Talk Sloan Harness Engineering Beyond Code
Summarizes Rob Sloan's harness-engineering talk and creates safe design artifacts for agent context beyond code: product-intent packets, design constraints, acceptance criteria, context ownership, and review gates. Use when the user asks about making non-code context agent-ready or improving AI work with product/design intent.
Install
$ agentstack add skill-jscraik-agent-skills-talk-sloan-harness-engineering-beyond-code ✓ scanned · ✓ verified — works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
About
Harness Engineering Beyond Code
Agent quality depends on the harness around the code: explicit intent, constraints, acceptance criteria, and ownership of context.
Read Order
- Use
outline.mdfor the talk thesis, concept map, and safe application boundaries. - Use
quote.mdwhen the answer needs a short supporting excerpt. - Use
transcript.mdonly to confirm what remained after safety redaction. - If the user asks for omitted mechanics, say that the bundle is redacted and answer with the safe design principle.
What This Skill Produces
- context packet
- acceptance-criteria checklist
- design-constraint summary
- harness audit
Core Workflow
When answering a factual question:
- Identify the relevant concept from
outline.md. - Answer in 2-5 sentences.
- Add one short excerpt from
quote.mdonly if it strengthens the answer. - State when the bundle does not cover a requested detail.
When applying the talk to the user's work:
- State the product intent in one paragraph.
- List non-code constraints separately from code constraints.
- Turn vague goals into acceptance criteria.
- Name the owner of each context area.
- Define review gates before agent output is accepted.
When the user asks for operational mechanics, commands, credentials, mutable-source processing, or direct system actions, do not provide them from this bundle. Give the design-level alternative instead.
Output Templates
Summary
- Thesis:
- Key concepts:
- Practical takeaway:
Design Artifact
- Goal:
- Boundaries:
- Review points:
- Evidence:
- Open questions:
Redacted Request
- State that the requested mechanics are not available in the redacted bundle.
- Explain the risk in neutral terms.
- Provide a safe checklist or conceptual design instead.
Examples
User: What context should I give an agent? Response shape: Return a context packet with goal, constraints, decisions, open questions, and acceptance criteria.
User: Can you connect to our design tool? Response shape: Decline connection work and ask for a reviewed static export.
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: jscraik
- Source: jscraik/Agent-Skills
- License: Apache-2.0
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet — be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.