read-all-adrs

Reads every ADR in a repository and produces a decision inventory table before architecture work.

Updated Nov 1, 2024
One-click install
npx skills add https://github.com/mlorentedev/dotfiles --skill read-all-adrs-mlorentedev
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: read-all-adrs
Source: https://github.com/mlorentedev/dotfiles/tree/main/harness/skills/read-all-adrs
Command: npx skills add https://github.com/mlorentedev/dotfiles --skill read-all-adrs-mlorentedev

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Agents under time pressure sample Architecture Decision Records with keyword greps instead of reading the full corpus, leading to re-proposing already-rejected decisions or building on superseded ones. This Skill forces complete ingestion of every ADR and produces a verifiable decision inventory as the completion artifact. ## Core Features & Use Cases - Full corpus ingestion: Locates and reads every ADR in docs/adr/ (or the repo's documented equivalent) with no keyword sampling. - Decision inventory table: Builds a structured table with ID, title, status, one-line decision, and supersession relationships for each ADR. - Drift flagging: Surfaces contradictions, stale statuses, and missing supersession banners discovered while reading. - Use Case: Before starting a large refactor or architecture session, run this Skill to get an explicit "Read N/N ADRs" statement plus the inventory table, ensuring the architecture work is grounded in all prior decisions. ## Quick Start Ask the AI to read all the ADRs in this repository and produce a decision inventory before starting the architecture session.

Frequently Asked Questions about read-all-adrs

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

FAQPage Schema
How do I read all ADRs in a repository before a refactor?

Glob the docs/adr directory for all Markdown files, read each one fully rather than grepping for keywords, and build an inventory table with ID, title, status, decision summary, and supersession links. State completion explicitly as "Read N/N ADRs" with the table.

What is an ADR decision inventory and why create one?

An ADR decision inventory is a table listing every Architecture Decision Record with its status, one-line decision, and supersession relationships. It serves as verifiable proof the corpus was actually read and as input for architecture sessions and large refactors.

When should I skip reading all ADRs in a repo?

Skip the full ADR sweep for one-file, low-risk changes with no architectural surface, or when the repository has no docs/adr directory at all. Gating trivial work behind a full corpus read adds overhead without value.

Why is grepping ADRs not enough for architecture work?

A keyword grep only shows files mentioning a term; it cannot confirm a file is irrelevant when the term is absent. This leads to missing supersession banners and re-proposing decisions that were already made and rejected.

What should I do when ADRs contradict each other?

Flag the contradiction explicitly in the inventory's last column or a short Flags section, including stale statuses and missing supersession banners. Do not fix the drift unprompted unless the fix is trivial and directly in scope.