Quack Workflow Engine

ReviewAudited by ClawScan on May 10, 2026.

Overview

The skill is coherent, but it can send workflow files to an external service and includes a workflow that posts generated web-search content to Twitter without clear approval or account-scope safeguards.

Review workflow files before running them, especially any workflow that posts, deletes, purchases, or changes external-account data. Treat Orchestrate as an external service that may receive your workflow contents and use linked accounts, and require a manual approval step before public posting.

Findings (5)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

What this means

A workflow file can cause remote actions to run, including high-impact actions if the remote platform supports them.

Why it was flagged

The runner submits the selected workflow file directly to the remote execution API without local validation, action allowlisting, or a confirmation gate before execution.

Skill content
const content = await readFile(workflowPath, 'utf8'); ... const res = await fetch(`${BASE_URL}/api/v1/workflows/run`, { method: 'POST', ... body: content });
Recommendation

Add explicit confirmation for mutating/public actions, validate allowed workflow actions, and make dry-run or preview the default for risky workflows.

ConcernHigh Confidence
ASI08: Cascading Failures
What this means

Bad or manipulated web content could be transformed into a public post under the user's connected account.

Why it was flagged

The template chains untrusted web-search output into an LLM summary and then into a social-media post, with no shown human review or containment step.

Skill content
action: web_search ... prompt: "Summarize these search results into a concise social media post: {{research.output}}" ... action: social_post ... text: "{{summarize.output}}"
Recommendation

Insert a mandatory review/approval step between generated content and posting, and treat web-search results as untrusted input.

ConcernMedium Confidence
ASI03: Identity and Privilege Abuse
What this means

Users may not know which account will be used or what permissions the remote workflow service has when posting.

Why it was flagged

Posting to Twitter implies delegated social-account authority, but the artifacts do not declare which credential, account, or permission scope is used.

Skill content
action: social_post ... platform: twitter
Recommendation

Document required Orchestrate and social-account credentials, account-linking behavior, permission scopes, and revocation steps.

What this means

Sensitive information placed in a workflow file may be shared with Orchestrate and any downstream providers it uses.

Why it was flagged

The selected workflow YAML is sent to an external orchestration provider. This is purpose-aligned, but prompts, inputs, or embedded data in the workflow leave the local environment.

Skill content
const BASE_URL = 'https://orchestrate.us.com'; ... body: content
Recommendation

Avoid putting secrets or private data in workflow YAML unless the provider's data handling is acceptable; document downstream data sharing.

What this means

Users have less registry-level context for verifying the provider and may hit an undeclared runtime requirement.

Why it was flagged

The skill has limited provenance metadata and under-declares the Node runtime required by the documented command. No hidden install code or dependency download is shown.

Skill content
Source: unknown; Homepage: none; Required binaries (all must exist): none
Recommendation

Declare the source/homepage and Node runtime requirement, and verify the Orchestrate provider before using workflows with sensitive or public effects.