkillEvaluator / functions

Functions24,741 in github.com/NVIDIA/SkillEvaluator

↓ 1 callersFunction_safe_id
(raw: str)
src/skillevaluator/tier3/plugin_prepare.py:829
↓ 1 callersFunction_safe_path
(value)
ci/feedback.py:110
↓ 1 callersFunction_safe_repo_context_rel
(path: Path, root: Path)
src/skillevaluator/tier3/harbor/adapter.py:1126
↓ 1 callersFunction_safe_skip_reason_text
Normalize one untrusted reason without changing identity-safe text rules.
src/skillevaluator/reporting/base.py:465
↓ 1 callersFunction_safe_terminal_text
(value: str)
src/skillevaluator/model_commands.py:75
↓ 1 callersFunction_safe_trial_source_component
Return an exact safe child name without applying output normalization.
src/skillevaluator/tier3/harbor/collector.py:5895
↓ 1 callersFunction_safe_validation_location
(location: tuple[Any, ...])
src/skillevaluator/tier3/external_metrics.py:153
↓ 1 callersMethod_safety_severity
Return Safety's severity: a level word, else its CVSS object, else medium. Safety prints null without an API key and, with one, an object suc
src/skillevaluator/validators/dependencies.py:972
↓ 1 callersFunction_salvage_behavior_results
Recover complete per-behavior entries from a truncated ``results`` array. Reasoning judges that hit the output-token cap emit ``{"results": [...`
src/skillevaluator/tier3/eval_core/llm_judge.py:455
↓ 1 callersFunction_salvage_behavior_results
Recover complete per-behavior entries from a truncated ``results`` array. Reasoning judges that hit the output-token cap emit ``{"results": [...`
src/skillevaluator/tier3/harbor/templates/eval.py:2752
↓ 1 callersFunction_same_location_variants
Return True when two refs cast one location and one body of text in different roles.
src/skillevaluator/tier3/harbor/report.py:721
↓ 1 callersFunction_same_skill_revision
(current: dict[str, Any], prior: dict[str, Any])
src/skillevaluator/tier3/version_comparison.py:47
↓ 1 callersFunction_sanitize_analyzer_revision
(revision: object)
src/skillevaluator/skillspector_provenance.py:229
↓ 1 callersFunction_sanitize_control
(name: str, value: object, *, expected_adapter_support: bool)
src/skillevaluator/skillspector_provenance.py:121
↓ 1 callersFunction_sanitize_error_value
(value, extra_secret_values=())
src/skillevaluator/tier3/harbor/templates/eval.py:1196
↓ 1 callersFunction_sanitize_items_with_evidence
(raw: object, *, evidence_keys: tuple[str, ...])
src/skillevaluator/reporting/harbor_viewer.py:215
↓ 1 callersFunction_sanitize_json_numbers
Copy a payload while replacing numbers browsers cannot represent safely.
src/skillevaluator/evaluation/tier3_report.py:159
↓ 1 callersFunction_sanitize_routing
(raw_routing: Mapping, *, resolved_adapter: str)
src/skillevaluator/skillspector_provenance.py:178
↓ 1 callersFunction_sanitize_rule_component
Return a SARIF-safe rule identifier fragment.
src/skillevaluator/reporting/sarif_reporter.py:33
↓ 1 callersFunction_sanitize_tier3_case_diagnostics
(value: object)
src/skillevaluator/reporting/html.py:1317
↓ 1 callersFunction_sanitize_tier3_comparison_basis
Keep an evaluator card's judge-model lift exclusions only when they name a case count.
src/skillevaluator/reporting/html.py:939
↓ 1 callersFunction_sanitize_tier3_dataset_case
(value: dict[str, Any])
src/skillevaluator/reporting/html.py:1309
↓ 1 callersFunction_sanitize_tier3_dimension_comparison
Project recorded native comparisons without recomputing or trusting shaped values.
src/skillevaluator/reporting/html.py:584
↓ 1 callersFunction_sanitize_tier3_harbor_viewer
(value: object)
src/skillevaluator/reporting/html.py:1290
↓ 1 callersFunction_sanitize_tier3_insight_items
(value: object)
src/skillevaluator/reporting/html.py:1299
↓ 1 callersFunction_sanitize_tier3_mcnemar_exact
(value: object)
src/skillevaluator/reporting/html.py:795
↓ 1 callersFunction_sanitize_tier3_paired_comparison
(value: object)
src/skillevaluator/reporting/html.py:858
↓ 1 callersFunction_sanitize_tier3_pass_condition
(value: object)
src/skillevaluator/reporting/html.py:662
↓ 1 callersFunction_sanitize_tier3_probability_fields
( value: dict[str, Any], *, numeric_key: str, text_key: str, underflow_key: str, )
src/skillevaluator/reporting/html.py:750
↓ 1 callersFunction_sanitize_tier3_probability_text
(value: object)
src/skillevaluator/reporting/html.py:725
↓ 1 callersFunction_sanitize_value
(value, paths: tuple[Path | None, ...])
src/skillevaluator/utils/tier2_paths.py:143
↓ 1 callersFunction_save_current_stage
()
src/skillevaluator/tier3/harbor/adapter.py:2616
↓ 1 callersFunction_save_state
Persist state to disk in markdown format, returning error if any.
src/skillevaluator/tier3/reference_skills/task-list/scripts/tasks.py:79
↓ 1 callersFunction_saved_case_fixes
The run's ``what_to_fix.json`` when the report can show it as is, else ``None``.
src/skillevaluator/evaluation/tier3_report.py:4289
↓ 1 callersFunction_says_missing_tool
Whether a failed result says the called tool (or skill, command, agent) does not exist.
src/skillevaluator/tier3/eval_core/plugin_signals.py:3380
↓ 1 callersFunction_scalar
(raw: object)
src/skillevaluator/skillspector_provenance.py:135
↓ 1 callersFunction_scale_timeout_values
(data: dict[str, Any], multiplier: float)
src/skillevaluator/tier3/nemo_gym/runner.py:285
↓ 1 callersFunction_scaled_task_timeouts
Scale a staged task.toml's timeouts by Harbor's multiplier, checking nothing else changed.
src/skillevaluator/tier3/nemo_gym/runner.py:162
↓ 1 callersFunction_scan_descriptor_tree
( source: Path, descriptor: int, parts: tuple[str, ...], root_device: int, entries: list[_
src/skillevaluator/publication_identity.py:312
↓ 1 callersFunction_scan_descriptor_tree
( *, source: Path, descriptor: int, parts: tuple[str, ...], root_device: int, role: st
src/skillevaluator/tier3/harbor/secure_copy.py:407
↓ 1 callersFunction_scan_env_entries
Return child names, remaining entry budget, and whether every child was inspectable.
src/skillevaluator/plugin_member_paths.py:363
↓ 1 callersFunction_scan_fallback_tree
( source: Path, parts: tuple[str, ...], root_device: int, entries: list[_PublicationEntry],
src/skillevaluator/publication_identity.py:491
↓ 1 callersFunction_scan_fallback_tree
( source: Path, *, parts: tuple[str, ...], root_device: int, role: str, ignore: Ignore
src/skillevaluator/tier3/harbor/secure_copy.py:583
↓ 1 callersMethod_scan_file
Scan a single file for invisible Unicode characters.
src/skillevaluator/validators/unicode_smuggle.py:169
↓ 1 callersMethod_scan_for_pii
Scan files for PII using regex patterns (emails, paths, SSNs, etc.).
src/skillevaluator/validators/security.py:5814
↓ 1 callersMethod_scan_lines_for_pattern
Check all lines against a single compiled PII pattern. Only the first remaining match on a line is reported. Matches are filtered bef
src/skillevaluator/validators/security.py:6554
↓ 1 callersMethod_scan_source_headers
Tier 3: Scan source files for SPDX-License-Identifier headers. Files under Tier 1 artifact directories (``evals/``, ``results/``, ``v
src/skillevaluator/validators/license.py:637
↓ 1 callersMethod_scan_truncated
Record that ``source`` has handlers past the scan limits: a ``truncated`` row and a blocking finding.
src/skillevaluator/plugin_component_risk.py:3545
↓ 1 callersFunction_scan_units
(skills: list[str], excluded: frozenset[str])
src/skillevaluator/tier1/plugin_checks.py:146
↓ 1 callersFunction_scanner_env
A strict process allowlist: no auth, proxies, host configs or contexts.
src/skillevaluator/plugin_dependency_audit.py:420
↓ 1 callersFunction_scanner_row
Report one scanner's resolved executable and version without a shell.
src/skillevaluator/doctor.py:68
↓ 1 callersFunction_schema_validator_for
Return the schema validator matching the (forced or detected) content type. Rules, workflows, and plugins use their dedicated schema validators;
src/skillevaluator/tier1/commands.py:280
↓ 1 callersFunction_scope
(result: ValidationResult)
src/skillevaluator/reporting/tier2_views.py:40
↓ 1 callersMethod_scope
What an approval hook can approve: ``all``, ``bash``, ``scoped`` (write, fetch, or MCP tools, listed), or ``narrow``; ``None`` for an event th
src/skillevaluator/plugin_component_risk.py:3500
↓ 1 callersFunction_scope_events
The groups and compound openers of one segment, in the order they open. ``{ ( for f in x`` opens a brace group, a subshell group and a loop, one
src/skillevaluator/tier3/eval_core/script_invocation.py:1944
↓ 1 callersFunction_scope_harness_call_ids
Keep independent step sessions' reused call IDs distinct in the merged document.
src/skillevaluator/tier3/plugin_trace_artifacts.py:470
↓ 1 callersFunction_scope_rows
(scope: Mapping, *, case: str, latency: bool, runtime: Mapping, text: _Text)
src/skillevaluator/reporting/telemetry_view.py:120
↓ 1 callersFunction_scoped_targets
(fake_mcp: str)
tests/tier3/test_mcp_proof.py:646
↓ 1 callersFunction_score
(value: float | None)
src/skillevaluator/reporting/external_metrics.py:21
↓ 1 callersFunction_score_band
Return the dimension verdict band: 0 for FAIL, 1 for NEUTRAL, 2 for PASS.
src/skillevaluator/reporting/benchmark.py:1913
↓ 1 callersFunction_score_bar
(score: float)
src/skillevaluator/tier3/result_display.py:70
↓ 1 callersMethod_score_cell
Format a dimension score cell with color coding.
src/skillevaluator/reporting/cli.py:336
↓ 1 callersFunction_score_definition_for_policy
(metrics: tuple[str, ...], score_policy: str)
src/skillevaluator/evaluation/tier3_report.py:2025
↓ 1 callersFunction_score_from_txt
()
src/skillevaluator/tier3/harbor/templates/custom_grader_runner.py:225
↓ 1 callersFunction_score_mapping
( value: object, *, minimum: float = 0.0, maximum: float = 1.0, )
src/skillevaluator/reporting/html.py:481
↓ 1 callersFunction_score_mapping_is_valid
(value: object)
src/skillevaluator/tier3/results_location.py:290
↓ 1 callersFunction_score_policy_for_rewards
(rewards: list[dict], metrics: list[str])
src/skillevaluator/tier3/harbor/templates/metric.py:56
↓ 1 callersFunction_score_style
(score: float)
src/skillevaluator/tier3/commands.py:1851
↓ 1 callersFunction_score_style
(score: float)
src/skillevaluator/tier3/result_display.py:62
↓ 1 callersFunction_score_to_grade
Convert numeric score to letter grade (module-level helper for folder aggregation).
src/skillevaluator/validators/quality_score.py:1001
↓ 1 callersFunction_score_transition_problem
Return why a score transition is malformed or inconsistent, or ``None`` when it is exact.
scripts/ci/check_public_benchmarks.py:882
↓ 1 callersFunction_scored_metric
(reward: Mapping[str, Any], metric_name: str)
src/skillevaluator/tier3/attempt_verdict.py:57
↓ 1 callersFunction_scored_pass_flags
(trials: Sequence[TrialObservation], pass_threshold: float)
src/skillevaluator/tier3/harbor/stats.py:1443
↓ 1 callersFunction_scored_trial
( job: Path, name: str, case: str, scores: dict[str, float] | None = None, details: dict[str, Any] | None
tests/tier3/test_baseline_timeouts.py:108
↓ 1 callersFunction_scored_trials
(agent: Mapping[str, Any], key: str)
src/skillevaluator/tier3/case_summary.py:296
↓ 1 callersFunction_screen_rules
Every screen style rule in document order, as (selector, declarations).
tests/reporting/test_report_redesign.py:541
↓ 1 callersMethod_script_evidence
Facts of the plugin scripts a command runs (and of the scripts those name, in turn), why any of them was not analyzed, and whether any script
src/skillevaluator/plugin_component_risk.py:3253
↓ 1 callersFunction_script_exit_status
(call: dict[str, Any], script: str)
src/skillevaluator/tier3/eval_core/declarative_checks.py:90
↓ 1 callersMethod_script_paths
Root-relative paths of the plugin script a command token may name (none when it escapes the root or still names a variable).
src/skillevaluator/plugin_component_risk.py:3125
↓ 1 callersMethod_script_references
Plugin scripts a script runs: named by root placeholder, relative to its own directory, or (in shell and Python scripts) by a script file name
src/skillevaluator/plugin_component_risk.py:3312
↓ 1 callersFunction_script_safe_json
Serialize JSON for an HTML raw-text script element without expansion attacks.
src/skillevaluator/reporting/html.py:1613
↓ 1 callersMethod_scripts
Preserve referenced resources and their transitive dependencies under the same root.
src/skillevaluator/tier3/loose_parts.py:1138
↓ 1 callersFunction_scrub
Final pass over every string, including names and statuses that were not shortened. Key-based redaction would also blank measured ``tokens`` tota
src/skillevaluator/evaluation/tier3_feedback_chat.py:940
↓ 1 callersFunction_scrub_untrusted_tier3_agents
Retain agent identity/coverage while removing untrusted score evidence.
src/skillevaluator/reporting/html.py:1202
↓ 1 callersFunction_sdk_targets_native_openai
(client: Any)
src/skillevaluator/inference/client.py:109
↓ 1 callersFunction_sdk_transport_class
Return the transport class behind the OpenAI SDK's default async client, or fail the test. The stub answers httpx requests only. If the SDK build
tests/tier3/test_ragas_judge_evidence.py:200
↓ 1 callersMethod_search_cue
Return whether *pattern* matches any form of the cue text.
src/skillevaluator/validators/security.py:6238
↓ 1 callersFunction_seatbelt_argv
Build the macOS Seatbelt launcher argv. Reads: deny host HOME except canonical run/runtime paths. Writes: confined to the run dirs. Network:
src/skillevaluator/tier3/harbor/local_sandbox.py:797
↓ 1 callersFunction_seatbelt_available
()
src/skillevaluator/tier3/harbor/local_sandbox.py:498
↓ 1 callersFunction_seatbelt_home_read_rule
Deny host-HOME reads while allowing metadata-only path traversal.
src/skillevaluator/tier3/harbor/local_sandbox.py:739
↓ 1 callersFunction_seatbelt_path_filter
Return the narrow SBPL filter for one existing canonical path.
src/skillevaluator/tier3/harbor/local_sandbox.py:681
↓ 1 callersFunction_seatbelt_smoke_test
()
src/skillevaluator/tier3/harbor/local_sandbox.py:502
↓ 1 callersFunction_seconds
(milliseconds: float | None)
src/skillevaluator/tier3/harbor/trajectory_timing.py:90
↓ 1 callersFunction_secret_exposure_finding
( observation: str, *, tool: str | None, target_skill_used_before: bool | None, )
src/skillevaluator/tier3/eval_core/checks.py:5312
↓ 1 callersFunction_secret_reference_name_error
(value: Any, *, label: str)
src/skillevaluator/tier3/evals_config.py:614
↓ 1 callersFunction_sections
(raw: object, *, latency: bool)
src/skillevaluator/reporting/telemetry_view.py:168
↓ 1 callersMethod_secure_compose_exec
Run a sidecar command without exposing values or main-only state.
src/skillevaluator/tier3/harbor/secure_docker_environment.py:2936
↓ 1 callersFunction_security_alias
(path)
src/skillevaluator/tier3/eval_core/plugin_security.py:2444
↓ 1 callersFunction_security_chain
``steps`` match in order inside one segment of ``text`` between ``breaks``, each after the one before. Each segment is read once, from its first
src/skillevaluator/tier3/eval_core/plugin_security.py:3268
↓ 1 callersFunction_security_change_directory
The directory after a top-level ``cd``/``pushd`` in ``unit`` (``cd`` alone goes home).
src/skillevaluator/tier3/eval_core/plugin_security.py:3011
← previousnext →7,601–7,700 of 24,741, ranked by callers