Glossary

Terms used in Jnkn documentation.

A

Artifact : A node in the dependency graph representing code, configuration, or infrastructure. Examples: env var, file, Terraform resource.

B

Blast Radius : The set of artifacts that would be affected by a change to a given artifact. Computed by traversing downstream dependencies.

Blocked Token : A token (like "id", "key") that provides no signal for matching and is ignored during stitching.

C

Confidence : A score from 0.0 to 1.0 indicating how likely a cross-domain match is correct. Higher = more certain.

Cross-Domain Dependency : A dependency that spans different technology domains (e.g., Python code depending on a Terraform resource).

D

Dependency Graph : A directed graph where nodes are artifacts and edges are dependencies between them.

E

Edge : A connection between two nodes in the dependency graph. Has a type (reads, imports, etc.) and optional metadata.

Extractor : A component that detects specific patterns in source code (e.g., os.getenv() calls).

G

Graph : See Dependency Graph.

M

Matcher : A component that determines if two artifacts should be linked based on name similarity.

N

Node : An entity in the dependency graph. Has an ID, type, and metadata.

Normalized Name : A standardized form of an artifact name used for matching. Example: DATABASE_URL → databaseurl.

P

Parser : A component that reads source files and extracts nodes and edges.

Pattern : A code construct that Jnkn recognizes (e.g., os.getenv("VAR")).

Penalty : A multiplier that reduces confidence (e.g., 0.5 for short tokens).

R

Rule : See Stitching Rule.

S

Signal : A factor that increases confidence in a match (e.g., exact token overlap).

Stitching : The process of creating edges between nodes from different domains based on name similarity.

Stitching Rule : A rule that defines which node types can be matched and how confidence is calculated.

Suppression : A user-defined rule to ignore specific matches (false positives).

T

Token : A word extracted from an artifact name. DATABASE_URL → ["database", "url"].

Token Matching : The technique of linking artifacts by comparing their tokenized names.

Artifact Types

code_file : A source code file (.py, .js, etc.).

code_entity : A function or class definition within a file.

env_var : An environment variable.

infra_resource : A Terraform resource, variable, output, or data source.

k8s_resource : A Kubernetes resource (Deployment, ConfigMap, Secret, etc.).

data_asset : A dbt model, source, or seed.

Edge Types

reads : Source reads a value from target (e.g., file reads env var).

imports : Source imports target (e.g., Python import).

provides : Source provides target (e.g., Terraform outputs an env var value).

configures : Source configures target (e.g., K8s deployment uses ServiceAccount).

contains : Source contains target (e.g., file contains function definition).

references : Generic reference between artifacts.