Adr
Drop-in ADR toolkit for AI coding agents (Claude Code, Cursor, Copilot, Codex). Skill + agent + instructions + template. MIT.
npx -y skills add rvdbreemen/adr-kit --skill adrAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 4 stars4 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Create or review Architecture Decision Records with ADR Kit's four verification gates and lifecycle rules.
The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.
SKILL.md
1.5 KB, as published. Nobody here has run it
ADR authoring
Select adr-kit:adr from /skills, or ask Copilot to use the adr ADR Kit skill.
Resolve <plugin-root> from this installed skill. Use only bundled,
local ADR Kit tools and follow this canonical workflow:
- Query the generated ADR index for relevant Accepted ADRs before writing, open only returned sources, and report conflicts.
- Run
python <plugin-root>/bin/adr profiles --format json; use only a returned profile withavailable: true. - Use
madras the default; supportnygardandcanonicalonly when returned as available, and never invent a profile or template. - Qualify
<subject>as a consequential, difficult-to-reverse decision before creating a record; do not create an ADR for an ordinary reversible implementation choice. - Create a Proposed record, then follow the canonical grill protocol for context, decision, alternatives, consequences, risks, mitigations, evidence, scope, ownership, conflicts, and Open Questions.
- Show the final acceptance packet and require an explicit
yesin this active session before invokingadr accept; require Completeness, Evidence, Clarity, and Consistency and supersede rather than rewriting Accepted decisions.
Do not contact another model or mutate user-owned instructions.