agentsclimarketplace
Sign in

Review workflow determinism

Skill diagrid-labs/dapr-skills/skills/review-workflow-determinism

This skill reviews Dapr Workflow code for non-determinism hazards that would cause stuck workflows on replay. Use this skill when the user asks to "review workflow for determinism", "check non-determinism in workflow", "audit Dapr workflow replay safety", or similar.From its SKILL.md

Install
npx -y skills add diagrid-labs/dapr-skills --skill review-workflow-determinism

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 11 stars11 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

3.3 KB, 720 tokens by cl100k_base, as published. Nobody here has run it

Review Dapr Workflow — Determinism

Overview

Scans the bodies of Dapr Workflow classes/functions for code that is not safe to run during replay. Read-only: this skill never modifies source files. Activity bodies, management endpoints, and configuration are out of scope and are covered by review-workflow-activity and review-workflow-management.

Execution Order

You MUST follow these phases in strict order. Do not load files outside the agreed scope, and do not write or edit any files.

  1. Resolve scope — Read ../shared/review-scope-prompt.md and follow it to set scope_root.
  2. Detect target — Read ../shared/review-detect-target.md and follow it to produce language and workflow_files. If workflow_files is empty, emit a single critical finding (no workflow files found) and stop.
  3. Load checklist — Based on language, read exactly one of:
  4. Scan — For each workflow file in scope, apply every rule from the loaded checklist using Grep and Read. For each match, capture file:line, the rule id, and a short evidence snippet. Confirm the match is inside workflow scope (see "Cross-reference" in each checklist) before promoting it from "Please verify" to a graded severity.
  5. Report — Format every finding using the canonical template from ../shared/review-report-format.md. Group by severity, then rule id, then file path.
  6. Show final message — Your last output is the report. Do not append a summary, follow-up question, or next-action prompt other than the ## Next steps block defined by the report format.

Prerequisites

  • Read access to the project directory.
  • No build, compile, or run step is required — this skill is fully static.

Allowed tools

Read, Grep, Glob only. The skill MUST NOT call Bash, Write, or Edit.

Rules

The full rule list, including detection patterns, severities, and suggested fixes, lives in the loaded language checklist:

Rule ids are stable across releases. New rules append; deprecated rule ids are reserved.

Show final message

The last thing you emit MUST be the report from step 5, including the ## Next steps section. The ## Next steps block should suggest:

  • Run review-workflow-activity next if any activity files exist.
  • Run review-workflow-management next if any management files exist.
  • Re-run this skill after fixes if any critical findings were reported.

Do not add any text after the report.

What ships with it: 1 file

2.4 KB alongside SKILL.md

Gives 0 of the 12 instructions most review quality skills give in 720 tokens

Counted across 1,273 of the 2,403 authors here whose files we hold, read 2026-09-06

  • Ask one question at a timein 63 of 1273, across 62 files
  • Provide a recommended answer for each questionin 47 of 1273, across 45 files
  • Rank findings by severityin 44 of 1273
  • Use parameterized queries for database accessin 38 of 1273, across 20 files
  • Validate all user input with schemasin 33 of 1273, across 15 files
  • Store secrets in environment variablesin 32 of 1273, across 14 files
  • Explore the codebase to answer questionsin 31 of 1273, across 29 files
  • Store tokens in httpOnly cookiesin 30 of 1273, across 12 files
  • Implement rate limiting on API endpointsin 30 of 1273, across 12 files
  • Sanitize user-provided HTMLin 29 of 1273, across 11 files
  • Return generic error messages to usersin 28 of 1273, across 10 files
  • Cite file and line for every findingin 28 of 1273, across 25 files

Said here and by no other author read

  • Follow phases in strict order
  • Scan activity files using grep and read
  • Format findings using canonical template
  • Output report as final message
  • Read scope and detect target files
  • Confirm matches are inside workflow scope

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 325,949. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.