MCPcopy Create free account

hub / github.com/NVIDIA/SkillEvaluator / functions

Functions24,741 in github.com/NVIDIA/SkillEvaluator

↓ 1 callersFunction_private_environment_labels
Return imported non-public environment labels that must not escape in free text.
src/skillevaluator/reporting/benchmark.py:2188
↓ 1 callersFunction_private_reward_inputs
Known verifier layouts of a bounded, explicitly selected set of regular trial roots.
src/skillevaluator/tier3/harbor/collector.py:3622
↓ 1 callersFunction_probability_exact
Return a bounded reduced rational, or None when decimal rendering is unsafe.
src/skillevaluator/tier3/harbor/collector.py:4929
↓ 1 callersFunction_probability_text_matches_number
(probability: Decimal, number: float)
src/skillevaluator/reporting/html.py:740
↓ 1 callersFunction_probe_bedrock_model
Run one Bedrock catalog request inside the outer deadline worker.
src/skillevaluator/tier3/harbor/runtime_preflight.py:1282
↓ 1 callersFunction_probe_bedrock_model_with_deadline
Bound session creation, credential providers, retries, and request I/O.
src/skillevaluator/tier3/harbor/runtime_preflight.py:1376
↓ 1 callersFunction_probe_chat
(http_client: Any, timeout: float)
src/skillevaluator/doctor.py:151
↓ 1 callersFunction_probe_embedding
(http_client: Any, timeout: float)
src/skillevaluator/doctor.py:134
↓ 1 callersFunction_probe_manifest_files
Every supported manifest file in ``root``, probed without following links. Only root entries that can name a manifest (any letter case of a clien
src/skillevaluator/plugin_manifest.py:461
↓ 1 callersFunction_probe_run_options
Launch a version probe with the environment the scanner's validator uses.
src/skillevaluator/doctor.py:55
↓ 1 callersFunction_probe_runtime_version
Run a selected CLI's version command in a credential-free temp HOME.
src/skillevaluator/tier3/harbor/local_runtime.py:590
↓ 1 callersFunction_probe_session
( url: str, kind: str, headers: dict[str, str], *, addresses: tuple[str, ...], total_timeout: float )
src/skillevaluator/tier3/mcp_proof.py:625
↓ 1 callersFunction_probe_vertex_anthropic_model_with_deadline
Bound Vertex AI credential acquisition and rawPredict I/O within a single deadline.
src/skillevaluator/tier3/harbor/runtime_preflight.py:1795
↓ 1 callersFunction_probe_vertex_openapi_model_with_deadline
Bound Vertex AI OpenAPI credential acquisition and chat/completions I/O within a single deadline.
src/skillevaluator/tier3/harbor/runtime_preflight.py:1876
↓ 1 callersMethod_process_bandit_output
Parse Bandit JSON output and report issues.
src/skillevaluator/validators/code_risk.py:208
↓ 1 callersMethod_process_safety
Parse Safety output and report additional findings. ``reported`` holds the pip-audit advisories of the same file as ``(package, versi
src/skillevaluator/validators/dependencies.py:771
↓ 1 callersFunction_produced_tier3_result
Build Tier 3 evidence through the real payload producer.
tests/reporting/test_advisory_skip_verdict.py:2359
↓ 1 callersFunction_program_runs_stdin
Whether a program passed as an argument evaluates its standard input (the download). True for code that evaluates text it reads from standard inp
src/skillevaluator/plugin_component_risk.py:2430
↓ 1 callersFunction_projection_includes
( relative: tuple[str, ...], *, exact_paths: frozenset[tuple[str, ...]], whole_roots: frozense
src/skillevaluator/tier3/harbor/adapter.py:6652
↓ 1 callersFunction_prompt_skill_files_re
A prompt that points the agent at the skill's own files, which the no-skill baseline is never given.
src/skillevaluator/tier3/evals_spec.py:738
↓ 1 callersFunction_prompt_texts
The task prompt: the user and system messages before the agent's first step. Every harness puts the task (and context such as ``AGENTS.md``) in f
src/skillevaluator/tier3/eval_core/plugin_signals.py:3388
↓ 1 callersFunction_proof_census
Only confirmed harness states can contradict the result of a host probe.
src/skillevaluator/tier3/plugin_evidence.py:212
↓ 1 callersFunction_protect_key_for_storage
Bind key bytes to the current Windows user with DPAPI.
src/skillevaluator/tier3/output_provenance.py:253
↓ 1 callersMethod_protected_main_environment_values
(self)
src/skillevaluator/tier3/harbor/secure_docker_environment.py:2228
↓ 1 callersMethod_prove_candidate
(self, emitted: list[str], discard_re: re.Pattern[str])
src/skillevaluator/tier3/harbor/stream_redaction.py:407
↓ 1 callersFunction_provider
()
src/skillevaluator/tier3/eval_core/llm_judge.py:93
↓ 1 callersFunction_provider_id
(value: object)
src/skillevaluator/skillspector_provenance.py:422
↓ 1 callersFunction_provider_log_rejection
Return why a timed-out trial's logs cannot clear it of provider trouble, or None when they do. ``provider_degradation_observed`` when any text lo
src/skillevaluator/tier3/harbor/collector.py:7348
↓ 1 callersFunction_provider_url
(base_url: str, *, ensure_v1: bool = False)
src/skillevaluator/model_catalog.py:564
↓ 1 callersFunction_prune_declarative_check_rows
Cap per-check coverage rows when the payload is over its limits; counts stay exact. Coverage grows with every check of every reward, so it goes b
src/skillevaluator/evaluation/tier3_report.py:1586
↓ 1 callersFunction_prune_history
Bound deployed files while preserving current main and open-PR links. Git history is not rewritten. Shallow workflow checkouts download only the
ci/publish_reports.py:836
↓ 1 callersFunction_prune_non_best_agent_details
Drop duplicated lower-priority details before touching best-agent evidence.
src/skillevaluator/evaluation/tier3_report.py:1518
↓ 1 callersFunction_prune_pass_at_k_pairing_diagnostics
Bound legacy full mismatch-ID arrays while preserving counts and samples.
src/skillevaluator/evaluation/tier3_report.py:1546
↓ 1 callersFunction_pruned_folder_findings
Say which shipped folders Tier 1 does not scan, and refuse links out of the plugin inside them.
src/skillevaluator/plugin_paths.py:620
↓ 1 callersFunction_public_provider_error
()
src/skillevaluator/tier3/harbor/templates/eval.py:1487
↓ 1 callersFunction_public_provider_fix
The problem and actions for the provider behind ``SKILL_EVAL_LLM_*``: a missing key or a failed call.
src/skillevaluator/reporting/console_ui.py:1406
↓ 1 callersFunction_publication_safe_location
(finding: Finding)
src/skillevaluator/reporting/benchmark.py:2413
↓ 1 callersFunction_publication_safe_skill_name
Return an exact NFC target identity or a non-injectable fallback.
src/skillevaluator/reporting/benchmark.py:2178
↓ 1 callersFunction_publish
( tmp_path: Path, baseline_not_applicable: tuple[str, ...], *, judge_fallback: bool = False,
tests/test_overall_cross_format_consistency.py:78
↓ 1 callersFunction_publish
(tmp_path: Path, monkeypatch, reports: Path)
tests/ci/test_secret_redaction.py:631
↓ 1 callersFunction_published_job_failure
Return one validation/launch failure safe for generated results.
src/skillevaluator/tier3/harbor/collector.py:343
↓ 1 callersFunction_published_trial_labels_by_root
Assign one stable, distinct published identity to each physical trial root.
src/skillevaluator/tier3/harbor/collector.py:379
↓ 1 callersFunction_pull_identity
(pull)
ci/benchmark_publish.py:91
↓ 1 callersFunction_python_dotenv_prerequisite_error
Require the python-dotenv release that honors ``PYTHON_DOTENV_DISABLED``.
src/skillevaluator/tier3/harbor/runner.py:2556
↓ 1 callersFunction_python_grader_template
()
src/skillevaluator/tier3/commands.py:222
↓ 1 callersFunction_python_import_paths
Files of the modules a Python script imports, other than the standard library's. ``from core.rule_engine import RuleEngine`` names ``core/rule_en
src/skillevaluator/plugin_component_risk.py:2976
↓ 1 callersFunction_python_reconstructions
( text: str, )
scripts/check_oss_boundary.py:376
↓ 1 callersFunction_python_source
(inputs: AuditInputs, source: str, text: str, *, pyproject: bool)
src/skillevaluator/plugin_dependency_audit.py:150
↓ 1 callersFunction_qualified_label_key_prefix
Return the prefix of a well-formed ``prefix/name`` Kubernetes label key, or None.
src/skillevaluator/tier3/harbor/gke_environment.py:727
↓ 1 callersFunction_quality_config_label
Return the Tier 1 quality config line, or ``None`` when no QUALITY result recorded one.
src/skillevaluator/reporting/benchmark.py:1589
↓ 1 callersFunction_quality_min_score
Return the Tier 1 quality pass threshold, or ``None`` when no QUALITY result recorded one.
src/skillevaluator/reporting/benchmark.py:1609
↓ 1 callersFunction_quality_result
(payload: str)
tests/reporting/test_rich_markup_escaping.py:105
↓ 1 callersFunction_quality_score
(results: Sequence[ValidationResult])
src/skillevaluator/precommit.py:139
↓ 1 callersFunction_quality_stats
(results: list[ValidationResult])
src/skillevaluator/reporting/console_ui.py:1153
↓ 1 callersMethod_quantifier
(self)
src/skillevaluator/plugin_component_risk.py:1290
↓ 1 callersFunction_question_matches_skill_domain
Return True when an author-provided negative still looks on-skill for this skill.
src/skillevaluator/tier3/generate_dataset.py:372
↓ 1 callersFunction_quiet_probe_program
(words: list[str])
src/skillevaluator/tier3/eval_core/checks.py:7486
↓ 1 callersFunction_quote_lenient_values
Quote the plain frontmatter values strict YAML rejects, the way Claude Code retries a failed parse. Claude Code 2.1.x parses frontmatter with a Y
src/skillevaluator/plugin_markdown.py:38
↓ 1 callersFunction_quoted_directory_variants
Return the quoted top-level directory name of a nested file, such as `` `policies` ``. Guidance names a directory as code or a string ("Load the
src/skillevaluator/validators/skillspector_scope.py:687
↓ 1 callersFunction_ragas_evidence_http_client
Return the SDK's default async HTTP client with the evidence hook, or None to score without evidence.
src/skillevaluator/tier3/harbor/templates/eval.py:3549
↓ 1 callersFunction_ragas_message_text
(message)
src/skillevaluator/tier3/harbor/templates/eval.py:3475
↓ 1 callersFunction_ragas_prompt
Return a RAGAS prompt that holds a Bash call running *command*, as the judge sends it.
tests/tier3/test_security_finding_evidence.py:2300
↓ 1 callersFunction_ragas_reply_text
Return the judge's answer in one chat-completions reply: its text, or the tool-call arguments RAGAS parses. A reply with neither, such as a refus
src/skillevaluator/tier3/harbor/templates/eval.py:3491
↓ 1 callersFunction_ragas_request_text
Return the judge model and the messages of one RAGAS chat-completions request as text.
src/skillevaluator/tier3/harbor/templates/eval.py:3482
↓ 1 callersFunction_raise_directory_depth_limit
(relative: Path)
src/skillevaluator/utils/secure_fs.py:942
↓ 1 callersFunction_random_ladder_arms
Four random arms: 0-24 cases, 1-3 attempts, some cases missing from an arm, some N/A metrics.
tests/tier3/test_private_ladder_stats.py:388
↓ 1 callersFunction_random_reward
(case_id: str, rng: random.Random, *, skill: bool)
tests/tier3/test_private_ladder_stats.py:373
↓ 1 callersFunction_ranks_like_the_canonical_report
Return whether the canonical report ranks the raw agent data ``raw``. Data the canonical report cannot read, such as a string in place of a s
src/skillevaluator/tier3/result_display.py:223
↓ 1 callersFunction_raw_skip_reason
Return the untrusted reason text that ``get_skip_reason`` would select.
src/skillevaluator/reporting/console_ui.py:984
↓ 1 callersFunction_raw_structural_heading_relative_lines
Return line numbers for every real raw-HTML h1/h2 start tag.
scripts/ci/check_public_benchmarks.py:1307
↓ 1 callersFunction_read
(root: SecureRoot, path: str)
src/skillevaluator/tier3/history.py:68
↓ 1 callersFunction_read
(root: SecureRoot, relative: Path, metadata: os.stat_result)
ci/diagnostics.py:169
↓ 1 callersFunction_read
(directory: Path | None)
ci/report_views.py:128
↓ 1 callersFunction_read
(path: Path, limit: int)
ci/comment_report.py:48
↓ 1 callersFunction_read
(path)
tests/tier3/test_shared_check_adapters.py:157
↓ 1 callersMethod_read
(self, rel: PurePosixPath, max_bytes: int = CONTENT_DEDUP_MAX_FILE_BYTES)
src/skillevaluator/plugin_discovery.py:506
↓ 1 callersFunction_read_attempt_policy
(run_dir: Path)
src/skillevaluator/evaluation/tier3_report.py:4883
↓ 1 callersFunction_read_backend_response
(response: Any, deadline: float)
src/skillevaluator/tier3/harbor/nvidia_build_bridge.py:404
↓ 1 callersFunction_read_bounded
Read a validated regular file through the anchored skill root.
src/skillevaluator/deduplication/utils/skill_collector.py:399
↓ 1 callersFunction_read_bounded_manifest
Read one regular UTF-8 manifest through its anchored parent directory.
src/skillevaluator/embedding/extractor.py:423
↓ 1 callersFunction_read_canary_bytes
Anchored no-follow read in both package and staged standalone modes.
src/skillevaluator/tier3/eval_core/plugin_security.py:3826
↓ 1 callersFunction_read_catalog_text_posix
(catalog_path: Path)
src/skillevaluator/embedding/registry.py:1252
↓ 1 callersFunction_read_catalog_text_windows
(catalog_path: Path)
src/skillevaluator/embedding/registry.py:1285
↓ 1 callersFunction_read_comparison
Read the cross-agent ``comparison.json`` for the Diagnostics tab, if present.
src/skillevaluator/evaluation/tier3_report.py:4925
↓ 1 callersFunction_read_config_text
Read the config as one bounded, single-link regular UTF-8 file without following links. As on the Harbor run, ``evals/`` and the config file must
src/skillevaluator/tier3/evals_config.py:176
↓ 1 callersFunction_read_dataset
(path: Path)
src/skillevaluator/tier3/plugin_prepare.py:786
↓ 1 callersFunction_read_fixtures
(files_dir: Path, owner: str, fixtures: dict[str, bytes | None], owners: dict[str, str])
src/skillevaluator/tier3/plugin_prepare.py:804
↓ 1 callersFunction_read_float
Read a non-negative float from the environment variable or return default.
src/skillevaluator/tier3/harbor/templates/eval.py:1774
↓ 1 callersFunction_read_inline_program
``(runs an inline program, the program)`` for a shell's arguments, its options read with ``read_option``. A shell reads options only until its fi
src/skillevaluator/validators/mcp_static.py:1834
↓ 1 callersFunction_read_int
Read a non-negative integer from the environment variable or return default.
src/skillevaluator/tier3/harbor/templates/eval.py:1763
↓ 1 callersFunction_read_js_string
(source: str, start: int)
src/skillevaluator/tier3/eval_core/codex_tool_call_normalizer.py:387
↓ 1 callersFunction_read_json_mapping_no_follow
Read a regular, unlinked JSON object file, or return ``None``.
src/skillevaluator/tier3/harbor/runner.py:4398
↓ 1 callersFunction_read_key_after_concurrent_publish
(path: Path)
src/skillevaluator/tier3/output_provenance.py:435
↓ 1 callersMethod_read_lenient
Like :meth:`_read`, but a file that is not UTF-8 is decoded with replacement characters. Claude Code still lists such a component (a Latin-1
src/skillevaluator/plugin_discovery.py:513
↓ 1 callersFunction_read_lenient_manifest
Read a discovered manifest inode up to ``max_bytes`` and decode it leniently.
src/skillevaluator/plugin_manifest.py:177
↓ 1 callersMethod_read_only_passwd_mount_lines
Return lines whose every ``/etc/passwd`` belongs to an exact read-only mount. The file is read once, without following links, from beneath th
src/skillevaluator/validators/security.py:5471
↓ 1 callersFunction_read_ready_file
Read and validate a private regular ready file through a no-follow fd.
src/skillevaluator/tier3/harbor/nvidia_build_bridge.py:672
↓ 1 callersFunction_read_referenced_trajectory
( agent_dir: Path, reference: Any, state: _TrajectoryReferenceState, *, count_against_refe
src/skillevaluator/tier3/harbor/collector.py:2099
↓ 1 callersFunction_read_regular_text
Read a single-link regular file without following a link, refusing more than ``max_bytes``.
src/skillevaluator/tier3/plugin_native.py:2447
↓ 1 callersFunction_read_response_body
Read one response without letting byte trickles reset the wall-clock limit. ``urllib``'s timeout is a socket-inactivity timeout. A peer that send
src/skillevaluator/model_catalog.py:686
← previousnext →7,201–7,300 of 24,741, ranked by callers