agentleFS
timeyourtimeyour/agentskills-audit-collection/.claude/skills/harness/SKILL.md

harness

Skill1 starsChanged 6 months agoMITView on GitHub

Build an engineering delivery harness for AI-assisted work: decompose business goals into multi-level steps, route each step to prompt, AgentSkill, Dify workflow, RPA, code, manual human intervention, or external tool, and define automatic checkpoints, retry strategy, escalation paths, evidence gates, and acceptance criteria. Use before implementing, auditing, or operationalizing a complex workflow.

36 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use retro, verify and 3 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • AWSSkill

    retro

    Conduct a retrospective on a coding session.

    Also in 3 projects that use this

  • Microsoft AzureSkill

    verify

    End-to-end app testing pipeline for browser-based web apps on cloud browsers via Azure Playwright Workspaces: verify the live app, author tests for working flows, heal failing tests, then re-run with video and produce one consolidated report. Three phases (app verification, test authoring, test healing) plus a scoped final sanity run; each phase emits a sub-report. Use when asked to "verify my app", "make sure my app works and has good test coverage", "do a full QA pass", "add tests where I'm missing coverage", "fix failing tests and add new ones", "smoke test before deploy", "QA my deployment", or any combined verify-and-test ask. Requires Azure CLI (`az login`) and a Playwright Workspace resource ID — bootstraps configs on first invocation. Browser-based web apps with Playwright only — not for unit/API tests, native mobile/desktop, load/perf, or full accessibility audits.

    Also in 3 projects that use this

  • GoogleSkill

    code-review

    Adversarial systems, concurrency, memory-model, and architecture code review protocol. Use this skill whenever asked to review code, PRs, diffs, or branches in Filament.

    Also in 4 projects that use this

  • GoogleSkill

    grill-me

    A relentless interview to sharpen a plan or design.

    Also in 3 projects that use this

  • SentrySkill

    skill-creator

    Create new agent skills following the Agent Skills specification. Use when asked to "create a skill", "add a new skill", "write a skill", "make a skill", "build a skill", or scaffold a new skill with SKILL.md. Guides through requirements, writing, registration, and verification.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

11 rules. 0 are common, 6 are distinctive, and 5 are very specific to this project.

  • 0

    Common

    Most of the 24 edited versions we read keep these.

  • 6

    Distinctive

    This author's own choices. Worth a look.

  • 5

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use this skill to turn a messy business objective into an executable delivery harness.

    line 13
  2. Distinctive:

    Start from the business process, not from the tool.

    line 23
  3. Distinctive:

    Keep source evidence, live evidence, tool evidence, and human decisions separate.

    line 29
  4. Very specific:

    If the goal is to convert a manual business process into a process agent, also use references/process-agent-pattern.md.

    line 44
  5. Distinctive:

    Add retry and escalation

    line 63
  6. Distinctive:

    Never retry destructive actions blindly.

    line 65
  7. Very specific:

    Run /audit for overall readiness.

    line 73
  8. Very specific:

    Run /flow-test for user-visible workflows.

    line 74
  9. Very specific:

    Run /deploy-check for production dependencies.

    line 75
  10. Very specific:

    Run /accept-five for important workflows.

    line 76
  11. Distinctive:

    No checkpoint between AI output and real-world action.

    line 129

“Common” is measured: kept by at least 6 in 10 of the 24 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Harness

Use this skill to turn a messy business objective into an executable delivery harness.

The goal is to prevent AI work from becoming one giant prompt. Break the work into business stages, substeps, execution units, checkpoints, retries, and human decision points.

business objective -> stage tree -> execution router -> checkpoints -> retries/escalations -> acceptance harness

Core Rules

  • Start from the business process, not from the tool.
  • Decompose in layers until each unit has one clear owner, input, output, tool mode, checkpoint, and retry rule.
  • Route each unit to the right execution mode: prompt, AgentSkill, Dify workflow, RPA, code task, manual human decision, or external system.
  • Insert automatic checkpoints after every unit that can fail silently.
  • Insert human checkpoints where judgment, permissions, money, private data, or irreversible actions appear.
  • Define retry limits and escalation paths before execution.
  • Keep source evidence, live evidence, tool evidence, and human decisions separate.
  • Feed final outputs into /audit, /flow-test, /deploy-check, or /accept-five for verification.

Workflow

  1. Scope the objective
    • Name the business goal.
    • Identify user/customer, operator, system, and decision owner.
    • Define done, blocked, and out-of-scope.
  2. Build the stage tree
    • Split into business stages.
    • Split each stage into substeps.
    • Continue until each unit is small enough to execute and verify independently.
    • Use references/business-decomposition.md.
    • If the goal is to convert a manual business process into a process agent, also use references/process-agent-pattern.md.
  3. Route execution modes
    • Choose the lightest reliable mode for each unit:
      • PROMPT: one-off reasoning or drafting.
      • SKILL: repeatable agent workflow.
      • DIFY: structured multi-step LLM workflow, API chain, or business automation.
      • RPA: brittle UI-only workflow with no API or direct integration.
      • CODE: deterministic transform, parser, test, or integration.
      • HUMAN: judgment, approval, credential entry, payment, legal, brand, or safety decision.
      • EXTERNAL: SaaS/API/platform action outside the agent.
    • Use references/execution-router.md.
  4. Add checkpoints
    • Define automatic checkpoint after each unit.
    • Define human checkpoint before irreversible, private, paid, or high-judgment actions.
    • Each checkpoint needs expected evidence, pass/fail condition, and owner.
    • For process agents, add a signal ledger and iteration loop: collect signals, diagnose root cause, test change, verify reduction.
  5. Add retry and escalation
    • Define retryable failures, retry count, backoff, fallback mode, and escalation owner.
    • Never retry destructive actions blindly.
    • Use references/checkpoint-retry-policy.md.
  6. Output the harness
    • Produce stage tree, execution matrix, checkpoint table, retry table, human-intervention map, risk register, and acceptance handoff.
    • For process-agent work, also produce lifecycle and signal/iteration tables.
  7. Verify the harness
    • Run /audit for overall readiness.
    • Run /flow-test for user-visible workflows.
    • Run /deploy-check for production dependencies.
    • Run /accept-five for important workflows.

Output Format

## Harness Summary

- Business goal:
- Scope:
- Done:
- Out of scope:
- Main risk:

## Stage Tree

| Level | Stage | Step | Input | Output | Owner | Dependency |
| --- | --- | --- | --- | --- | --- | --- |

## Execution Matrix

| Step | Mode | Why this mode | Tool/Skill/Workflow | Evidence | Fallback |
| --- | --- | --- | --- | --- | --- |

## Checkpoints

| Step | Checkpoint | Automatic/Human | Pass condition | Fail condition | Evidence | Owner |
| --- | --- | --- | --- | --- | --- | --- |

## Retry And Escalation

| Failure | Retryable | Retry limit | Backoff/Fallback | Human escalation | Stop condition |
| --- | --- | --- | --- | --- | --- |

## Acceptance Handoff

| Target | Next skill | What to verify |
| --- | --- | --- |

## Process Agent Lifecycle

| Stage | Business Question | Output | Checkpoint | Owner |
| --- | --- | --- | --- | --- |

## Signal And Iteration Loop

| Signal | Root Cause | Change Tried | Verification | Result |
| --- | --- | --- | --- | --- |

Anti-Patterns

  • One giant prompt for a multi-stage business process.
  • Choosing Dify, RPA, or skills before mapping the business steps.
  • No checkpoint between AI output and real-world action.
  • Retrying the same bad prompt without changing input, context, or mode.
  • Building a process agent before defining the business flow and data templates.
  • Optimizing symptoms instead of diagnosing root causes from repeated signals.
  • Treating RPA as robust when an API or deterministic code path exists.
  • Letting the agent approve its own high-risk action.
  • Hiding manual steps instead of marking them as human checkpoints.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.