ha-coding-common

Guides code implementation, fixing, and refactoring with evidence-first repository inspection and scoped edits.

1.6k|149|Updated Mar 13, 2026
One-click install
npx skills add https://github.com/shiwenwen/hope-agent --skill ha-coding-common
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ha-coding-common
Source: https://github.com/shiwenwen/hope-agent/tree/main/skills/ha-coding-common
Command: npx skills add https://github.com/shiwenwen/hope-agent --skill ha-coding-common

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Coding tasks often go wrong when changes are made without understanding the existing codebase, when unrelated user work gets overwritten, or when completion is claimed without real verification. This Skill establishes a disciplined baseline for implementing, fixing, refactoring, and maintaining code so changes stay scoped, safe, and verifiable.

Core Features & Use Cases

  • Evidence-First Inspection: Requires reading the worktree, relevant files, nearby tests, and docs (using rg where available) before choosing an implementation, and protects uncommitted user changes from being reverted or reformatted.
  • Right-Sized Workflow: Distinguishes small, clear tasks (act directly) from multi-step or uncertain work, and routes specialized needs to narrower skills such as debugging, test strategy, code review, multi-agent execution, verification, and workflow scripting.
  • Scoped Change Discipline: Keeps edits limited to the requested behavior and owning subsystem, avoids speculative abstractions and unrelated cleanup, and finishes with diff inspection plus requirement-to-evidence verification.
  • Use Case: When asked to implement a feature in a Rust or TypeScript repository, the Skill ensures the agent first inspects existing helpers and conventions, makes a minimal diff, and reports exactly what was verified and what residual risk remains.

Quick Start

Ask the agent to implement a small feature or fix a bug in your repository while keeping the diff minimal and verifying the result against the original requirements.

Frequently Asked Questions about ha-coding-common

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

FAQPage Schema
How do I make an AI agent implement a feature without breaking existing code?

Require the agent to inspect the worktree, relevant files, and nearby tests before editing, and to prefer existing helpers and conventions. This Skill enforces that evidence-first approach and keeps edits scoped to the requested behavior and owning subsystem.

How to protect uncommitted changes when an AI edits my repository?

Treat all uncommitted changes as belonging to the user and never revert, overwrite, or reformat unrelated work. This Skill makes that protection an explicit rule, so only the files tied to the requested task are modified.

When should a coding task use a plan instead of acting directly?

Act directly when the behavior, owning file, and verification path are obvious. Use a formal plan only when the change spans ownership boundaries, has ordering constraints, carries migration risk, or needs explicit completion criteria.

Does this coding workflow support languages like Rust, TypeScript, and Python?

Yes, it applies to Rust, TypeScript, JavaScript, Python, Go, Java, Kotlin, Swift, C, C++, Ruby, PHP, and shell files. The methodology is language-agnostic and focuses on inspection, scoped edits, and verification rather than any specific toolchain.

Why does an agent claim a task is done when nothing was verified?

Completion claims fail when they rely on intent or unrelated passing commands. This Skill requires inspecting the final diff, mapping each requirement to concrete verification evidence, and reporting any real residual risk before declaring completion.