Code
Hub
Workspaces
Following
Trending
Connect
MCP
copy
Create free account
hub
/
github.com/NVIDIA/SkillEvaluator
/ functions
Functions
24,741 in github.com/NVIDIA/SkillEvaluator
⨍
Functions
24,741
◇
Types & classes
1,292
↳
Endpoints
703
↓ 1 callers
Function
_claude_rows
(inventory: PluginInventory, data: dict[str, Any], repo_root: Path | None)
src/skillevaluator/plugin_dependency_records.py:84
↓ 1 callers
Function
_claude_url_ok
Whether a WHATWG URL parser (Claude Code's zod ``url()``) accepts ``value`` as an absolute URL. It checks what makes ``new URL()`` fail: a missin
src/skillevaluator/plugin_format_schema.py:562
↓ 1 callers
Function
_claude_user_config_problems
(option: Any)
src/skillevaluator/plugin_format_schema.py:764
↓ 1 callers
Function
_clauses
(text: str)
src/skillevaluator/evaluation/dataset_case_removal.py:208
↓ 1 callers
Function
_clean_aliases
Bounded ``staged name (casefolded) -> declared name`` pairs whose target is a declared subagent.
src/skillevaluator/tier3/eval_core/plugin_signals.py:1538
↓ 1 callers
Function
_cleanup_backup_best_effort
(parent_descriptor: int, backup_name: str)
src/skillevaluator/tier3/harbor/secure_copy.py:1405
↓ 1 callers
Method
_client_is_authorized
(self)
src/skillevaluator/tier3/harbor/nvidia_build_bridge.py:230
↓ 1 callers
Function
_client_reading
How WHATWG clients read an ambiguous URL, for messages: no userinfo, query, or fragment.
src/skillevaluator/validators/mcp_static.py:412
↓ 1 callers
Function
_clip
``text`` cut to :data:`MAX_REF_LABEL_CHARS` characters, ending in ``...`` when cut.
src/skillevaluator/plugin_dependencies.py:336
↓ 1 callers
Function
_closest_field
The known field a misspelled key most likely means (``mcpServer`` -> ``mcpServers``), if any.
src/skillevaluator/plugin_format_schema.py:154
↓ 1 callers
Function
_cluster_location_text
Return the finding scope and member location list, bounded in length.
src/skillevaluator/deduplication/intra_skill/intra_skill_validator.py:121
↓ 1 callers
Function
_cmd_references_exact_target
Detect a target reference, returning ``None`` when a parser bound is hit.
src/skillevaluator/tier3/eval_core/checks.py:3058
↓ 1 callers
Function
_cmd_text
(tc: dict[str, Any])
src/skillevaluator/tier3/eval_core/checks.py:7605
↓ 1 callers
Function
_code_forms
``(pattern, by_name)`` for each form, once with its literal file and once with a name.
src/skillevaluator/tier3/eval_core/plugin_signals.py:2496
↓ 1 callers
Function
_code_paths
(code: str, forms: Sequence[tuple[re.Pattern[str], bool]], names: Mapping[str, str])
src/skillevaluator/tier3/eval_core/plugin_signals.py:2533
↓ 1 callers
Function
_code_span_reference_text
Return extra reference text for names only a shell command or glob spells. Code spans and fenced blocks name directories as bare words (``fin
src/skillevaluator/validators/skillspector_scope.py:713
↓ 1 callers
Function
_codex_call_outputs
Tool call id -> the start of its result text in the trajectory (bounded).
src/skillevaluator/tier3/plugin_trace_artifacts.py:186
↓ 1 callers
Function
_codex_collab_tool_arguments
Map Codex ``CollabToolCallItem`` wire fields into ATIF tool arguments. Schema fields: ``sender_thread_id``, ``receiver_thread_ids``, ``prompt``,
src/skillevaluator/tier3/eval_core/log_converters.py:724
↓ 1 callers
Function
_codex_exec_records
Split a completed outer observation into exactly ``count`` checked records, in output order.
src/skillevaluator/tier3/eval_core/codex_tool_call_normalizer.py:1349
↓ 1 callers
Function
_codex_file_change_step
(item: dict[str, Any], evt: dict[str, Any], tool_index: int)
src/skillevaluator/tier3/eval_core/log_converters.py:775
↓ 1 callers
Function
_codex_function_arguments
(function_call: dict[str, Any])
src/skillevaluator/tier3/eval_core/log_converters.py:619
↓ 1 callers
Function
_codex_headers
Merge a Codex server's ``http_headers`` into ``headers`` for the static policy and Tier 3. Every header either map declares is kept, so a literal
src/skillevaluator/plugin_mcp.py:746
↓ 1 callers
Function
_codex_implicit_disabled
(skill_path: Path)
src/skillevaluator/harness/overlay.py:367
↓ 1 callers
Function
_codex_item_status
``failed`` or ``completed`` for a finished ``mcp_tool_call`` item, else ``None``.
src/skillevaluator/tier3/plugin_trace_artifacts.py:127
↓ 1 callers
Function
_codex_item_text
The start of a finished item's first text block (the tool's own answer), or ``""``.
src/skillevaluator/tier3/plugin_trace_artifacts.py:176
↓ 1 callers
Function
_codex_item_text
(item: dict[str, Any])
src/skillevaluator/tier3/eval_core/log_converters.py:601
↓ 1 callers
Function
_codex_mcp_calls
Codex MCP calls read from the raw Codex logs: ``(call id -> server, call id -> status)``. Harbor's Codex trajectory keeps only the bare MCP tool
src/skillevaluator/tier3/plugin_trace_artifacts.py:243
↓ 1 callers
Function
_codex_mcp_observation
(item: dict[str, Any])
src/skillevaluator/tier3/eval_core/log_converters.py:699
↓ 1 callers
Function
_codex_mcp_result_text
(item: dict[str, Any])
src/skillevaluator/tier3/eval_core/log_converters.py:677
↓ 1 callers
Function
_codex_observation_body
(item: dict[str, Any], *, include_result: bool = False)
src/skillevaluator/tier3/eval_core/log_converters.py:655
↓ 1 callers
Function
_codex_session_files
Return the first Codex session logs (``*.jsonl``) under ``<trial>/agent/sessions``. The tree is agent-writable, so the walk never follows a link
src/skillevaluator/tier3/plugin_trace_artifacts.py:69
↓ 1 callers
Function
_codex_status_error_prefix
(item: dict[str, Any])
src/skillevaluator/tier3/eval_core/log_converters.py:637
↓ 1 callers
Function
_codex_thread_item
Return the Codex thread item payload from a completed event.
src/skillevaluator/tier3/eval_core/log_converters.py:626
↓ 1 callers
Function
_codex_tool_step
( *, item: dict[str, Any], evt: dict[str, Any], tool_index: int, function_name: str, a
src/skillevaluator/tier3/eval_core/log_converters.py:743
↓ 1 callers
Function
_codex_txt_items
``mcp_tool_call`` items of ``codex.txt`` in start order: server, tool, arguments key, status. ``codex.txt`` item ids (``item_N``) are not tool ca
src/skillevaluator/tier3/plugin_trace_artifacts.py:140
↓ 1 callers
Function
_codex_web_search_arguments
Map Codex ``WebSearchItem`` wire fields into ATIF tool arguments. Schema shape (``codex exec --json``): required top-level ``query`` plus an
src/skillevaluator/tier3/eval_core/log_converters.py:703
↓ 1 callers
Function
_coerce_category
(value: Any)
src/skillevaluator/evaluation/insights_judge.py:374
↓ 1 callers
Function
_coerce_email_regex
(value: Any, source: str)
src/skillevaluator/validators/policy.py:159
↓ 1 callers
Function
_coerce_metadata_id_requirement
Return a supported ``metadata_id_requirement``; reject anything else. Unlike an unknown severity, a misspelled requirement is an error: silently
src/skillevaluator/validators/policy.py:176
↓ 1 callers
Function
_coerce_severity
(value: Any, key: str)
src/skillevaluator/validators/policy.py:144
↓ 1 callers
Function
_coerce_severity
(value: Any)
src/skillevaluator/evaluation/insights_judge.py:366
↓ 1 callers
Function
_collapsed_prefix
Return the whole words of ``value`` that fit in ``_MAX_SCAN_CHARS``, whitespace collapsed. A whitespace run that contains a line break becomes on
src/skillevaluator/evaluation/tier3_findings.py:66
↓ 1 callers
Function
_collect
(value: object, key: str | None = None, depth: int = 0)
src/skillevaluator/validators/skillspector_scope.py:804
↓ 1 callers
Function
_collect_bundle_manifest_servers
The ``mcp`` list of a bundle-reference ``agent_plugin.yaml``, kept as declared (Pydantic validates it).
src/skillevaluator/plugin_mcp.py:292
↓ 1 callers
Function
_collect_case_inputs
Reuse the engine's path and control-file rules, preserving input-relative names.
src/skillevaluator/tier3/plugin_prepare.py:704
↓ 1 callers
Function
_collect_dataset
( package: PreparedRuntimePackage, evals_source: Path | None, *, max_cases: int )
src/skillevaluator/tier3/plugin_prepare.py:656
↓ 1 callers
Function
_collect_default_files
The format's default MCP files at the root, unless a declared path names one (then it loads as declared). A path Codex drops does not count as na
src/skillevaluator/plugin_mcp.py:357
↓ 1 callers
Function
_collect_files_anchored
Walk a skill directory and collect all analyzable text files. Args: skill_root: Path to the skill directory. excluded_dirs: Direc
src/skillevaluator/deduplication/utils/skill_collector.py:443
↓ 1 callers
Function
_collect_inline_map
An inline ``mcpServers`` server map (Cursor and Codex may also wrap it in ``{"mcpServers": {...}}``).
src/skillevaluator/plugin_mcp.py:395
↓ 1 callers
Function
_collect_path_ref
Collect one ``mcpServers`` string: a remote bundle URL, a ``.json`` config file, or a bundle file. A reference that cannot be loaded is a broken
src/skillevaluator/plugin_mcp.py:773
↓ 1 callers
Function
_collect_secret_variants
Extract sensitive environment values and their shell/URL-encoded representations.
src/skillevaluator/tier3/harbor/gke_environment.py:384
↓ 1 callers
Function
_collect_url_ref
Collect one remote ``mcpServers`` reference; only an ``.mcpb``/``.dxt`` bundle may be named by URL. The bundle is recorded but not inspected. One
src/skillevaluator/plugin_mcp.py:872
↓ 1 callers
Function
_collector_owned_agent_dirs
(output_root: Path)
src/skillevaluator/tier3/harbor/collector.py:559
↓ 1 callers
Function
_collector_pass_at_k
One arm's pass@k as the collector records it: ``scored`` attempts per case, ``planned`` under a run budget.
tests/tier3/test_token_usage.py:58
↓ 1 callers
Function
_combine
Sum arm summaries of several agents; the sum is complete only when every part is. A part whose total is out of the exact JSON range makes the sum
src/skillevaluator/tier3/token_usage.py:389
↓ 1 callers
Function
_combine_skillspector_applicability
Total per-skill SkillSpector applicability counters for a folder of skills.
src/skillevaluator/validators/security.py:1078
↓ 1 callers
Function
_combined
(outcomes: Iterable[Status])
src/skillevaluator/reporting/run_status.py:137
↓ 1 callers
Function
_command_executes_skill
(command: str, expected_script: str | None, paths: list[str])
src/skillevaluator/tier3/eval_core/checks.py:7380
↓ 1 callers
Method
_command_hook
(self, handler: dict[str, Any], site: _HookSite)
src/skillevaluator/plugin_component_risk.py:3583
↓ 1 callers
Function
_command_inline_credential
Whether hook command words carry a credential outside a URL: the value of a credential-named flag (``--api-key x``, the MCP command-argument rule)
src/skillevaluator/plugin_component_risk.py:2793
↓ 1 callers
Method
_command_map
The object form of ``commands``: each entry has a ``source`` file or folder, or inline ``content`` text.
src/skillevaluator/plugin_discovery.py:996
↓ 1 callers
Method
_command_map_folder
A ``commands`` map entry whose source is a folder: one command per Markdown file directly in it. Claude Code accepts "a command file or skill
src/skillevaluator/plugin_discovery.py:1041
↓ 1 callers
Function
_command_run_before_refusal
The first command of ``text``, when the shell runs it before refusing a later one and the walk reads it as the shell does; empty otherwise. A
src/skillevaluator/tier3/eval_core/script_invocation.py:1889
↓ 1 callers
Method
_command_source
A ``commands`` map entry's ``source``: a command file, or a folder of them (:meth:`_command_map_folder`).
src/skillevaluator/plugin_discovery.py:1024
↓ 1 callers
Function
_command_text
(handler: dict[str, Any], keys: tuple[str, ...] = ("command",))
src/skillevaluator/plugin_component_risk.py:1695
↓ 1 callers
Function
_command_to_dict
(command: click.Command)
tests/golden/_cli_surface.py:59
↓ 1 callers
Function
_command_tokens
(handler: dict[str, Any], keys: tuple[str, ...] = ("command",))
src/skillevaluator/plugin_component_risk.py:1683
↓ 1 callers
Function
_command_url_credentials
Whether a command line embeds credentials in a URL (``https://user:password@host``, ``?token=...``).
src/skillevaluator/plugin_component_risk.py:2784
↓ 1 callers
Function
_command_words
(cmd: str)
src/skillevaluator/tier3/eval_core/checks.py:7608
↓ 1 callers
Function
_commands
(command: click.Command, ctx: click.Context)
src/skillevaluator/cli_help.py:217
↓ 1 callers
Function
_commands
(text: str)
tests/test_skillevaluator_skill.py:23
↓ 1 callers
Function
_commands_kwargs
(**overrides: Any)
tests/tier3/test_skip_with_skill.py:158
↓ 1 callers
Function
_commands_similar
(cmd1: str, cmd2: str)
src/skillevaluator/tier3/eval_core/checks.py:7614
↓ 1 callers
Function
_commands_that_ran
Which commands of a shell list ran: ``a && b`` skips ``b`` when ``a`` failed, ``a || b`` when it did not.
src/skillevaluator/tier3/eval_core/plugin_signals.py:2694
↓ 1 callers
Function
_comment_identity
(comment: dict)
ci/comment_report.py:144
↓ 1 callers
Function
_comments_for_language
(language: str, lines: list[str])
src/skillevaluator/validators/security.py:393
↓ 1 callers
Function
_commit
(repository: Path, message: str)
ci/smoke_image.py:162
↓ 1 callers
Function
_compact_agent_result
Project duplicated agent detail only when its canonical artifact exists.
src/skillevaluator/tier3/harbor/runner.py:464
↓ 1 callers
Function
_compact_agents
Keep per-agent comparisons, measured resources and failure scope without trials.
src/skillevaluator/evaluation/tier3_feedback_chat.py:757
↓ 1 callers
Function
_compact_behavior_conversation
Keep both setup context and late outcome evidence in behavior prompts.
src/skillevaluator/tier3/harbor/templates/eval.py:848
↓ 1 callers
Function
_compact_condition_detail
Retain condition truth while sampling diagnostics stored in its summary.
src/skillevaluator/tier3/harbor/runner.py:442
↓ 1 callers
Function
_compact_evidence_refs
(raw_refs: object)
src/skillevaluator/evaluation/tier3_report.py:2924
↓ 1 callers
Function
_compact_grid
One line per flagged harness: its most serious finding, then how many more.
src/skillevaluator/harness/report.py:87
↓ 1 callers
Function
_compact_json
Return compact ASCII JSON for the shared schema budget (one char per byte).
src/skillevaluator/tier3/mcp_proof.py:962
↓ 1 callers
Function
_compact_pass_detail
Retain exact pass aggregates while referencing persisted case details.
src/skillevaluator/tier3/harbor/runner.py:426
↓ 1 callers
Function
_compact_resources
Keep measured agent tokens and seconds per arm; the scope text is in the system prompt.
src/skillevaluator/evaluation/tier3_feedback_chat.py:750
↓ 1 callers
Function
_compact_summary
Keep the Actions landing view small; detailed evidence lives in reports.
ci/report_views.py:513
↓ 1 callers
Method
_compare_fields
Field-level differences between the two sides, as message fragments (empty when they match). Claude Code's manifest errors are compared with
src/skillevaluator/validators/claude_plugin_validate.py:520
↓ 1 callers
Function
_comparison_column_config
(columns: tuple[str, ...])
src/skillevaluator/reporting/dashboard_app.py:181
↓ 1 callers
Function
_comparison_views
(projection: _RuntimeProjection, comparisons: dict)
src/skillevaluator/reporting/plugin_runtime.py:262
↓ 1 callers
Function
_compatibility_alias_target
Return the recognized contained CLAUDE.md -> AGENTS.md alias target.
src/skillevaluator/utils/secure_fs.py:926
↓ 1 callers
Function
_complete_metric_mean
(scores: dict[str, Any], metrics: list[str])
src/skillevaluator/evaluation/tier3_report.py:2046
↓ 1 callers
Function
_complete_result
(skill: str, names: tuple[str, str], scores: tuple[float, float])
tests/reporting/test_report_browser.py:34
↓ 1 callers
Function
_completion_token_limit
(model, max_tokens)
src/skillevaluator/tier3/harbor/templates/eval.py:1292
↓ 1 callers
Function
_component_for_ref
The component whose root-relative path a declared ref names, compared after normalization. ``./agents/x.md``, ``agents/x.md``, and ``${CLAUDE_PLU
src/skillevaluator/plugin_inventory.py:585
↓ 1 callers
Function
_component_name
``(name, foreign)``: a declared component's own spelling when the name is this plugin's. ``release-kit:version-bump`` and ``version-bump`` both b
src/skillevaluator/tier3/eval_core/plugin_signals.py:2740
↓ 1 callers
Function
_compose
A binding nested two or three scopes deep, read at a random level. Which value the interpreter reads depends on which scopes keep a binding,
scripts/script_invocation_differential.py:1017
↓ 1 callers
Function
_compose_additional_build_contexts
Return named Compose build contexts after structural validation.
src/skillevaluator/tier3/harbor/adapter.py:3105
↓ 1 callers
Function
_compose_build_arg_overrides
Return literal build-arg overrides for Compose services that build *dockerfile*.
src/skillevaluator/tier3/harbor/adapter.py:2970
↓ 1 callers
Function
_compose_build_context_values
Return validated primary and named build-context paths.
src/skillevaluator/tier3/harbor/adapter.py:3092
← previous
next →
5,901–6,000 of 24,741, ranked by callers