spectralangel/ai-plugins (code-change-discipline)

How to find, read, edit and verify code in an existing repository without burning context or guessing. Invoke it BEFORE the first search and BEFORE the first edit of any code task, at any size — a one-line fix and a thousand-site migration use the same ladder. Triggers on: fix this bug, why is this failing, debug this, refactor X, rename X, extract this, add a field, add a parameter, add an endpoint, change how X works, wire up X, update the handler, remove this code, delete dead code, where is X handled, find every place that X, which files do I need to touch, is this still used, make this change, apply this fix. Covers the codebase-memory-mcp discovery ladder and its silent truncation signals, when to fall back to LSP and when to fall back to grep, the narrow-window Read-then-Edit protocol, and verification by test count. Not for prose or docs-only edits, config-file tweaks, brand-new standalone scripts, or conversation that touches no repository.

Claude Skills are markdown + scripts loaded into Claude Code; scanned with the same pipeline.

Source:

github.com/spectralangel/ai-plugins

Provenance:

Found by crawling (provenance, not a security rating)

We found this by crawling public sources and nobody has attested to it. That is how most of the registry gets here, and it is not a mark against the artifact — it only means no authenticated act of publication is on record. An artifact can be the vendor's own and still appear here.