Experiment
Generated from workflows/modes/experiment.md, the mode contract the HELIX skill loads. Edit that file, not this page.
Use for metric-driven optimization loops.
- Define the goal, metric, baseline, intervention, and decision that the measurement will change. State the stopping condition before running.
- Bound the whole experiment, including retries and parallel runs: maximum runs, elapsed time, retained temporary storage and paid-service usage where applicable. Honor existing user and tool limits. With no baseline, choose a small probe with explicit size and duration limits before expanding.
- Before each launch, bound its workload, timeout and outstanding parallel jobs to the remaining cumulative allowance. Measure results and resource use after each iteration, before launching another. Expand only to resolve a named remaining uncertainty. Stop when the decision is supported or any limit is reached; preserve results and stop outstanding work using available controls or operator intervention. Report work still running rather than imply successful cancellation. Resuming does not reset the allowance.
- Keep changes or revert/adjust based on metric evidence.
- To validate a methodology or skill change (a workflow prompt, template, or this routing skill), use a regression evaluation: record a committed baseline, run a fixed brief from the bare prompt with the improved skill installed (never by redirecting reads), score intrinsic metrics against the baseline, and keep what moved, cut what didn’t. That evaluation is the standing answer to “how do we know this change is impactful.”
Known prices and metered usage can support a monetary estimate; unknown rates
or delayed billing cannot support a claimed dollar cap. Use available workload,
time and storage bounds instead, disclose the uncertainty once, and do not
start unbounded paid runs. Use automated enforcement and metering when available;
otherwise monitor the bounded work and arrange manual stopping with the operator.
Disclose any limits that cannot be enforced or usage that cannot be measured;
never claim a hard cap or successful stop without evidence. HELIX defines the
stopping decision independently of the execution tool. Record limits and
measured use in the existing experiment record, not a second budget document. Token
spending still follows workflows/references/work-item-first.md.
Procedure: workflows/actions/experiment.md (deeper step detail; this file is the contract).
Innsigle seal: model-primary by HELIX
The signature covers the markdown source of this page, not these HTML bytes. This page quotes that seal; verify it against the source file.
- Composition
- model-primary
- Issuer
- HELIX
helix - Signing key
ed25519:b0865d76d834a52c48506414d16f4e5a(build key)- Signed source
reference/workflow-modes/experiment.md- Signed
- 2026-11-16T15:52:45Z
- Content digest
sha256:88e8113d…5c04956c
This build key is endorsed by the human key for build signing; the signature is not a detector and not a truth guarantee.
Raw attestation JSON
{
"payload": {
"innsigle": "1",
"type": "https://innsigle.dev/claim/colophon/v1",
"issued_at": "2026-11-16T15:52:45Z",
"issuer": {
"id": "helix",
"name": "HELIX",
"key_id": "ed25519:b0865d76d834a52c48506414d16f4e5a",
"key_url": "https://documentdrivendx.github.io/helix/.well-known/innsigle/keys.json"
},
"subjects": [
{
"uri": "https://documentdrivendx.github.io/helix/reference/workflow-modes/experiment/",
"digest": {
"alg": "sha256",
"value": "88e8113d019c2312a33e809ea73c676759ec105e01826ac3ec124f685c04956c"
}
}
],
"colophon": {
"schema_version": "1",
"composition": "model-primary",
"ingredients": [
{
"kind": "model",
"name": "Claude",
"role": "draft"
},
{
"kind": "tool",
"name": "sloptimizer",
"role": "rewrite"
},
{
"kind": "human",
"name": "operator",
"role": "structure-edit"
}
],
"notes": null
}
},
"payload_encoding": "json",
"signatures": [
{
"key_id": "ed25519:b0865d76d834a52c48506414d16f4e5a",
"alg": "ed25519",
"sig": "Nk5YLtxkvaKleXJxC390VZ1BfNwVpP4A0nHuzRdiEXjY8gIhCnWdsLFfuDXdCZ17UUnYBl8K_ddkWAgiOHLyAA",
"signed_at": "2026-11-16T15:52:45Z"
}
]
}