dev-guide-skill-stewardship

Guides distilling repeated experience into de-privatized, benchmarked SKILL.md files for the LingTai preset tree.

692|60|Updated Mar 15, 2026
One-click install
npx skills add https://github.com/Lingtai-AI/lingtai --skill dev-guide-skill-stewardship
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: dev-guide-skill-stewardship
Source: https://github.com/Lingtai-AI/lingtai/tree/main/tui/internal/preset/skills/lingtai-dev-guide/reference/skill-stewardship
Command: npx skills add https://github.com/Lingtai-AI/lingtai --skill dev-guide-skill-stewardship

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Teams repeatedly re-derive the same workflows across sessions and agents, or ship skills containing private paths, secrets, and unverified commands. This reference defines when a recurring pattern deserves a SKILL.md, how to distill session journals into triggered instructions, and how to scrub and validate a skill before it ships in a PR.

Core Features & Use Cases

  • Write-vs-skip triage: Criteria for deciding whether an experience becomes a skill, stays a session-journal entry, or remains a shared-library candidate.
  • Router vs nested-reference pattern: Rules for keeping root SKILL.md routers compact while nested references under reference/<topic>/ carry the detail, with same-commit catalog, routing-table, and test updates.
  • De-privatization and pre-publish checklist: Canonical rules for parameterizing paths, people, secrets, and IDs, plus a benchmark checklist covering frontmatter, triggers, verified commands, and last_changed_at timestamps.
  • Use Case: After noticing three agents independently re-implemented the same release workflow, use this reference to distill the pattern into a nested reference, scrub private details, run the checklist, and bundle it into a reviewable PR.

Quick Start

Ask the agent to evaluate whether a recently repeated workflow should become a skill and, if so, distill and de-privatize it following this stewardship reference.

Frequently Asked Questions about dev-guide-skill-stewardship

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

FAQPage Schema
When should I write a skill instead of leaving notes?

Write a skill when a non-obvious technique recurs across sessions or agents and would otherwise be re-derived ad hoc. Skip one-off solutions, practices already documented elsewhere, project-specific conventions, and constraints a test or regex could enforce.

How do I structure a router skill with nested references?

Keep the root SKILL.md as a compact router with a nested-reference catalog and routing table, and place detailed procedures in reference/<topic>/SKILL.md files. Adding a nested reference requires updating the parent catalog, routing table, and related tests in the same commit.

How do I remove private information from a skill before publishing?

Replace user-specific paths with placeholders like <your-lingtai-checkout>, named individuals with generic roles, secrets with <REDACTED> or environment variables, and concrete chat IDs or PR numbers with parameterized examples. Repo-relative paths and generic commands are acceptable.

What checks should a skill pass before shipping in a PR?

Verify the YAML frontmatter parses with a valid name, a trigger-only description, and a last_changed_at timestamp; confirm cross-linked skills exist; actually execute any commands to confirm claimed output; and complete de-privatization. Nested references also need catalog, routing-table, and test updates.

When should a skill go to the shared library instead of the preset tree?

Promote a skill to the shared library only when it is broadly reusable agent behavior that is not specific to LingTai development. Domain-specific or privacy-contaminated skills should stay in their project, and uncertain cases should start in the smallest local home.