Feature · Coding Automation

Coding automation inside a visual runtime, when the canvas is not enough.

The query coding automation covers everything from CI/CD orchestration to code generation. Strkr is neither. Strkr Flows is a visual automation runtime for revenue operations with a code step that lets engineers extend the canvas for edge cases without rebuilding the whole workflow in a repository.

Start free trial See pricing
What this page covers

Narrow pivot: code steps inside Strkr Flows.

If you reached this page looking for a replacement for GitHub Actions, CircleCI, GitLab pipelines, or any developer CI tool, Strkr is the wrong product. If you reached this page looking for a way to extend a no-code business automation runtime with a short piece of code when the visual canvas does not fit the problem, every section below is written for you.

Out of scope

We are not a CI tool.

Strkr does not build, test, or deploy code. We do not run on push events to a Git branch, we do not spin up build matrices, and we do not report back to a pull request. GitHub Actions, CircleCI, GitLab CI, Jenkins, and Buildkite are the names your engineering team already knows.

In scope

Business automation with a code escape hatch.

Strkr Flows is a visual canvas for revenue automation (lead routing, deal updates, marketing cadences, SMS drip, project handoffs). The code step is for the ten percent of workflows that need a function to transform data, call a specialized service, or implement a non-trivial business rule.

The split

Visual by default, code on exception.

Default to the canvas. Use a code step when the visual logic becomes awkward (nested conditions, iteration over a list, non-standard enrichment, a legacy system without a native connector). The code step is a node on the same canvas, not a parallel system.

Primary reader

A RevOps lead with occasional engineering support.

The RevOps lead owns the Flow. Engineering reviews the code step when one gets added. The sales floor sees a working automation on day one, and the engineering team keeps a short review loop instead of maintaining a parallel scripts repository.

Why we built it

The ten-percent problem is real.

Every no-code platform hits a wall at some workflow shape. The classic answer is a parallel scripts folder that nobody maintains. The Strkr answer is a code step that keeps the exceptional logic on the same canvas as the normal logic, with the same audit and governance posture.

What follows

Mechanics, patterns, governance, limits.

The rest of this page covers how the code step works inside Flows, when to reach for it, how review and governance operate, and what alternatives exist for teams whose actual need is dev tooling rather than revenue automation.

How the code step works

A function on the canvas, not a repository elsewhere.

A code step is one node on the Flows canvas. It accepts the Flow payload as input and returns a structured output the next step consumes. It runs sandboxed with explicit credential grants and resource ceilings. The function body lives in Strkr; there is no external repository to maintain.

Language choice

JavaScript or Python at the step level.

A code step runs either a short JavaScript function or a short Python function. The runtime is pinned to a minor version per tenant so a dependency upgrade in Node or Python cannot silently change behavior under a production Flow.

Input shape

The Flow payload, typed.

The step receives the Flow payload: triggering record, upstream node outputs, actor context, and any variables set earlier in the chain. The schema is visible in the step configuration so the engineer writes against a known shape rather than guessing field names.

Output shape

Structured return the next step reads.

The function returns a JSON object with keys the next step expects. Schema violations get caught at the step boundary, not downstream in a mystery branch. Engineers define the output schema inline with the function so a review captures both the logic and the contract.

Resource ceilings

CPU, memory, and wall-clock caps.

Each code step has a maximum wall-clock duration, a memory ceiling, and a CPU-time ceiling. The caps prevent a runaway function from affecting Flow throughput for the rest of the tenant. If a function hits the cap, the step errors into the dead-letter queue with the context attached.

Credential grants

Explicit secrets, admin approved.

A code step can call external services only via credentials explicitly granted to the step. Admins approve each grant. The step configuration shows which credentials are in scope. The exfiltration pattern (a function that posts tenant data to an attacker-controlled URL) is blocked by the credential model.

Logs and audit

Every run captures the function activity.

Each code step run writes the input (with sensitive fields redacted per tenant config), the output, the execution timing, and any thrown exceptions to the audit log. The log is queryable and exportable. Compliance teams review code-step activity on the same surface as the rest of the Flow.

When to reach for a code step

The patterns that justify the escape hatch.

Not every awkward Flow needs a code step. Reaching for one too often turns Strkr Flows into a scripts repository with a visual wrapper. Below are the patterns we see land cleanly with a code step, and the patterns to leave on the visual canvas.

Bespoke parser

Legacy data with a weird shape.

A partner sends CSVs with inconsistent column names and some embedded JSON in one of the cells. The visual canvas has no clean way to parse the shape. A code step accepts the raw CSV, returns the normalized record. The Flow continues with a predictable payload.

Complex pricing

Multi-tier math with non-linear rules.

A deal needs a quote with volume tiers, seasonal discounts, and partner markup. The pricing is a page of logic. A code step implements the math; the Flow uses the output to populate the quote line items and send to the prospect for signature.

Non-native integration

An API with no connector.

The team uses a niche vertical tool that Strkr does not integrate natively. A code step calls the vendor API using an admin-granted credential and returns the result for downstream Flow actions. The code step is the smallest useful bridge; building a native connector is overkill for a single-tenant need.

Enrichment choice

Dynamic enrichment vendor selection.

Routing enrichment vendors based on target geography (vendor A for North America, vendor B for Europe, vendor C for Asia) is a short function instead of a tangle of conditional branches. The code step returns the enriched record from whichever vendor matched.

Leave on canvas

Routing, scoring, drafting, scheduling.

The core patterns stay on the canvas. Lead routing uses the Flow condition builder. Scoring uses the Strkr AI step. Drafting uses the Strkr AI step. Scheduling uses the Flow scheduler. Reaching for a code step for any of these means the canvas is not being used right.

Hard stop

Not for product features.

If the logic is something ten customers would want, file it with engineering as a product request, not as a code step. The code step is a safety valve for one-off tenant-specific logic, not a way to backfill missing product capability.

Review and governance

Engineering sees what RevOps shipped.

A code step is reviewed before it goes live. The review captures the function body, the test fixture, and the expected output shape. The approval sits on the Flow detail page alongside the approver name and timestamp. The pattern mirrors the pull-request flow engineering already uses.

Review required

Admin configurable per tenant.

Tenants can require review on every code step, or on code steps in specific Flow categories, or on code steps that call outbound credentials. The requirement is enforced before publish; a draft that includes an unreviewed code step cannot go live.

Diff view

Changes show line by line.

When a reviewer opens a code step for review, the diff view shows the current production function alongside the proposed change. Added lines are green, removed lines are red. The reviewer can approve, request changes, or reject with a comment that lands on the Flow detail page.

Test fixtures

Capture the expected behavior inline.

A code step configuration includes test fixtures: sample inputs with their expected outputs. The reviewer sees the fixtures. On publish the fixtures run to confirm the function still produces the expected shape. Fixtures that fail block publish; the Flow stays on the prior version.

Audit coverage

Same log, same export.

Code step runs log the same way visual step runs log: trigger payload, step input, step output, timing, errors. The audit log is immutable and exportable. Compliance teams see code-step activity on the same surface as the rest of the Flow activity, with no separate telemetry tool.

Rollback

Prior function version is one click away.

Code steps are versioned. If a production run misbehaves, one click reverts the step to the prior version and the next Flow runs pick up the rollback. The rollback is captured on the Flow detail page with the actor and timestamp.

Permission inheritance

Code steps obey Flow permissions.

A code step runs as the Flow owner by default and respects the owner permissions. Admins can elevate a Flow (and the included code step) to run as a service account with broader access, logged separately with the elevation visible on the detail page. The owner cannot grant the elevation; only admins can.

Alternatives in dev tooling

Honest redirects when Strkr is not the right vendor.

If what you actually need is developer automation, below is the honest routing. We would rather send you to the right tool than hold you through an evaluation that will end in a no.

CI/CD orchestration

GitHub Actions, CircleCI, GitLab, Jenkins.

For build, test, deploy on code events, the mainstream list is GitHub Actions, CircleCI, GitLab CI, Jenkins, and Buildkite. For infrastructure as code, Terraform or Pulumi. Strkr belongs on exactly none of those evaluation grids.

Code generation

Developer-facing AI tools.

For code completion, generation, or refactoring, the developer tooling catalog covers the space (your engineering lead knows the names). Strkr AI generates sales drafts, meeting summaries, and classification outputs, not code.

Scheduled scripts

Cron plus a managed runtime.

If the actual need is a scheduled script with no business context, a managed cron (Cloud Scheduler, EventBridge, GitHub Actions cron) is often the right answer. Strkr Flows is overkill for a daily cron job that touches nothing in your CRM.

External glue layer

Third-party platforms connect code to apps.

Third-party external platforms let you add a code step between SaaS actions. The approach works at low volume but gets expensive at scale because of per-task fees. We do not recommend them as the strategic answer for a growing revenue org.

Strkr Flows

If the automation touches revenue data, Flows fits.

If the workflow interacts with the CRM, marketing, messaging, or project modules, Strkr Flows is the right surface. The visual canvas covers the majority, the code step covers the exceptions, and the audit log covers the whole stack for compliance.

the answer

Pick the tool that matches the problem shape.

Dev tooling for dev workflows. Business automation for business workflows. The categories overlap only at the margins. Strkr is not trying to be your CI tool, your package manager, or your IaC runner. We are trying to be the automation layer under your revenue stack.

Need a canvas for business automation with a code escape hatch? Flows is the surface.

Build a routing Flow this afternoon. Add a code step when the business rule needs one. Review loops stay short, compliance stays covered, and the whole chain lives on one canvas.

Start free trial Explore Flows
Common questions

What buyers ask about this feature.

Is Strkr a replacement for GitHub Actions or Jenkins?

No. Strkr is a revenue operations platform; Flows automates CRM, marketing, messaging, and project work. We do not build code, run test suites, or deploy artifacts. For CI/CD orchestration the mainstream names are GitHub Actions, CircleCI, GitLab CI, Jenkins, and Buildkite. Strkr belongs on exactly zero of those shortlists. If what you actually need is a canvas that automates business work with a short code step for the hard cases, keep reading.

When does it make sense to add a code step to a Flow?

Reach for a code step when the visual canvas becomes awkward: a bespoke CSV parser, a multi-tier pricing calculation, a call to a vendor API without a native connector, or dynamic enrichment vendor selection. Leave routing, scoring, drafting, scheduling, and notification on the canvas; those have first-class support that outperforms rolling your own. If the logic is something ten customers would want, file it with engineering as a product request instead of building it into a code step.

What languages can a code step run?

A code step runs either a short JavaScript function or a short Python function. The runtime is pinned to a minor version per tenant to prevent silent behavior changes under a production Flow. The step runs sandboxed with CPU, memory, and wall-clock ceilings. Explicit credential grants cover any outbound API calls; a step cannot read tenant secrets it was not granted.

How does code step review work?

Code steps go through pull-request-style review before publish. The reviewer sees the function body, the test fixtures (sample inputs with expected outputs), and the expected output shape. On publish, the fixtures run to confirm the function still produces the expected result; a failing fixture blocks the publish and the Flow stays on the prior version. The review approval is captured on the Flow detail page with the approver name and timestamp.

What does the audit log capture for code step activity?

Each code step run writes the step input (with sensitive fields redacted per tenant config), the step output, the execution timing, and any thrown exceptions. The log is immutable and queryable, exportable to CSV or Parquet. Compliance teams review code-step activity on the same surface as the rest of the Flow activity. SOC 2 Type II attestation is in progress through 2026-Q4; the audit log export covers both the SOC 2 and ISO review patterns.

Can I use Strkr Flows for a nightly data sync job?

Yes, if the data lives in a Strkr module or touches a system a Flow can reach. A scheduled Flow fires at a configured time, runs the chain (fetch, transform, write), and logs the outcome. For a nightly job that touches nothing in your CRM, marketing, messaging, or project surfaces, a managed cron tool is often the simpler answer. We would rather send a nightly backup job to Cloud Scheduler than hold the use case artificially inside Flows.

Related features

Pair this with...

AI Workflow Automation Codeless Automation (Scope Note) CRM Automation Content Workflow Software

Try it free. Bring your team next week.

No sales call, no migration consultant, no four-month implementation. Enter your card, get 14 days of the full Pro tier, cancel any time before day 14 with zero charge. Spin up a workspace, import your CSV, and have something useful before lunch.

Start free trial See pricing