Alan_IFT avatar

Alan_IFT

Community

@Alan-IFT

1Followers
|
9Public Repos
|
17Published Skills

Keep thinking & Coding for fun.

Skills Distribution
DomainDeveloper To...Pipeline Orchestra.. (40%)Requirements Analy.. (25%)Code Quality Verif.. (20%)Project Bootstrapp.. (15%)

Agent Skills by Alan_IFT

Showing 17 vetted skills indexed across 1 GitHub repositories.

Alan-IFTAlan-IFT

harness-grill

Interviews users one question at a time to produce an aligned requirement brief.

Community
Intermediate
Alan-IFTAlan-IFT

harness-goal

Runs Developer and QA agents in a bounded loop until a measurable success criterion is met.

Community
Advanced
Alan-IFTAlan-IFT

harness-deflate

Scans the whole codebase for structural entropy and dispatches authorized refactorings to a goal-driven loop.

Community
Advanced
Alan-IFTAlan-IFT

harness-adopt

Adopts Harness Engineering workflow into existing projects through non-destructive scanning and planning.

Community
Advanced
Alan-IFTAlan-IFT

harness-intervene

Writes intervention signals to redirect, pause, or extend a running Harness pipeline.

Community
Intermediate
Alan-IFTAlan-IFT

harness-supervise

Analyzes 7-stage task folders and generates supervision reports flagging pipeline anti-patterns.

Community
Advanced
Alan-IFTAlan-IFT

harness-decision-mode

Switch a harness project's decision and escalation mode between human, preset-rubric, and custom-rubric autonomy.

Community
Intermediate
Alan-IFTAlan-IFT

harness-explore

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

Community
Intermediate
Alan-IFTAlan-IFT

harness

Routes feature, bug, and refactor tasks through a 7-stage agent pipeline with verify_all gating.

Community
Advanced
Alan-IFTAlan-IFT

harness-stream

Drains a continuously-growable task pool through a 7-stage pipeline one task at a time.

Community
Advanced
Alan-IFTAlan-IFT

harness-language

Switches a harness project's output-language policy between English and Chinese.

Community
Advanced
Alan-IFTAlan-IFT

harness-plan

Runs requirement analysis, solution design, and gate review stages to produce a vetted design verdict.

Community
Intermediate
Alan-IFTAlan-IFT

harness-batch

Runs a fixed task list through a 7-stage pipeline sequentially with isolated contexts.

Community
Advanced
Alan-IFTAlan-IFT

harness-status

Inspects Harness project assets, hook wiring, baseline state, and task progress to report a health snapshot.

Community
Advanced
Alan-IFTAlan-IFT

harness-init

Bootstrap new projects with the Harness Engineering skeleton and Claude Code binding.

Community
Advanced
Alan-IFTAlan-IFT

harness-verify

Runs the project's verify_all script and reports PASS, WARN, or FAIL results.

Community
Basic
Alan-IFTAlan-IFT

harness-upgrade

Upgrade stale harness projects to the current plugin layout non-destructively.

Community
Advanced

Frequently Asked Questions About Alan_IFT

FAQPage Schema
What tasks can I run using Alan_IFT's Harness skills?▼

You can run full 7-stage feature pipelines, design-only planning, feasibility research, fixed-list batch execution, continuously-growing task pools, budgeted goal loops, whole-codebase anti-entropy sweeps, requirement interviews, and structured PASS/WARN/FAIL verification of claimed-done work.

Who are the Harness Engineering skills designed for?▼

They target developers and engineering leads using Claude Code who want disciplined AI-driven development: solo builders bootstrapping new fullstack or backend projects, teams adopting conventions into existing repos, and operators supervising autonomous multi-stage task execution.

How do I start using the Harness skills on a project?▼

Run /harness-init to bootstrap a new project with the full skeleton, or /harness-adopt to non-destructively apply Harness conventions to an existing repo. Stale projects use /harness-upgrade with a dry-run preview, then prove health via a green verify_all run.

Can I control how much the AI decides on its own?▼

Yes. /harness-decision-mode sets three escalation modes: human decides every judgment call, AI decides per a preset rubric escalating only red lines, or AI decides per your custom rubric. /harness-intervene also injects redirect, pause, or add-task signals into running pipelines.

What prerequisites do the Harness skills require?▼

Skills run inside Claude Code and rely on its allowed-tools permissions including Read, Write, Edit, Glob, Grep, Bash, AskUserQuestion, TodoWrite, Task, and WebFetch. Verification requires the project's .harness/scripts/verify_all script, and supervision reads in-flight or archived task folders.