code

Implements WBS plans and investigation reports by editing files, running tests, and logging decisions.

Updated Mar 24, 2026
One-click install
npx skills add https://github.com/hanh-nd/agent-kit --skill code-hanh-nd
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: code
Source: https://github.com/hanh-nd/agent-kit/tree/main/plugins/agent-kit/.gemini/skills/code
Command: npx skills add https://github.com/hanh-nd/agent-kit --skill code-hanh-nd

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Turning a written plan or root-cause investigation into actual code often drifts: invented symbols, out-of-scope edits, style mismatches, and untracked decisions. This Skill executes a validated contract end-to-end while enforcing scope discipline, reuse of existing code, and auditable decision records. ## Core Features & Use Cases - Contract-driven implementation: Executes a WBS plan or Investigation Report exactly as written, halting on logic gaps, contract conflicts, or unauthorized dependencies instead of guessing. - Reuse-first discipline: Runs a mandatory reuse sweep against the repo's shared utilities and ARCHITECTURE.md Reuse Map before writing any new helper, preventing duplicate code. - Auditable execution records: Tracks material decisions in DECISIONS.md and produces a structured Code Execution Report covering progress, files modified, tests, and acceptance criteria. - Use Case: Hand the Skill a WBS plan for a new feature; it reads the affected files, implements each task in dependency order, runs the project's lint and test scripts, and saves a handoff report with any gaps surfaced. ## Quick Start Use the code skill to implement the WBS plan in the handoff folder for the user-settings feature and report the results.

Frequently Asked Questions about code

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

FAQPage Schema
How do I implement a WBS plan with an AI coding agent?▼

Provide the WBS plan as the input contract. The agent reads every file the plan names, executes tasks in layer and dependency order, runs the project's own lint and test scripts, and saves a Code Execution Report with per-task status and acceptance criteria verification.

How to fix a bug from a root-cause investigation report?▼

Pass the Investigation Report as input. The agent applies the smallest change targeting the documented root cause in the affected files, adds a regression test where a surface exists, and halts if the report status is INCONCLUSIVE or the evidence no longer matches the code.

What happens when the plan references code that does not exist?▼

The agent halts with a Logic Gap notice quoting the claimed reference versus reality, continues any unblocked tasks, and surfaces the gap in the final report. It never invents missing symbols, paths, or files.

Does the code skill write tests automatically?▼

Only when the contract includes a TESTS.md file or explicit test tasks. Otherwise it skips test writing and records the omission, since the planner owns the decision of whether tests are required.

When should I not use contract-based code execution?▼

Avoid it when no validated plan or investigation exists, since the agent stops and asks for a contract. Exploratory refactors, open-ended design work, or cleanup without a defined scope belong to planning or refactoring skills instead.