CEL Libraries

kro includes a rich set of CEL function libraries from three sources: kro's own custom libraries, the cel-go standard extensions, and Kubernetes apiserver libraries.

Library Overview

Library	Provider	Docs
Hash	kro	kro, Go doc
JSON	kro	kro, Go doc
Random	kro	kro, Go doc
Maps	kro	kro, Go doc
Index Mutation	kro	kro, Go doc
Omit	kro	kro, Go doc
Strings	cel-go	kro, Go doc
Lists (cel-go)	cel-go	kro, Go doc
Encoders	cel-go	kro, Go doc
Two-Variable Comprehensions	cel-go	kro, Go doc
URLs	kubernetes	kro, Go doc, k8s.io
Regex	kubernetes	kro, Go doc, k8s.io
Quantity	kubernetes	kro, Go doc, k8s.io
Lists (k8s)	kubernetes	kro, Go doc, k8s.io
IP	kubernetes	kro, Go doc, k8s.io
CIDR	kubernetes	kro, Go doc, k8s.io
Semver	kubernetes	kro, Go doc, k8s.io

kro Libraries

Hash

Cryptographic and non-cryptographic hash functions. All return raw bytes; use "%x".format(...) for hex or base64.encode() for base64.

Function	Returns	Description
hash.fnv64a(string)	bytes	FNV-1a 64-bit hash. Fast, non-cryptographic. Recommended for most use cases.
hash.sha256(string)	bytes	SHA-256 cryptographic hash.
hash.md5(string)	bytes	MD5 hash.

Examples:

# SHA-256 as hex string (like sha256sum output)
checksum: ${"%x".format([hash.sha256(schema.spec.config)])}

# FNV-1a as hex (short, fast identifier)
id: ${"%x".format([hash.fnv64a(schema.spec.name + "-" + schema.spec.namespace)])}

# Base64-encoded hash (more compact)
encoded: ${base64.encode(hash.fnv64a(schema.metadata.uid))}

JSON

Parse and serialize JSON strings.

Function	Returns	Description
json.unmarshal(string)	dyn	Parse a JSON string into a CEL value (map, list, string, number, bool, or null).
json.marshal(dyn)	string	Serialize a CEL value to a JSON string.

Examples:

# Parse a JSON string from a ConfigMap or user input
config: ${json.unmarshal(schema.spec.jsonConfig)}

# Access nested fields after parsing
dbHost: ${json.unmarshal(configmap.data.settings).database.host}

# Serialize structured data into a JSON string (e.g. for an annotation or env var)
configJson: ${json.marshal({"name": schema.spec.name, "replicas": schema.spec.replicas})}

Random

Deterministic random generation seeded by a stable value (e.g. resource UID), so results are reproducible across reconcile loops.

Function	Returns	Description
random.seededString(int, string)	string	Random alphanumeric string of given length, seeded by the second argument.
random.seededInt(int, int, string)	int	Random integer in [min, max), seeded by the third argument.

Examples:

# Generate a stable random suffix
suffix: ${random.seededString(8, schema.metadata.uid)}

# Generate a stable random port
port: ${random.seededInt(30000, 32768, schema.metadata.uid)}

Maps

Map manipulation functions.

Function	Returns	Description
<map>.merge(map)	map	Merge two maps. Keys from the second map overwrite keys in the first. Keys must be strings.

Examples:

# Merge user labels with default labels
labels: ${{"app": schema.spec.name, "managed-by": "kro"}.merge(schema.spec.extraLabels)}

# Layer overrides on top of defaults
config: ${{"timeout": "30s", "retries": "3"}.merge(schema.spec.overrides)}

Index Mutation

Pure list functions that return a new list without modifying the input.

Function	Signature	Description
lists.setAtIndex	list(T), int, T -> list(T)	Replace the element at index. Index must be in [0, size(list)).
lists.insertAtIndex	list(T), int, T -> list(T)	Insert value before index. Use index == size(list) to append. Index must be in [0, size(list)].
lists.removeAtIndex	list(T), int -> list(T)	Remove the element at index. Index must be in [0, size(list)).

Examples:

# Replace the second tag
tags: ${lists.setAtIndex(schema.spec.tags, 1, "new-tag")}

# Prepend an environment variable
envVars: ${lists.insertAtIndex(schema.spec.envVars, 0, "DEBUG=true")}

# Remove the first port
ports: ${lists.removeAtIndex(schema.spec.ports, 0)}

# Chain operations: swap first two elements
swapped: ${lists.setAtIndex(lists.setAtIndex(schema.spec.items, 0, schema.spec.items[1]), 1, schema.spec.items[0])}

Omit

The omit() sentinel tells kro to remove a field from the rendered resource. This is useful for conditionally excluding fields.

Function	Returns	Description
omit()	omit	Returns a sentinel value that causes the field to be excluded from the output.

Examples:

# Conditionally include a field
nodeSelector: ${schema.spec.pinToNode ? {"kubernetes.io/hostname": schema.spec.nodeName} : omit()}

note

omit() is only allowed in resource template fields. It is rejected in includeWhen, readyWhen, and forEach expressions.

Runtime

The runtime variable builds and reads instance status conditions. It is only available in the schema's status.conditions block.

Function	Returns	Description
runtime.newCondition({type, status, reason, message})	Condition	Builds a condition. type and status are required; status must be True, False, or Unknown.
runtime.condition(schema, type)	Condition	Reads kro's internal value for a built-in condition type (Ready, InstanceManaged, GraphResolved, ResourcesReady).

Examples:

# Publish an application-level condition
- ${runtime.newCondition({type: 'AppReady', status: deployment.status.readyReplicas > 0 ? 'True' : 'False', reason: 'ReplicaCount', message: ''})}

# Compose kro's lifecycle signal with a domain check
- ${runtime.newCondition({type: 'Ready', status: runtime.condition(schema, 'ResourcesReady').status == 'True' && deployment.status.readyReplicas > 0 ? 'True' : 'False', reason: 'Health', message: ''})}

See Custom Status Conditions for details.

cel-go Libraries

Strings

Extended string manipulation functions from cel-go/ext.

Function	Returns	Description
<string>.charAt(int)	string	Character at position.
<string>.indexOf(string)	int	Index of first occurrence, or -1.
<string>.indexOf(string, int)	int	Same, starting search from offset.
<string>.lastIndexOf(string)	int	Index of last occurrence, or -1.
<string>.lastIndexOf(string, int)	int	Same, with max search position.
<string>.lowerAscii()	string	Lowercase ASCII characters.
<string>.upperAscii()	string	Uppercase ASCII characters.
<string>.replace(string, string)	string	Replace all occurrences.
<string>.replace(string, string, int)	string	Replace up to N occurrences.
<string>.split(string)	list(string)	Split by separator.
<string>.split(string, int)	list(string)	Split with limit.
<string>.substring(int)	string	Substring from position to end.
<string>.substring(int, int)	string	Substring from start (inclusive) to end (exclusive).
<string>.trim()	string	Remove leading and trailing whitespace.
<string>.reverse()	string	Reverse the string.
<list(string)>.join()	string	Concatenate list elements.
<list(string)>.join(string)	string	Concatenate with separator.
<string>.format(list)	string	Printf-style formatting (%s, %d, %f, %e, %b, %x, %X, %o).
strings.quote(string)	string	Escape a string for safe printing.

Examples:

# Printf-style formatting
roleArn: ${"arn:aws:iam::%s:role/%s".format([schema.spec.accountId, schema.spec.roleName])}

# Join list elements with a separator
labelStr: ${schema.spec.tags.join(",")}

# Split a domain name
parts: ${schema.spec.fqdn.split(".")}

# Substring extraction
prefix: ${schema.spec.name.substring(0, 5)}

# Case conversion
lower: ${schema.spec.input.lowerAscii()}
upper: ${schema.spec.input.upperAscii()}

# Search within strings
hasPort: ${schema.spec.endpoint.indexOf(":") >= 0}

# Replace characters
sanitized: ${schema.spec.name.replace("_", "-")}

# Trim whitespace
cleaned: ${schema.spec.input.trim()}

# Hex encoding via %x (useful for hash output)
checksum: ${"%x".format([hash.sha256(schema.spec.data)])}

Lists (cel-go)

Extended list manipulation functions from cel-go/ext.

Function	Returns	Description
<list>.slice(int, int)	list	Sub-list from start (inclusive) to end (exclusive).
<list>.flatten()	list	Flatten nested lists one level deep.
<list>.flatten(int)	list	Flatten up to N levels.
<list>.sort()	list	Sort comparable elements.
<list>.sortBy(var, keyExpr)	list	Sort by a key expression.
<list>.distinct()	list	Remove duplicate elements.
<list>.reverse()	list	Reverse element order.
lists.range(int)	list(int)	Generate [0, 1, ..., n-1].

Examples:

# Slice a list
subset: ${schema.spec.items.slice(0, 3)}

# Sort and deduplicate
tags: ${schema.spec.tags.sort().distinct()}

# Sort by a field
ordered: ${items.sortBy(i, i.data.priority)}

# Generate index range
indices: ${lists.range(schema.spec.count)}

# Flatten nested lists
flat: ${schema.spec.nestedPorts.flatten()}

Encoders

Base64 encoding and decoding from cel-go/ext.

Function	Returns	Description
base64.encode(bytes)	string	Encode bytes to base64 string.
base64.decode(string)	bytes	Decode base64 string to bytes.

Examples:

# Encode a hash to base64
encoded: ${base64.encode(hash.sha256(schema.spec.data))}

# Encode a string as base64 (e.g. for a Kubernetes Secret)
secret: ${base64.encode(bytes(schema.spec.password))}

# Decode a base64-encoded string
decoded: ${string(base64.decode(schema.spec.encodedValue))}

Two-Variable Comprehensions

Two-variable versions of list/map comprehension macros from cel-go/ext. These provide access to both the key/index and value in each iteration.

Macro	Description
<list|map>.all(k, v, predicate)	True if all entries satisfy the predicate.
<list|map>.exists(k, v, predicate)	True if any entry satisfies the predicate.
<list|map>.existsOne(k, v, predicate)	True if exactly one entry satisfies the predicate.
<list|map>.transformList(k, v, transform)	Convert map/list to a list by evaluating transform for each entry.
<list|map>.transformList(k, v, filter, transform)	Same with a filter predicate.
<list|map>.transformMap(k, v, transform)	Transform values while preserving keys.
<list|map>.transformMap(k, v, filter, transform)	Same with a filter predicate.
<list|map>.transformMapEntry(k, v, transform)	Build new key-value pairs (transform must return a single-entry map).
<list|map>.transformMapEntry(k, v, filter, transform)	Same with a filter predicate.

Examples:

# Extract keys from a map
keys: ${{ "app": "nginx", "version": "1.19" }.transformList(k, v, k)}
# -> ["app", "version"]

# Transform map values
scaled: ${{ "cpu": 2, "memory": 4 }.transformMap(k, v, v * 2)}
# -> {"cpu": 4, "memory": 8}

# Filter and transform
filtered: ${{ "a": 1, "b": 5, "c": 3 }.transformMap(k, v, v > 1, v * 10)}
# -> {"b": 50, "c": 30}

# Swap keys and values
swapped: ${{ "us-east-1": "primary", "eu-west-1": "secondary" }.transformMapEntry(k, v, {v: k})}
# -> {"primary": "us-east-1", "secondary": "eu-west-1"}

# Build formatted strings from key-value pairs
envVars: ${{ "APP": "nginx", "PORT": "8080" }.transformList(k, v, k + "=" + v)}
# -> ["APP=nginx", "PORT=8080"]

# Two-variable all/exists on a map
allDifferent: ${{ "hello": "world", "taco": "bell" }.all(k, v, k != v)}

Kubernetes Libraries

URLs

Parse and inspect URLs. The URL must be an absolute URI or an absolute path.

Function	Returns	Description
url(string)	URL	Parse a string into a URL. Errors if invalid.
isURL(string)	bool	Returns true if the string is a valid URL.
<URL>.getScheme()	string	Returns the URL scheme (e.g. https). Empty string if absent.
<URL>.getHost()	string	Returns the host including port (e.g. example.com:80). IPv6 addresses are bracketed.
<URL>.getHostname()	string	Returns the hostname without port. IPv6 addresses returned without brackets.
<URL>.getPort()	string	Returns the port, or empty string if not specified.
<URL>.getEscapedPath()	string	Returns the URL-escaped path.
<URL>.getQuery()	map(string, list(string))	Returns query parameters as a map.

Examples:

# Extract the host from an endpoint URL
host: ${url(schema.spec.endpoint).getHostname()}

# Validate a URL
valid: ${isURL(schema.spec.callbackUrl)}

# Build a connection using URL parts
port: ${url(database.status.endpoint).getPort()}

Regex

Regular expression matching and extraction via k8s.io/apiserver/pkg/cel/library.

Function	Returns	Description
<string>.find(string)	string	Returns the first regex match, or empty string.
<string>.findAll(string)	list(string)	Returns all non-overlapping matches.
<string>.findAll(string, int)	list(string)	Returns up to N matches.

note

The built-in CEL <string>.matches(string) function is always available (it's part of the CEL standard library, not this extension). It returns true if the entire string matches the regex.

Examples:

# Validate a resource name pattern
valid: ${schema.spec.name.matches('^[a-z][a-z0-9-]*$')}

# Extract version numbers from a string
versions: ${"v1.2.3 and v4.5.6".findAll('[0-9]+\\.[0-9]+\\.[0-9]+')}

# Find first match
version: ${schema.spec.image.find('[0-9]+\\.[0-9]+\\.[0-9]+')}

Quantity

Work with Kubernetes resource quantities (e.g. 100m, 1Gi, 500Mi).

Function	Returns	Description
quantity(string)	Quantity	Parse a Kubernetes quantity string.
isQuantity(string)	bool	True if the string is a valid quantity.
<Quantity>.add(Quantity|int)	Quantity	Sum of two quantities.
<Quantity>.sub(Quantity|int)	Quantity	Difference of two quantities.
<Quantity>.isLessThan(Quantity)	bool	Comparison.
<Quantity>.isGreaterThan(Quantity)	bool	Comparison.
<Quantity>.compareTo(Quantity)	int	Returns -1, 0, or 1.
<Quantity>.asInteger()	int	Scaled value as integer. Errors on overflow or precision loss.
<Quantity>.asApproximateFloat()	float	Approximate float value.
<Quantity>.isInteger()	bool	True if the quantity can be represented as an integer without loss.
<Quantity>.sign()	int	Returns 1, -1, or 0.

Examples:

# Compare memory requests
needsMore: ${quantity(schema.spec.memoryRequest).isLessThan(quantity('1Gi'))}

# Arithmetic on quantities
total: ${quantity(schema.spec.cpuRequest).add(quantity('100m'))}
remaining: ${quantity('2Gi').sub(quantity(schema.spec.memoryRequest))}

# Validate a resource quantity
valid: ${isQuantity(schema.spec.cpuLimit)}

# Convert whole-number quantities to integer
megabytes: ${quantity(schema.spec.storageSize).asInteger()}

Lists (k8s)

Kubernetes-specific list utility functions from k8s.io/apiserver/pkg/cel/library.

Function	Returns	Description
<list>.isSorted()	bool	True if elements are in sorted order.
<list>.sum()	T	Sum of numeric or duration elements. Returns 0 for empty lists.
<list>.min()	T	Minimum element. Errors on empty list.
<list>.max()	T	Maximum element. Errors on empty list.
<list>.indexOf(T)	int	Index of first occurrence, or -1.
<list>.lastIndexOf(T)	int	Index of last occurrence, or -1.

Examples:

# Check if ports are sorted
sorted: ${schema.spec.ports.isSorted()}

# Sum all replica counts
totalReplicas: ${schema.spec.replicaCounts.sum()}

# Find min/max
minPort: ${schema.spec.ports.min()}
maxPort: ${schema.spec.ports.max()}

# Find element index
idx: ${schema.spec.tags.indexOf("production")}

IP

Parse and inspect IPv4 and IPv6 addresses. IPv4-mapped IPv6 addresses and addresses with zones are not allowed.

Function	Returns	Description
ip(string)	IP	Parse a string into an IP address. Errors if invalid.
isIP(string)	bool	Returns true if the string is a valid IPv4 or IPv6 address.
ip.isCanonical(string)	bool	Returns true if the string is in canonical IP form.
<IP>.family()	int	Returns 4 or 6.
<IP>.isLoopback()	bool	True for 127.x.x.x (v4) or ::1 (v6).
<IP>.isGlobalUnicast()	bool	True for globally routable addresses.
<IP>.isLinkLocalUnicast()	bool	True for link-local unicast (169.254.x.x / fe80::/10).
<IP>.isLinkLocalMulticast()	bool	True for link-local multicast (224.0.0.x / ff00::/8).
<IP>.isUnspecified()	bool	True for 0.0.0.0 or ::.
string(<IP>)	string	Convert back to string.

Examples:

# Validate an IP address from user input
ready: ${isIP(schema.spec.clusterIP)}

# Check IP family for dual-stack logic
family: ${string(ip(schema.spec.podIP).family())}

# Gate on global unicast
isRoutable: ${ip(schema.spec.address).isGlobalUnicast()}

# Canonical form check
canonical: ${ip.isCanonical(schema.spec.ipv6Addr)}

CIDR

Parse and inspect CIDR subnet notation. Works with both IPv4 and IPv6.

Function	Returns	Description
cidr(string)	CIDR	Parse a CIDR string (e.g. 192.168.0.0/24). Errors if invalid.
isCIDR(string)	bool	Returns true if the string is valid CIDR notation.
<CIDR>.containsIP(string|IP)	bool	True if the CIDR contains the given IP.
<CIDR>.containsCIDR(string|CIDR)	bool	True if the CIDR fully contains the given subnet.
<CIDR>.ip()	IP	Returns the address part of the CIDR.
<CIDR>.prefixLength()	int	Returns the prefix length in bits.
<CIDR>.masked()	CIDR	Returns the CIDR in canonical masked form.
string(<CIDR>)	string	Convert back to string.

Examples:

# Check if a pod IP falls within the service CIDR
inRange: ${cidr(schema.spec.serviceCIDR).containsIP(pod.status.podIP)}

# Validate that a user-provided CIDR is valid
valid: ${isCIDR(schema.spec.subnetCIDR)}

# Extract prefix length
prefixLen: ${string(cidr(schema.spec.subnetCIDR).prefixLength())}

# Check subnet containment
isSubnet: ${cidr('10.0.0.0/8').containsCIDR(schema.spec.vpcCIDR)}

Semver

Parse and compare semantic versions. Supports optional normalization to handle common non-strict formats like v1.0 or 01.02.03.

Function	Returns	Description
semver(string)	Semver	Parse a strict semver string (e.g. 1.2.3).
semver(string, bool)	Semver	Parse with normalization (strips v prefix, fills missing parts, removes leading zeros).
isSemver(string)	bool	Returns true if the string is a valid semver.
isSemver(string, bool)	bool	Same with optional normalization.
<Semver>.major()	int	Major version number.
<Semver>.minor()	int	Minor version number.
<Semver>.patch()	int	Patch version number.
<Semver>.isGreaterThan(Semver)	bool	True if receiver > operand.
<Semver>.isLessThan(Semver)	bool	True if receiver < operand.
<Semver>.compareTo(Semver)	int	Returns -1, 0, or 1.

Examples:

# Validate a version string
validVersion: ${isSemver(schema.spec.version)}

# Compare versions
ready: ${semver(schema.spec.version).compareTo(semver('2.0.0')) >= 0}

# Extract major version for image tag
majorTag: ${string(semver(schema.spec.appVersion).major())}

# Tolerant parsing with normalization (accepts "v1.0", "01.02.03", etc.)
valid: ${isSemver(schema.spec.version, true)}
parsed: ${semver(schema.spec.version, true).minor()}

For the complete CEL language reference, see the CEL language definitions.