MCPcopy Create free account

hub / github.com/NVIDIA/SkillEvaluator / functions

Functions24,741 in github.com/NVIDIA/SkillEvaluator

↓ 1 callersFunction_security_clean_comparison
(info: dict[str, Any])
src/skillevaluator/evaluation/tier3_report.py:3523
↓ 1 callersFunction_security_command
The shell text a call runs: ``command``/``cmd``/``script``/``raw`` (argv lists joined), or Codex ``write_stdin`` ``chars`` typed into a running se
src/skillevaluator/tier3/eval_core/plugin_security.py:3234
↓ 1 callersFunction_security_data_indexes
Argument indexes that are data, not code: heredoc and here-string bodies of a non-interpreter.
src/skillevaluator/tier3/eval_core/plugin_security.py:2487
↓ 1 callersFunction_security_dumps_env
A simple command that prints environment variables: ``env``, ``printenv``, ``export -p``, ``declare -x``, ``set`` alone, or a ``/proc/<pid>/enviro
src/skillevaluator/tier3/eval_core/plugin_security.py:2907
↓ 1 callersFunction_security_find_parse
``find``'s expression as a tree, and the commands of its ``-exec``-style actions. Nodes: ``("and", nodes)``, ``("or", nodes)``, ``("not", node)``
src/skillevaluator/tier3/eval_core/plugin_security.py:2702
↓ 1 callersFunction_security_find_paths
Paths ``find`` may print for the store file ``target`` (``-path`` tests match these).
src/skillevaluator/tier3/eval_core/plugin_security.py:2799
↓ 1 callersFunction_security_find_reaches
A search from ``root`` can reach one of the credential files a ``find -name`` may look for.
src/skillevaluator/tier3/eval_core/plugin_security.py:2814
↓ 1 callersFunction_security_find_reads
Credential stores whose files a ``find`` reads (``find / -name 'id_rsa*' | xargs cat``). The search must start at or above the store, its express
src/skillevaluator/tier3/eval_core/plugin_security.py:2819
↓ 1 callersFunction_security_find_value
``True``/``False`` when ``node`` surely does or does not select the file, ``None`` when it may.
src/skillevaluator/tier3/eval_core/plugin_security.py:2778
↓ 1 callersFunction_security_finding_message
(result: Mapping[str, Any])
src/skillevaluator/tier3/attempt_verdict.py:100
↓ 1 callersFunction_security_finding_signature
(finding: dict[str, Any])
src/skillevaluator/tier3/harbor/collector.py:5851
↓ 1 callersFunction_security_glob_reaches
A search limited to file-name ``globs`` can read a file of the store ``entry``: a glob matches the store's name or the name of a credential file i
src/skillevaluator/tier3/eval_core/plugin_security.py:2640
↓ 1 callersFunction_security_hidden_below
``entry`` lies under a hidden (dot) directory or is a dot file, below the directory ``path``.
src/skillevaluator/tier3/eval_core/plugin_security.py:2651
↓ 1 callersFunction_security_indirect_injection
An info finding for the first tool output that tells the agent to drop its instructions, else ``None``.
src/skillevaluator/tier3/eval_core/plugin_security.py:3386
↓ 1 callersFunction_security_is_harness_text
(message)
src/skillevaluator/tier3/eval_core/plugin_security.py:3230
↓ 1 callersFunction_security_output_read
A command's output names files that get read: it goes to ``xargs <reader>``, or (with no pipe after it) into a ``$( ... )`` or ``<( ... )`` substi
src/skillevaluator/tier3/eval_core/plugin_security.py:2867
↓ 1 callersFunction_security_pipes_into_runner
For each of ``items``, whether its output reaches, through the rest of its pipeline, a command that may run it. One pass from the end, so a long p
src/skillevaluator/tier3/eval_core/plugin_security.py:2936
↓ 1 callersFunction_security_placeholder
A key-shaped placeholder (``sk-your-key-here``, ``sk-xxxxxxxx``): real keys always carry a digit.
src/skillevaluator/tier3/eval_core/plugin_security.py:3148
↓ 1 callersFunction_security_process_substitutions
``items`` with each ``<( ... )`` or ``>( ... )`` process substitution's commands moved in front of the command it belongs to, which gets a file (`
src/skillevaluator/tier3/eval_core/plugin_security.py:3022
↓ 1 callersFunction_security_remote_script
A download piped straight into a shell (``curl -fsSL URL | sudo bash``), read in one pass.
src/skillevaluator/tier3/eval_core/plugin_security.py:3286
↓ 1 callersFunction_security_report
Add one critical path finding, once per ``(type, evidence)`` in a call.
src/skillevaluator/tier3/eval_core/plugin_security.py:3754
↓ 1 callersFunction_security_result
(error)
src/skillevaluator/deduplication/plugin/commands.py:165
↓ 1 callersFunction_security_rm_outside_tmp
``rm`` with ``-r``, ``-R`` or ``-f`` and an operand outside ``/tmp`` (quotes and ``cd /tmp`` handled).
src/skillevaluator/tier3/eval_core/plugin_security.py:2890
↓ 1 callersFunction_security_runs_input
A command that may run the text piped into it, or open the names in it: ``xargs``, ``parallel``, a shell, a loop or group, or an interpreter with
src/skillevaluator/tier3/eval_core/plugin_security.py:2919
↓ 1 callersFunction_security_score_findings
(reward: dict[str, Any])
src/skillevaluator/tier3/harbor/collector.py:5844
↓ 1 callersFunction_security_secret_exposure
(observation, *, tool, before, environ=None, env_dump=False)
src/skillevaluator/tier3/eval_core/plugin_security.py:3196
↓ 1 callersFunction_security_shell_cwd
Claude Code's Bash directory after a call: where its ``cd`` left it, or where Claude Code reset it ("Shell cwd was reset to /workspace" after a co
src/skillevaluator/tier3/eval_core/plugin_security.py:3247
↓ 1 callersFunction_security_shell_session
A Codex session that runs a bare shell (``bash``, ``zsh -i``): text typed into it runs in that shell, so a ``cd`` in it moves the directory for th
src/skillevaluator/tier3/eval_core/plugin_security.py:3259
↓ 1 callersFunction_security_simple_command
Read one simple command into ``scan``; return its words without the data-only ones. ``piped`` says its output may be run or opened (a command sub
src/skillevaluator/tier3/eval_core/plugin_security.py:2949
↓ 1 callersFunction_security_tree_operands
Directory operands a command reads in full (an archive, a recursive copy, a transfer).
src/skillevaluator/tier3/eval_core/plugin_security.py:2540
↓ 1 callersFunction_security_tree_reads
Directories a command reads every file under, among them a credential store it reaches. An archive, a recursive copy or a transfer reads every fi
src/skillevaluator/tier3/eval_core/plugin_security.py:2657
↓ 1 callersFunction_security_write_message
(entry)
src/skillevaluator/tier3/eval_core/plugin_security.py:3773
↓ 1 callersFunction_security_write_targets
Files one simple command writes: redirections, tee, cp/mv/install/ln/dd targets, and sed -i operands.
src/skillevaluator/tier3/eval_core/plugin_security.py:2500
↓ 1 callersFunction_segment_timing_ms
Attribute each timestamp gap of one Harbor step's trajectory once.
src/skillevaluator/tier3/harbor/trajectory_timing.py:294
↓ 1 callersFunction_select_trials
Choose the trials to keep across all Harbor jobs before any file is read. Each job keeps its own first trials, failing ones first. Across jobs, f
ci/diagnostics.py:115
↓ 1 callersFunction_selected_environment_projection
Select the custom environment as one compatibility-preserving unit.
src/skillevaluator/tier3/harbor/adapter.py:6387
↓ 1 callersFunction_selected_grader_paths
(evals_dir: Path)
src/skillevaluator/tier3/harbor/adapter.py:6402
↓ 1 callersFunction_selected_rule_matches
( rule: BoundaryRule, candidates: list[_ReconstructedCandidate], *, line_number: int, defe
scripts/check_oss_boundary.py:533
↓ 1 callersFunction_selected_skills
(skills: Sequence[Path])
src/skillevaluator/tier3/interactions.py:242
↓ 1 callersFunction_semgrep_display_path
Return *path* relative to the scanned skill. Semgrep reports absolute paths for the absolute target ``_run_semgrep`` passes, so a raw path wo
src/skillevaluator/validators/code_risk.py:40
↓ 1 callersFunction_semgrep_error_line
(error: dict)
src/skillevaluator/validators/code_risk.py:58
↓ 1 callersFunction_semgrep_error_type
Return the Semgrep error type name; JSON reports use a string or ``["Name", ...]``.
src/skillevaluator/validators/code_risk.py:27
↓ 1 callersFunction_semgrep_file_excludes
Return generated file patterns using the target filesystem's case semantics.
src/skillevaluator/validators/code_risk.py:67
↓ 1 callersMethod_send_sse
(self, events: list[dict[str, Any]])
src/skillevaluator/tier3/harbor/nvidia_build_bridge.py:298
↓ 1 callersFunction_sensitive_command_write_match
Return the longest protected entry matched by the legacy or home-path write rule.
src/skillevaluator/tier3/eval_core/checks.py:1211
↓ 1 callersMethod_sensitive_exec
Require a transport that keeps sensitive values out of host argv and logs.
src/skillevaluator/tier3/harbor/local_agents.py:698
↓ 1 callersMethod_sensitive_host_path_reference
(self, command: str)
src/skillevaluator/tier3/harbor/local_environment.py:1688
↓ 1 callersFunction_sensitive_write_target
Return the longest protected entry a shell *command* writes through a redirection, tee, or sed -i.
src/skillevaluator/tier3/eval_core/checks.py:1143
↓ 1 callersFunction_sentence
Bound *text* to *limit* characters and end it as a sentence.
src/skillevaluator/reporting/html.py:1980
↓ 1 callersFunction_sentence_at
(text: str, index: int)
tests/tier3/test_dataset_baseline_fairness.py:1059
↓ 1 callersFunction_separates_folders
Whether exactly ``count`` backslashes are at ``index``, as between two folder names. In JSON the next name may start with an escaped letter or di
src/skillevaluator/evaluation/tier3_feedback_chat.py:324
↓ 1 callersMethod_sequence
(self)
src/skillevaluator/plugin_component_risk.py:1275
↓ 1 callersFunction_server_input_schemas
(spec: Mapping[str, Any] | None)
src/skillevaluator/tier3/eval_core/plugin_signals.py:4017
↓ 1 callersFunction_session_input_is_text
Whether session input types something; Enter or Ctrl-C alone only polls or interrupts.
src/skillevaluator/tier3/eval_core/checks.py:8298
↓ 1 callersFunction_set_counts
(element: ET.Element, *, tests: int, failures: int, errors: int, skipped: int)
src/skillevaluator/reporting/junit_reporter.py:218
↓ 1 callersMethod_set_license_metadata
Set license metadata on validation result.
src/skillevaluator/validators/license.py:708
↓ 1 callersFunction_set_model
(record: dict, model: object)
tests/test_skillspector_provenance.py:32
↓ 1 callersFunction_set_response_socket_timeout
Best-effort propagation of the remaining deadline to urllib's socket.
src/skillevaluator/model_catalog.py:732
↓ 1 callersFunction_set_table_numbers
Set numeric keys of one top-level task.toml table, adding the keys or the table when absent.
src/skillevaluator/tier3/nemo_gym/runner.py:232
↓ 1 callersFunction_set_task_docker_image
Inject ``docker_image`` into a task's ``task.toml`` so Harbor skips building.
src/skillevaluator/tier3/harbor/adapter.py:1618
↓ 1 callersFunction_setting
(value: object)
src/skillevaluator/skillspector_provenance.py:114
↓ 1 callersFunction_settings_findings
The checks on shipped Claude Code settings: a settings file, or the manifest's inline ``settings``.
src/skillevaluator/plugin_config_risk.py:134
↓ 1 callersFunction_settings_image_skill
(tmp_path: Path)
tests/validators/test_skillspector_hook_surface.py:245
↓ 1 callersFunction_settle
(value: str)
src/skillevaluator/reporting/plugin_runtime_io.py:17
↓ 1 callersFunction_settled_renderer
Read the one renderer of a settled batch and return the format of the records it prints. Accepted: `for (let i = 0; i < results.length; i++) text
src/skillevaluator/tier3/eval_core/codex_tool_call_normalizer.py:1019
↓ 1 callersFunction_settled_result
Return a `Promise.allSettled` result with exactly its own keys, else None.
src/skillevaluator/tier3/eval_core/codex_tool_call_normalizer.py:1295
↓ 1 callersFunction_setup_literals
Return the prefixes and exact paths that an inline setup predicate checks.
tests/ci/test_all_tier_workflows.py:547
↓ 1 callersFunction_sha256
(path: Path)
tests/validators/test_skillspector_hook_surface.py:338
↓ 1 callersFunction_sha256
(code: str)
tests/ci/test_native_reports.py:65
↓ 1 callersFunction_shapes
(svg: str)
tests/reporting/test_report_logo.py:15
↓ 1 callersFunction_shared_dimension_overall_lift
Return the canonical standard-policy Overall lift when shared metrics differ.
src/skillevaluator/tier3/commands.py:1679
↓ 1 callersFunction_shared_variants
(candidate: str, *, full_namespace: bool)
src/skillevaluator/validators/skillspector_scope.py:1128
↓ 1 callersFunction_sharing_saving
Characters saved by listing ``text`` once as shared block ``index``. Inline, each of the ``count`` occurrences carries the text. Shared, each
src/skillevaluator/deduplication/intra_skill/llm_analyzer.py:155
↓ 1 callersFunction_shebang_interpreters
Resolve a bounded shebang chain, including ``/usr/bin/env helper``.
src/skillevaluator/tier3/harbor/local_runtime.py:400
↓ 1 callersFunction_shell_apply_patch
(prefix: str, header: str)
tests/tier3/test_security_home_paths.py:89
↓ 1 callersFunction_shell_argv
A Codex ``shell`` call, which sends its command as an argv list.
tests/tier3/test_security_home_paths.py:79
↓ 1 callersFunction_shell_c_positional
The operands after the ``-c`` payload: ``bash -c '...' a b`` gives the payload positional parameters, so ``for f; do`` inside it iterates them
src/skillevaluator/tier3/eval_core/script_invocation.py:317
↓ 1 callersFunction_shell_commands
``line`` split into the separate commands a shell would run. Raises ``ValueError`` on a line no shell could read either, e.g. one holding an
tests/test_ci_workflows.py:697
↓ 1 callersFunction_shell_failure_finding
(path: str, line_number: int, error: _ShellDecodeError)
scripts/check_oss_boundary.py:580
↓ 1 callersFunction_shell_file_io
``(read_paths, written_paths)`` of shell ``text``, resolved against ``cwd`` (``cd`` moves it). Reads are operands of reader and consumer verbs (`
src/skillevaluator/tier3/eval_core/plugin_signals.py:2560
↓ 1 callersFunction_shell_grader_template
()
src/skillevaluator/tier3/commands.py:277
↓ 1 callersFunction_shell_payload
Return the ``-c`` payload of ``bash -lc '...'``-style invocations.
src/skillevaluator/tier3/eval_core/plugin_signals.py:2360
↓ 1 callersFunction_shell_reconstruction
(path: str, line: str)
scripts/check_oss_boundary.py:516
↓ 1 callersFunction_shell_segments
``line`` cut into command texts at every *unquoted* shell operator. Quoting is tracked by hand rather than left to ``shlex``'s lexer, because
tests/test_ci_workflows.py:650
↓ 1 callersFunction_shell_text_regions
Return the closed quoted strings, the comment and here-document body lines, and the comments of shell *text*. Each is a flat list of start and en
src/skillevaluator/tier3/eval_core/checks.py:953
↓ 1 callersMethod_shell_token_to_path
(self, token: str, *, exec_env: dict[str, str])
src/skillevaluator/tier3/harbor/local_environment.py:1753
↓ 1 callersFunction_shell_tokens
(text: str)
src/skillevaluator/tier3/eval_core/plugin_signals.py:2339
↓ 1 callersMethod_shell_write_redirect_targets
(command: str)
src/skillevaluator/tier3/harbor/local_environment.py:1714
↓ 1 callersFunction_shell_write_scans
Return shell *text* with its line continuations joined, and where it is searched for writes. A continuation is joined, except at the end of a com
src/skillevaluator/tier3/eval_core/checks.py:1062
↓ 1 callersFunction_shell_write_targets
Return the files *command* writes through a redirection, tee, or sed -i.
src/skillevaluator/tier3/eval_core/checks.py:1108
↓ 1 callersFunction_shell_write_texts
Return the command texts the home-path write rule reads. The first is ``_action_text`` with each argv-list argument quoted back into one shel
src/skillevaluator/tier3/eval_core/checks.py:901
↓ 1 callersFunction_should_try_fallback
(error: str)
src/skillevaluator/tier3/eval_core/llm_judge.py:201
↓ 1 callersFunction_should_try_fallback
(error)
src/skillevaluator/tier3/harbor/templates/eval.py:1244
↓ 1 callersFunction_show_comparisons
(rows: list[dict[str, Any]])
src/skillevaluator/reporting/dashboard_app.py:236
↓ 1 callersFunction_show_performance
(rows: list[dict[str, Any]])
src/skillevaluator/reporting/dashboard_app.py:213
↓ 1 callersFunction_side_channel_idents
Plugin components a shell call used without their own tool. * An MCP ``tools/call`` sent as JSON-RPC through the shell to a declared server
src/skillevaluator/tier3/eval_core/plugin_signals.py:3323
↓ 1 callersMethod_sidecar_exec_lock
(self, service: str)
src/skillevaluator/tier3/harbor/secure_docker_environment.py:2039
↓ 1 callersFunction_signal_harbor_process_group
(process: subprocess.Popen[bytes], value: signal.Signals)
src/skillevaluator/tier3/harbor/runner.py:3724
↓ 1 callersFunction_signal_scan
Signal an evaluator's process group, which also holds the scanner tools it started.
ci/run_tier1.py:353
↓ 1 callersFunction_signal_views
(projection: _RuntimeProjection, agents: object)
src/skillevaluator/reporting/plugin_runtime.py:173
← previousnext →7,701–7,800 of 24,741, ranked by callers