keyhog-core 0.5.73

keyhog-core: shared data model and detector specifications for the KeyHog secret scanner
Documentation
# Docs: https://docs.gitlab.com/ee/user/clusters/agent/
# Format: glagent- prefix followed by 20+ alphanumeric/_/- characters.
#   The GitLab Agent for Kubernetes authenticates to the agent server (KAS)
#   with this token; leaking one lets an attacker impersonate the agent and
#   reach the connected cluster, so the severity is critical.
# Verify: no public endpoint (cluster-internal / KAS authentication); unverified.
# Prefix: glagent-

[detector]
id = "gitlab-agent-token"
name = "GitLab Agent for Kubernetes Token"
service = "gitlab"
severity = "critical"
ml = { match_mode = "lift", entropy_mode = "disabled", weight = 1.0, context_radius_lines = 5 }
match_confidence = { literal_prefix_weight = 0.35, context_anchor_weight = 0.20, entropy_weight = 0.20, high_entropy_partial_weight = 0.12, moderate_entropy_threshold = 3.0, moderate_entropy_weight = 0.05, low_entropy_penalty_floor = 2.0, low_entropy_min_match_length = 10, low_entropy_penalty_multiplier = 0.60, keyword_nearby_weight = 0.10, sensitive_file_weight = 0.10, companion_weight = 0.05, very_high_entropy_margin = 1.2999999999999998, named_anchor_floor = 0.55, assignment_context_multiplier = 1.0, string_literal_context_multiplier = 0.9, unknown_context_multiplier = 0.8, documentation_context_multiplier = 0.3, comment_context_multiplier = 0.4, test_context_multiplier = 0.3, encrypted_context_multiplier = 0.05, soft_context_suppression_threshold = 0.5, encrypted_context_suppression_threshold = 0.8, post_match = { placeholder_multiplier = 0.05, minimum_byte_diversity = 0.1, low_diversity_multiplier = 0.1, maximum_repeat_ratio = 0.8, degenerate_run_min_length = 10, degenerate_repeat_multiplier = 0.1, fixture_path_multiplier = 0.5, ml_context_reapply_below = 0.95 } }
# "kas" removed: over-broad 3-char prefilter (kaspersky/alaska/…) that no regex
# references: the pattern requires the unambiguous `glagent-` prefix (keyword
# "glagent-"), so "kas" only forced wasted scan passes. Recall-neutral.
keywords = ["glagent-", "gitlab", "agentk"]

[[detector.patterns]]
regex = 'glagent-[a-zA-Z0-9_\-]{20,}'
description = "GitLab Agent for Kubernetes token (glagent- prefix)"

[[detector.tests]]
test_positive = "glagent-Hx7Kp2Qm9Rn4Sb6Tw8Vz1Yc3"
test_negative = "glagent-short"