harness-explore

Conducts lightweight feasibility research and produces a cited findings document without writing code.

Updated May 15, 2026
One-click install
npx skills add https://github.com/Alan-IFT/harness-kit --skill harness-explore-alan-ift
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: harness-explore
Source: https://github.com/Alan-IFT/harness-kit/tree/main/skills/harness-explore
Command: npx skills add https://github.com/Alan-IFT/harness-kit --skill harness-explore-alan-ift

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? When you need to know whether something is possible before committing to a build — "can library X handle our use case?", "is this API rate limit acceptable?" — a full design-and-development pipeline is overkill. This Skill runs a light-weight exploration that answers the question with evidence, skipping design, gate review, and code entirely. ## Core Features & Use Cases - Light Requirement Analysis: Dispatches only the Requirement Analyst to frame the question, success criteria, and candidates — no acceptance criteria or user stories. - Evidence-Gathering Exploration: Reads code, runs probe commands (pip show, npm info, curl), and fetches upstream docs to investigate the question directly. - Cited Findings Document: Produces findings.md with a direct answer, evidence with file:line or link citations, project implications, and a recommended next step. - Use Case: Before adopting a new provider API, run an exploration to verify its rate limits against your expected volume, then receive a findings document recommending whether to proceed via a design pipeline or abandon the approach. ## Quick Start Ask the AI to run /harness-explore to investigate whether a specific library or approach can handle your use case and produce a findings document with evidence.

Frequently Asked Questions about harness-explore

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I check if a library or API is feasible before building a feature?▼

Run a feasibility exploration that frames the question, probes the library or API with commands like pip show, npm info, or curl, and records evidence in a findings document. You get a direct answer with citations and a recommended next step instead of a full design.

What is the difference between research mode and a full development pipeline?▼

Research mode dispatches only a light requirement analysis and produces a findings.md with evidence — no architect, gate review, developer, or QA stages. It costs roughly 15-20% of the full pipeline and answers "can we do X" rather than building X.

When should I not use a feasibility exploration task?▼

Skip it when you already know the answer and can go straight to planning or implementation, when the investigation would take days and should be split into smaller tasks, or when the issue is a clear bug that belongs in a regular task.

Can I explore multiple unrelated questions in one research task?▼

No. Each exploration task should answer exactly one question. Bundling unrelated questions dilutes the findings document and makes the evidence and recommended next step harder to act on.

What should a findings document contain after a feasibility study?▼

A findings document contains a one-paragraph direct answer, an evidence section citing links or file:line sources, implications for the project, and a recommended next step such as proceeding to design, abandoning the approach, or opening a deeper follow-up exploration.