- default profile:
app/configs/default.yaml - strict profile:
app/configs/strict.yaml
Example usage:
sentilook scan <path> --config configs/default.yamlscan: performance and safety limitsignore: exclusion rulesdetectors: enabled detectors and optionsagent_threat_patterns: grouped pattern togglesmasking: preview masking presentation rulesseverity: default and path-based severity promotionreport: report behaviorsuppressions: detector-specific suppression rules
scan:
follow_symlinks: false
max_files: 50000
max_file_size_mb: 5
timeout_seconds: 90
include_hidden_files: true
detectors:
enabled:
- "dotenv_key_value"
- "jwt_like"
- "aws_iam_key"
- "openai_key"
- "stripe_key"
- "npm_token"
- "docker_pat"
- "slack_token"
- "gcp_service_key"
- "private_key_block"
- "generic_api_key"
- "high_entropy_string"
- "agent_threat_pattern"
options:
high_entropy_string:
min_length: 24
min_entropy_threshold: 3.5
max_findings_per_file: 5
generic_api_key:
min_length: 20
agent_threat_patterns:
enabled: true
match_download_exec: true
match_obfuscation: true
match_network: true
match_init_time: true
suppressions:
- detector: "high_entropy_string"
path_pattern: "*.test.js"
- detector: "generic_api_key"
value_pattern: "EXAMPLE_*"default.yaml: security-first baseline for normal local scans, including hidden dotfilesstrict.yaml: broader scan limits and denser detector output for deep review runs
min_length: minimum token lengthmin_entropy_threshold: Shannon entropy thresholdmax_findings_per_file: maximum findings per file
The legacy min_entropy key is still accepted for compatibility.
min_length: minimum accepted token length for the GitHub-style token detector bucket- this applies to the current
generic_api_keydetector only, not to provider-specific detectors such as AWS, OpenAI, or Stripe
enabled: default master switch when no per-group toggles are explicitly setmatch_download_exec: download-and-execute patterns such ascurl|bashandiwr|iexmatch_obfuscation: obfuscated execution patterns such asbase64 -d | shand-EncodedCommandmatch_network: suspicious public endpoint patterns such as publicIP:port,/dev/tcp, andnc -ematch_init_time: JavaScript and Python top-level execution patterns
Rules:
- if no per-group toggles are set,
enabledcontrols the fullagent_threat_patterngroup - if any per-group toggle is set, only explicitly enabled groups run
- if any per-group toggle is present,
enabled: falseno longer disables the group globally; the explicit per-group toggles become the only active controls - loopback, private-range, unspecified, and link-local
IP:portvalues are intentionally ignored to reduce developer-noise - standard Go
func init()declarations are intentionally not flagged
dotenv_key_valuejwt_likeaws_iam_keyopenai_keystripe_keynpm_tokendocker_patslack_tokengcp_service_keyprivate_key_blockgeneric_api_keyhigh_entropy_stringagent_threat_pattern
detector: detector identifierpath_pattern: glob path or filename patternvalue_pattern: glob pattern against the detected raw value
Suppressed findings do not appear in reports, but they remain in the audit log as finding_suppressed.
true(recommended default): share-safe output mode- HTML finding paths are relative or shortened
- SARIF artifact paths stay relative when possible, with basename fallback if not
- intended for screenshots, broader review, and safer circulation
false: local operator mode- HTML report keeps full local file paths for remediation work
- SARIF may contain local
file:///...artifact URIs - treat the generated output as local-only and do not distribute it
- Sentilook always masks secret previews in terminal, audit, HTML, and SARIF output
prefix_chars,suffix_chars, andreplace_charcontrol the preview format only- older custom configs that still include
masking.enabled,masking.strategy, ormasking.never_print_rawshould remove those keys
scan.include_hidden_files: trueis recommended for real repository scansreport.share_safe_mode: trueis recommended for any output that might be circulated beyond the operator machine- all previews are masked by runtime policy
- raw secret values must not appear in console, audit, or report outputs