Config File Reference

Full YAML schema for lmn.yaml.

Top-level fields

run:       # Request configuration
execution: # How many requests / how long
thresholds: # Pass/fail rules (optional)

run

Field	Type	Required	Description
host	string	Yes*	Target URL including scheme. * Required unless passed via -H
method	string	No	HTTP method. Default: get. Values: get, post, put, patch, delete
headers	map	No	Key-value headers. Values support ${ENV_VAR} substitution
request_template	path	No	Path to a JSON request body template
response_template	path	No	Path to a JSON response extraction template
alias	string	No	Name of a saved request template alias
output	string	No	Output format. Values: table (default), json
output_file	path	No	Write JSON report to this file (in addition to table output)

execution

Fixed mode — request_count and concurrency together:

Field	Type	Required	Description
request_count	int	Yes	Total requests to send (or scenario iterations, when scenarios is set). Max: 100_000_000
concurrency	int	Yes	Concurrent workers. Max: 10_000
rps	int	No	Aggregate requests-per-second cap across all VUs. Values are token-bucket-smoothed, so a rps: 100 run produces ~100 req/s rather than a once-per-second burst. Omit (or set null) for no rate limit.

Curve mode — stages only (cannot mix with fixed mode fields):

Field	Type	Required	Description
stages	array	Yes	List of load curve stages
rps	int	No	Aggregate requests-per-second cap, also valid in curve mode. Acts as a ceiling on top of the curve — if the curve would produce more than rps req/s, VUs wait.

Each stage:

Field	Type	Required	Description
duration	string	Yes	Stage duration. Format: 30s, 2m, 1m30s
target_vus	int	Yes	Target VU count at end of stage. Max: 10_000. Max stages per curve: 1000
ramp	string	No	Ramp type. Values: linear (default), step

thresholds

Array of threshold rules:

Field	Type	Required	Description
metric	string	Yes	Metric name. See Thresholds & CI Gating
operator	string	Yes	Comparison operator: lt, lte, gt, gte, eq
value	float	Yes	Threshold value. Must be finite. error_rate must be between 0.0 and 1.0

scenarios

An array of named scenarios, each with one or more steps. Scenarios are mutually exclusive with run.host, run.method, and the top-level request_template/response_template fields — use one or the other.

When scenarios are present, VUs are assigned to scenarios proportionally by weight. Each iteration of a VU runs all steps in its assigned scenario in order.

A config may define up to 64 scenarios.

scenarios:
  - name: checkout
    weight: 3
    on_step_failure: continue
    headers:
      X-Session: abc
    steps:
      - name: login
        host: https://api.example.com/auth/login
        method: post
        headers:
          Content-Type: application/json
        request_template: templates/login.json
        response_template: templates/login_resp.json
      - name: add_to_cart
        host: https://api.example.com/cart
        method: post
  - name: browse
    weight: 1
    steps:
      - name: list_products
        host: https://api.example.com/products

Scenario fields

Field	Type	Required	Description
name	string	Yes	Unique scenario name. Must not be empty
weight	int	No	Relative weight for VU assignment. Default: 1. Range: 1..=10_000
on_step_failure	string	No	What to do when a step fails. Values: continue (default), abort_iteration
headers	map	No	Headers applied to all steps in this scenario. Merged on top of run.headers. Subject to the same per-map caps as run.headers (≤ 64 entries, name ≤ 256 chars, value ≤ 8192 chars)
steps	array	Yes	Ordered list of steps. At least one step is required

Step fields

Field	Type	Required	Description
name	string	Yes	Unique step name within the scenario. Must not be empty
host	string	Yes	Full URL for this step including scheme. Supports ${ENV_VAR} substitution
method	string	No	HTTP method. Default: get. Values: get, post, put, patch, delete
headers	map	No	Step-level headers. Merged on top of scenario headers (last-wins, case-insensitive). Same per-map caps as above
request_template	path	No	Path to a JSON request body template for this step
response_template	path	No	Path to a JSON response extraction template for this step
body	string	No	Inline request body string. Mutually exclusive with request_template. Supports {{capture.KEY}} injection. Max length: 1 MiB (1,048,576 bytes)
capture	map	No	Capture definitions: alias → JSON path. Aliases must match [a-zA-Z0-9_]+. Paths must start with $.

Capture injection

Step headers values and body strings support {{capture.KEY}} references. The value is replaced at runtime with the string captured from a preceding step's response.

References are validated at config load time: if a step references {{capture.token}}, a preceding step must define capture.token. Runtime extraction failures (missing JSON path in response) cause the iteration to abort and remaining steps to be marked as skipped.

See Scenarios guide — Step chaining for examples.

Header merge order

Headers are merged in this priority order (later overrides earlier, case-insensitive):

1. run.headers — global headers applied to all scenarios and steps
2. Scenario headers — applied to all steps within the scenario
3. Step headers — applied to this step only

VU assignment

VUs are distributed across scenarios proportionally to their weight. With weights [3, 1] and 4 VUs: 3 VUs run checkout, 1 VU runs browse. With 8 VUs the pattern repeats: 6 run checkout, 2 run browse.

In fixed mode, the execution.request_count budget is shared across all VUs regardless of scenario, but one budget unit covers one scenario iteration — every step in the iteration is executed under that single claim. A 3-step scenario with request_count: 100 therefore produces 100 full iterations and ~300 HTTP requests (ignoring skipped or aborted steps).

See the Scenarios guide for usage patterns and the Scenarios internals page for the execution model, capture lifecycle, and validation pipeline.