MODEL DEFINITIONS
A model definition is a YAML file that configures an instance of a model type. Definitions are stored in one of two locations depending on how they are created.
To inspect a model type's methods, arguments, and output specs before creating a
definition, see model type describe below.
Definition Storage
models/ (via model create)
swamp model create command/shell my-shell --global-arg prefix="swamp says"Definitions created with model create are stored in models/{type}/{id}.yaml.
They are git-tracked and included in model search results.
models/
command/
shell/
83e5906c-3d3f-4ffe-84f9-5e01f6e9037c.yaml
53274ce2-c390-412b-93ac-48b205a13f4e.yamlThe directory path encodes the model type. The filename is the definition's UUID.
.swamp/auto-definitions/ (via direct type execution)
swamp model @command/shell method run execute my-shell --input 'run=echo hello'When a definition with the given name does not exist, one is created
automatically in .swamp/auto-definitions/{type}/{id}.yaml. On subsequent runs,
the existing definition is reused after verifying the type matches. --input
values are split between global arguments and method arguments using the type's
schemas — method arguments take precedence on ambiguous keys. Unknown keys are
rejected with an error listing valid inputs.
.swamp/
auto-definitions/
command/
shell/
a1b2c3d4-e5f6-7890-abcd-ef1234567890.yamlThe same directory and filename conventions apply. Auto-definitions are not
git-tracked and are not included in model search results. They are findable by
model get and are included in DEFAULT_DATASTORE_SUBDIRS for datastore sync.
Comparison
| Property | models/ |
.swamp/auto-definitions/ |
|---|---|---|
| Created by | swamp model create |
swamp model @type method run |
| Git-tracked | Yes | No |
Included in model search |
Yes | No |
Findable by model get |
Yes | Yes |
| Value source | Definition YAML file | --input flags at runtime |
| CEL expressions in globalArguments | Supported | Not supported |
| Datastore sync | Yes | Yes |
Top-Level Fields
type: command/shell
typeVersion: 2026.02.09.1
id: 83e5906c-3d3f-4ffe-84f9-5e01f6e9037c
name: hello-world
version: 1
tags: {}
globalArguments: {}
methods: {}Optional fields omitted above: inputs, checks, reports. These appear in
the YAML only when explicitly set.
type
Model type identifier.
| Property | Value |
|---|---|
| Type | string |
| Required | No |
| Default | Inferred from directory path |
The type value is normalized to a lowercase, slash-delimited path. Swamp
applies these transformations:
AWS::EC2::VPC→aws/ec2/vpcdocker run→docker/runMicrosoft.Resources/resourceGroup→microsoft/resources/resourcegroup
Consecutive, leading, and trailing slashes are removed.
typeVersion
CalVer version of the model type at definition creation time.
| Property | Value |
|---|---|
| Type | string |
| Required | No |
| Default | Set to the model type's current version on create |
| Format | YYYY.MM.DD.MICRO (e.g., 2026.02.09.1) |
Tracks which version of the model type schema this definition was created against. Used for detecting when model type upgrades are available.
id
Unique identifier for the definition.
| Property | Value |
|---|---|
| Type | string (UUID v4) |
| Required | Yes |
| Default | Auto-generated on create |
| Format | xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx |
name
Human-readable name for the definition.
| Property | Value |
|---|---|
| Type | string |
| Required | Yes |
| Default | None |
Constraints:
- Minimum 1 character.
- Must be unique within the repository.
- Must not contain
..,\, or null bytes (path traversal protection). /is only allowed in scoped names matching@[a-z0-9_-]+/[a-z0-9_-]+(\/[a-z0-9_-]+)*(e.g.,@john/pod-inventory).
version
Definition version number.
| Property | Value |
|---|---|
| Type | integer |
| Required | No |
| Default | 1 |
Must be a positive integer. Incremented when the definition is modified. This
tracks changes to the definition itself — it is separate from typeVersion,
which tracks the model type schema.
tags
Key-value labels attached to the definition.
| Property | Value |
|---|---|
| Type | Record<string, string> |
| Required | No |
| Default | {} |
Tags flow to data produced by the definition and can be overridden at runtime
with --tag key=value.
tags:
env: production
team: platform
owner: aliceglobalArguments
Shared arguments available to all methods.
| Property | Value |
|---|---|
| Type | Record<string, unknown> |
| Required | No |
| Default | {} |
Values can be any JSON-serializable type: string, number, boolean, object,
array, or null. CEL expressions are supported using the ${{ }} syntax.
globalArguments:
repo: systeminit/swamp
retries: 3
verbose: true
computed: "${{ 'https://' + self.tags.domain }}"
secret: "${{ vault.get('my-vault', 'api-key') }}"Global arguments are validated against the model type's globalArguments schema
(if one is defined) when the definition is validated.
A global argument whose schema marks it sensitive must not be given a literal
value. A literal is rejected when the definition is written — across
model create, model edit, model run / workflow run auto-definitions, and
the serve HTTP API — because it would be stored in cleartext in the definition
YAML. The value must instead be a vault.get('<vault>', '<key>') expression.
The write fails with:
Global argument 'apiKey' is marked sensitive and cannot be set to a literal value — it would be stored in cleartext in the definition YAML. Store the secret in a vault and reference it with a vault.get expression, e.g. --global-arg "apiKey=${{ vault.get('my-vault', 'api-key') }}".See the Vaults and CEL Expressions references for the expression syntax.
methods
Per-method argument overrides.
| Property | Value |
|---|---|
| Type | Record<string, MethodData> |
| Required | No |
| Default | {} |
Each key is a method name. The value is a MethodData object:
methods:
execute:
arguments:
run: "echo hello"
timeout: 5000
env:
MY_VAR: valueMethodData
| Field | Type | Required | Default |
|---|---|---|---|
arguments |
Record<string, unknown> |
No | {} |
arguments values support CEL expressions. They are merged with the model
type's method argument schema and validated at execution time.
inputs
JSON Schema definition for structured inputs that can be provided at runtime.
| Property | Value |
|---|---|
| Type | InputsSchema |
| Required | No |
| Default | None |
InputsSchema
| Field | Type | Required |
|---|---|---|
type |
"object" |
No |
properties |
Record<string, JsonSchemaProperty> |
No |
required |
string[] |
No |
additionalProperties |
boolean or JsonSchemaProperty |
No |
Additional JSON Schema keywords are allowed and preserved.
JsonSchemaProperty
| Field | Type | Description |
|---|---|---|
type |
"string", "number", "integer", "boolean", "array", "object" |
Value type |
description |
string |
Human-readable description |
default |
any | Default value |
enum |
unknown[] |
Allowed values |
items |
JsonSchemaProperty |
Schema for array items |
properties |
Record<string, JsonSchemaProperty> |
Nested object properties |
required |
string[] |
Required nested properties |
additionalProperties |
boolean or JsonSchemaProperty |
Extra property handling |
All fields are optional. Properties nest recursively for objects and arrays.
inputs:
type: object
properties:
environment:
type: string
enum: [dev, staging, production]
count:
type: integer
default: 1
config:
type: object
properties:
verbose:
type: boolean
default: false
additionalProperties: true
required:
- environmentInput values are available in CEL expressions as inputs.*.
checks
Pre-flight check selection.
| Property | Value |
|---|---|
| Type | CheckSelection |
| Required | No |
| Default | None |
CheckSelection
| Field | Type | Required | Default |
|---|---|---|---|
require |
string[] |
No | [] |
skip |
string[] |
No | [] |
require lists check names to include. skip lists check names to exclude.
Checks run before mutating methods (create, update, delete, action). Check names
must be defined by the model type.
checks:
require:
- validate-permissions
- check-resources
skip:
- expensive-validationreports
Post-execution report selection.
| Property | Value |
|---|---|
| Type | ReportSelection |
| Required | No |
| Default | None |
ReportSelection
| Field | Type | Required | Default |
|---|---|---|---|
require |
ReportRef[] |
No | [] |
skip |
string[] |
No | [] |
A ReportRef is either a report name (string) or an object:
| Field | Type | Required | Description |
|---|---|---|---|
name |
string |
Yes | Report name |
methods |
string[] |
No | Restrict to specific methods |
When methods is omitted, the report runs for all methods.
reports:
require:
- summary-report
- name: detailed-analysis
methods:
- execute
skip:
- verbose-debugA name in require that cannot be resolved to a loaded or installed report
(missing extension pull, typo, broken bundle) logs a warning naming the report
and the remedy, writes a searchable error artifact (visible via
swamp report search), and counts as a report failure —
swamp model method run exits non-zero. A name in skip that also appears in
require is exempted from the resolution check — skip wins over require.
CEL Expressions
String values in globalArguments and methods.*.arguments support CEL
expressions using the ${{ }} wrapper:
globalArguments:
url: "${{ 'https://' + self.tags.domain }}"
pool: "${{ model.config.definition.globalArguments.pool_size * 2 }}"
secret: "${{ vault.get('infra', 'api-key') }}"When the entire YAML value is a single ${{ }} expression, the result preserves
its type. When mixed with surrounding text, the result is coerced to a string.
Context Variables
| Variable | Description |
|---|---|
self |
The current definition: self.name, self.version, self.tags.*, self.globalArguments.* |
model |
Other definitions: model.<name>.definition.globalArguments.*, model.<name>.resource.<spec>.<instance>.attributes.* |
inputs |
Runtime input values: inputs.<property> |
env |
Process environment variables: env.<VAR_NAME> |
vault |
Secrets: vault.get('<vault>', '<key>') |
data |
Versioned data access: data.latest(...), data.version(...), data.findByTag(...) |
file |
File contents: file.contents('<model>', '<spec>') |
See the CEL Expressions reference for the full expression language.
Per-Model Method Locking
Concurrent swamp model method run invocations on the same model definition
serialize automatically. The engine acquires a per-model file lock before
executing the method and releases it when the method completes, fails, or is
cancelled. Different model definitions run in parallel — only runs targeting the
same definition contend.
The lock uses a ~30-second TTL with a heartbeat. The executing process renews the lock periodically; if the process dies without releasing the lock, stale-lock detection checks whether the owning PID is still alive and reclaims the lock once the TTL expires. This prevents a crashed process from permanently blocking a model.
When a second run cannot acquire the lock within the timeout window, the CLI
exits with code 75 and a lock_timeout error (see Exit codes
and Error codes). The caller can retry with
exponential backoff.
Breakglass: inspecting and releasing locks
If a lock is stuck — the owning process is gone but the TTL has not yet expired, or the PID check cannot reach the original host — use the datastore lock commands:
swamp datastore lock status # show who holds the lock
swamp datastore lock release --force # release the global lock
swamp datastore lock release --force --model <type/name> # release a specific model's lockSee Operational Commands for the full flag reference.
Execution Cancellation
A running model method can be cancelled via swamp model cancel. Cancellation
terminates the method run cooperatively and marks it as cancelled — a terminal
status alongside succeeded and failed.
Commands
| Command | Effect |
|---|---|
swamp model cancel <model> |
Cancel the running method run on the named model. |
swamp model cancel --all |
Cancel all running model method runs in the repository. |
swamp model cancel <model> --reason <text> |
Cancel with an audit-trail reason recorded on the run. |
Signal-based cancellation
Pressing Ctrl+C (SIGINT) or sending SIGTERM during swamp model method run
cancels the running method. The CLI intercepts the signal, sends a cancel
request, waits for the method to terminate cooperatively, and exits with a
non-zero status.
Examples
$ swamp model cancel my-server
$ swamp model cancel --all
$ swamp model cancel my-server --reason 'No longer needed'Validation
swamp model validate checks a definition against its model type schema:
$ swamp model validate greeter
Validating: greeter (command/shell)
✓ Definition schema
✓ Global arguments
✓ Method arguments
✓ Expression paths
✓ Check selection
Summary: 5/5 validations passed
Result: PASSEDOmit the name to validate all definitions in the repository.
Validation rules:
idmust be a valid UUID v4.namemust be at least 1 character, unique within the repository, with no path traversal sequences.versionmust be a positive integer.tagsvalues must be strings.globalArgumentsmust match the model type's global argument schema (if defined).- A
sensitiveglobal argument must be avault.get(...)expression, not a literal value. methods.*.argumentsmust match the model type's method argument schema.typeVersionmust be valid CalVer (YYYY.MM.DD.MICRO) or absent.- CEL expression syntax in
${{ }}wrappers must parse correctly. - Expression paths must reference existing models and valid schema paths.
checks.requireandchecks.skipnames must exist on the model type.
Complete Example
type: command/shell
typeVersion: 2026.02.09.1
id: 53274ce2-c390-412b-93ac-48b205a13f4e
name: greeter
version: 1
tags:
env: dev
team: platform
globalArguments:
prefix: "swamp says"
methods:
execute:
arguments:
run: "echo \"${{ self.globalArguments.prefix }}: hello!\""
inputs:
type: object
properties:
name:
type: string
description: Name to greet
default: world
required:
- name
checks:
require:
- validate-permissions
reports:
require:
- name: execution-summary
methods:
- executemodel type describe
swamp model type describe prints the schema of a model type — its methods,
arguments, data output specifications, and version. It accepts a type identifier
(e.g., command/shell, @swamp/issue-lifecycle) and renders a human-readable
summary by default.
$ swamp model type describe command/shell
Type: command/shell
Version: 2026.02.09.1
Data Outputs:
result [resource] - Shell command execution result (exit code, timing, command) (infinite)
log [file] - Shell command output (stdout and stderr) (text/plain, infinite)
Methods:
execute - Execute the shell command and capture stdout, stderr, and exit code
Inputs:
run (string) *required
workingDir (string)
timeout (integer)
env (object)
ignoreExitCode (boolean)For the YAML definition layer that configures model instances, see Definition Storage above.
--json
Produces structured JSON output. The full JSON includes complete JSON Schema definitions for every method's arguments and every data output spec's schema.
$ swamp model type describe command/shell --json{
"type": {
"raw": "command/shell",
"normalized": "command/shell"
},
"version": "2026.02.09.1",
"dataOutputSpecs": [
{
"specName": "result",
"kind": "resource",
"description": "Shell command execution result (exit code, timing, command)",
"schema": { ... },
"lifetime": "infinite",
"garbageCollection": 10
},
{
"specName": "log",
"kind": "file",
"description": "Shell command output (stdout and stderr)",
"contentType": "text/plain",
"lifetime": "infinite",
"garbageCollection": 10,
"streaming": true
}
],
"methods": [
{
"name": "execute",
"description": "Execute the shell command and capture stdout, stderr, and exit code",
"arguments": {
"$schema": "https://json-schema.org/draft/2020-12/schema",
"type": "object",
"properties": {
"run": {
"type": "string",
"minLength": 1,
"description": "The shell command to execute"
},
...
},
"required": ["run"],
"additionalProperties": false
}
}
]
}Full JSON shape
| Field | Type | Description |
|---|---|---|
type.raw |
string |
Type identifier as provided. |
type.normalized |
string |
Lowercased, slash-delimited form. |
version |
string |
CalVer version of the model type. |
dataOutputSpecs |
array |
Full output spec objects including schema, lifetime, kind. |
methods |
array |
Method entries with full JSON Schema arguments. |
Each dataOutputSpecs entry includes the complete JSON Schema for resource
specs, or contentType and streaming fields for file specs.
Each methods entry includes the full JSON Schema for that method's arguments,
including description, type constraints, and validation rules for every
property.
--compact
Produces a minimal digest suited to agent discovery. Strips full JSON Schema definitions down to property names and types, and reduces data output specs to names only.
$ swamp model type describe command/shell --compact --json{
"type": {
"raw": "command/shell",
"normalized": "command/shell"
},
"version": "2026.02.09.1",
"dataOutputSpecs": [
"result",
"log"
],
"methods": [
{
"name": "execute",
"description": "Execute the shell command and capture stdout, stderr, and exit code",
"arguments": {
"properties": {
"run": { "type": "string" },
"workingDir": { "type": "string" },
"timeout": { "type": "integer" },
"env": { "type": "object" },
"ignoreExitCode": { "type": "boolean" }
},
"required": ["run"]
}
}
]
}Compact JSON shape
| Field | Type | Description |
|---|---|---|
type.raw |
string |
Type identifier as provided. |
type.normalized |
string |
Lowercased, slash-delimited form. |
version |
string |
CalVer version of the model type. |
dataOutputSpecs |
array |
Spec names only (strings, not objects). |
methods |
array |
Method entries with simplified arguments (names and types). |
The compact form omits JSON Schema metadata ($schema, description,
minLength, additionalProperties, constraints) from method arguments, and
replaces full data output spec objects with their specName strings.
model type search
swamp model type search searches for model types available in the repository's
installed extensions.
swamp model type search [query]| Argument | Required | Description |
|---|---|---|
query |
No | Text to match against type names |
| Flag | Description |
|---|---|
--repo-dir |
Repository directory (env: SWAMP_REPO_DIR; not required for type search) |
$ swamp model type search shellReturns matching model types with their version and description. Use --json
for structured output.
CLI Commands
swamp model create
Create a new model definition.
swamp model create <type> <name>| Argument | Required | Description |
|---|---|---|
type |
Yes | Model type identifier (e.g., command/shell) |
name |
Yes | Human-readable name for the definition |
| Flag | Description |
|---|---|
--repo-dir |
Repository directory (env: SWAMP_REPO_DIR) |
--global-arg |
Set a global argument (repeatable, key=value) |
--server |
Swamp server URL |
--token |
Authentication token |
Creates a definition file in models/{type}/{id}.yaml. See
Definition Storage for the directory layout.
$ swamp model create command/shell greeter --global-arg prefix="swamp says"swamp model get
Show details of a model definition.
swamp model get <model_id_or_name>| Argument | Required | Description |
|---|---|---|
model_id_or_name |
Yes | Definition name or UUID |
| Flag | Description |
|---|---|
--repo-dir |
Repository directory (env: SWAMP_REPO_DIR) |
--server |
Swamp server URL |
--token |
Authentication token |
Use --json for structured output. Resolves definitions from both models/ and
.swamp/auto-definitions/.
swamp model search
Search model definitions in the repository.
swamp model search [query]| Argument | Required | Description |
|---|---|---|
query |
No | Text to match against definition names |
| Flag | Description |
|---|---|
--repo-dir |
Repository directory (env: SWAMP_REPO_DIR) |
--server |
Swamp server URL |
--token |
Authentication token |
Returns definitions stored in models/. Auto-definitions in
.swamp/auto-definitions/ are excluded. Use --json for structured output.
swamp model edit
Open a model definition in $EDITOR.
swamp model edit [model_id_or_name]| Argument | Required | Description |
|---|---|---|
model_id_or_name |
No | Definition name or UUID |
| Flag | Description |
|---|---|
--repo-dir |
Repository directory (env: SWAMP_REPO_DIR) |
When the name is omitted, Swamp presents a selector. The definition is validated after the editor exits.
swamp model delete
Delete a model definition and its related artifacts.
swamp model delete <model_id_or_name>| Argument | Required | Description |
|---|---|---|
model_id_or_name |
Yes | Definition name or UUID |
| Flag | Description |
|---|---|
--repo-dir |
Repository directory (env: SWAMP_REPO_DIR) |
-f, --force |
Skip confirmation prompt |
--server |
Swamp server URL |
--token |
Authentication token |
swamp model validate
Validate a definition against its model type schema.
swamp model validate [model_id_or_name]| Argument | Required | Description |
|---|---|---|
model_id_or_name |
No | Definition name or UUID |
| Flag | Description |
|---|---|
--repo-dir |
Repository directory (env: SWAMP_REPO_DIR) |
--all |
Validate all definitions in the repository |
--server |
Swamp server URL |
--token |
Authentication token |
When the name is omitted and --all is not set, Swamp presents a selector. See
Validation above for the rules checked.
swamp model evaluate
Evaluate CEL expressions in a definition without executing any method.
swamp model evaluate [model_id_or_name]| Argument | Required | Description |
|---|---|---|
model_id_or_name |
No | Definition name or UUID |
| Flag | Description |
|---|---|
--repo-dir |
Repository directory (env: SWAMP_REPO_DIR) |
--all |
Evaluate all definitions in the repository |
--server |
Swamp server URL |
--token |
Authentication token |
Resolves all ${{ }} expressions in globalArguments and methods.*.arguments
and prints the evaluated values.
swamp model cancel
Cancel a running method run.
swamp model cancel [model_id_or_name]| Argument | Required | Description |
|---|---|---|
model_id_or_name |
No | Definition name or UUID |
| Flag | Description |
|---|---|
--repo-dir |
Repository directory (env: SWAMP_REPO_DIR) |
--all |
Cancel all running method runs |
--reason |
Audit-trail reason recorded on the run |
See Execution Cancellation for signal-based cancellation and examples.
Method Execution
swamp model method run
Execute a method on a model. This is the single most important command in Swamp.
swamp model method run <model_or_type> <method> [definition_name]| Argument | Required | Description |
|---|---|---|
model_or_type |
Yes | Definition name or @type identifier |
method |
Yes | Method name to execute |
definition_name |
No | Name for auto-definition when using @type (see auto-definitions) |
| Flag | Description |
|---|---|
--repo-dir |
Repository directory (env: SWAMP_REPO_DIR) |
--last-evaluated |
Reuse the last evaluated definition |
--input |
Method or global argument (repeatable, key=value) |
--input-file |
Read inputs from a JSON or YAML file |
--stdin |
Read inputs from stdin |
--tag |
Runtime tag override (repeatable, key=value) |
--skip-check |
Skip a check by name (repeatable) |
--skip-check-label |
Skip checks by label (repeatable) |
--skip-checks |
Skip all checks |
--skip-reports |
Skip all reports |
--skip-report |
Skip a report by name (repeatable) |
--skip-report-label |
Skip reports by label (repeatable) |
--report |
Run a specific report (repeatable) |
--report-label |
Run reports by label (repeatable) |
--timeout |
Method execution timeout |
--server |
Swamp server URL |
--token |
Authentication token |
Exit codes
| Code | Meaning |
|---|---|
0 |
Success |
1 |
General error |
75 |
Lock contention — retry with exponential backoff |
Error codes (JSON on stderr)
| Code | Description |
|---|---|
lock_timeout |
Another method run holds the model lock |
model_not_found |
Definition does not exist |
unknown_model_type |
Model type is not installed |
unknown_method |
Method name not found on the type |
no_evaluated_definition |
No evaluated definition available |
missing_deps |
Required dependencies not satisfied |
method_execution_failed |
Method ran but failed |
not_authenticated |
Authentication required |
cancelled |
Run was cancelled |
Example
$ swamp model method run greeter execute
Running method: execute
Model: greeter (command/shell)
...
$ swamp model method run @command/shell execute my-shell \
--input 'run=echo hello' --tag env=devswamp model method describe
Describe a method with its arguments, types, and constraints.
swamp model method describe <model_id_or_name> <method>| Argument | Required | Description |
|---|---|---|
model_id_or_name |
Yes | Definition name or UUID |
method |
Yes | Method name |
| Flag | Description |
|---|---|
--repo-dir |
Repository directory (env: SWAMP_REPO_DIR) |
--server |
Swamp server URL |
--token |
Authentication token |
Use --json for the full JSON Schema of the method's arguments.
swamp model method history get
Show details of a model method run.
swamp model method history get <output_id_or_model_name>| Argument | Required | Description |
|---|---|---|
output_id_or_model_name |
Yes | Output ID or model name (latest run) |
| Flag | Description |
|---|---|
--repo-dir |
Repository directory (env: SWAMP_REPO_DIR) |
--server |
Swamp server URL |
--token |
Authentication token |
When a model name is given, shows the most recent run for that model.
swamp model method history search
Search method run history.
swamp model method history search [query]| Argument | Required | Description |
|---|---|---|
query |
No | Text to match against run history |
| Flag | Description |
|---|---|
--repo-dir |
Repository directory (env: SWAMP_REPO_DIR) |
--server |
Swamp server URL |
--token |
Authentication token |
Use --json for structured output.
swamp model method history logs
Show logs for a method run.
swamp model method history logs <output_id_or_model_name>| Argument | Required | Description |
|---|---|---|
output_id_or_model_name |
Yes | Output ID or model name (latest run) |
| Flag | Description |
|---|---|
--repo-dir |
Repository directory (env: SWAMP_REPO_DIR) |
--tail |
Show only the last N lines |
--server |
Swamp server URL |
--token |
Authentication token |
Output Management
swamp model output get
Show details of a model output.
swamp model output get <output_id_or_model_name>| Argument | Required | Description |
|---|---|---|
output_id_or_model_name |
Yes | Output ID or model name (latest output) |
| Flag | Description |
|---|---|
--repo-dir |
Repository directory (env: SWAMP_REPO_DIR) |
--server |
Swamp server URL |
--token |
Authentication token |
When a model name is given, shows the most recent output for that model.
Example
$ swamp model output get greeter
$ swamp model output get greeter --jsonswamp model output search
Search model outputs.
swamp model output search [query]| Argument | Required | Description |
|---|---|---|
query |
No | Text to match against outputs |
| Flag | Description |
|---|---|
--repo-dir |
Repository directory (env: SWAMP_REPO_DIR) |
--server |
Swamp server URL |
--token |
Authentication token |
Use --json for structured output.
swamp model output logs
Show log artifact content for an output.
swamp model output logs <output_id>| Argument | Required | Description |
|---|---|---|
output_id |
Yes | Output ID |
| Flag | Description |
|---|---|
--repo-dir |
Repository directory (env: SWAMP_REPO_DIR) |
--tail |
Show only the last N lines |
--server |
Swamp server URL |
--token |
Authentication token |
swamp model output data
Show data artifact content for an output.
swamp model output data <output_id>| Argument | Required | Description |
|---|---|---|
output_id |
Yes | Output ID |
| Flag | Description |
|---|---|
--repo-dir |
Repository directory (env: SWAMP_REPO_DIR) |
--field |
Extract a specific field from the data |
--name |
Filter by data output spec name |
--server |
Swamp server URL |
--token |
Authentication token |