CommunityRédaction et éditiongithub.com

Karthick-Ramachandran/recall-os

🧠 Local-first CLI that gives your repo durable, AI-ready memory — features, ADRs, modules, architecture, standards & tests — and a 🩺 deterministic doctor that validates health, evidence & drift. Git tracks what changed; Recall OS tracks why.

Compatible avec~Claude Code~Codex CLI~Cursor
npx skills add Karthick-Ramachandran/recall-os

Ask in your favorite AI

Open a new chat with this agent skill pre-loaded.

Documentation

Skill: Create ADR

Purpose

Record durable architecture decisions so future humans and agents do not rediscover or contradict them.

Inputs

  • Decision topic.
  • Context and constraints.
  • Options considered.
  • Recommended or selected decision.

Required Reading

  • docs/10-architecture/ARCHITECTURE.md
  • docs/10-architecture/FILE_WRITE_POLICY.md
  • docs/20-security/SECURITY_MODEL.md
  • Relevant existing ADRs in docs/adrs/

Output Files

  • docs/adrs/ADR-####-<slug>.md

Process

  1. Determine whether the change needs an ADR.
  2. Find the next ADR number without reusing existing numbers.
  3. Write context, decision, alternatives, consequences, and related docs.
  4. Mark the status as Proposed unless the human has accepted the decision.
  5. Link to affected PRDs, architecture docs, feature docs, and security docs.

Stop Conditions

Stop and request human decision if:

  • The decision changes security posture, network behavior, telemetry, file writes, auth, secrets, cloud, or runtime MCP.
  • The requested decision conflicts with an accepted ADR.
  • The decision is not actually made yet and should remain a proposal.

Quality Bar

  • The ADR explains why the decision exists, not just what changed.
  • Alternatives and consequences are honest.
  • Status is clear.
  • Related documents are linked.

Skills associés

youwei16/akubun-writing-skill

Agent-ready Japanese writing kit based on 『悪文』 for review and revision

community

raja21068/AutoResearch

An automated AI research-code writer and AI research-paper writer implementation through a skills+ autoraters using any coding agent (Claude Code, Cursor, Antigravity, Cline, Aider). No API keys, no LLM SDKs.

community

calm-north/fix-linking

Design or audit internal link structure for a website. Use when the user asks about internal linking, link architecture, orphan pages, PageRank flow, anchor text, site structure, silo architecture, or why pages aren't ranking despite good content. For external link building, see build-links.

community

kostja94/medium-posts

When the user wants to write, publish, republish, or optimize posts on Medium.com (canonical tags, distribution, Medium SEO). Also use when the user mentions "Medium," "Medium article," "Medium story," "Medium publishing," "canonical Medium," or "publish on Medium." Not for general parasite SEO when Medium is not the target platform—use parasite-seo. For AI search visibility strategy, use generative-engine-optimization.

community

iliyanivanovmp-stack/linkedin-profile-audit-claude-skill

Claude Code skill that scores your LinkedIn profile and returns copy-paste headline rewrites, banner recommendations, and a full about section rewrite.

community

agricidaniel/seo-schema

Detect, validate, and generate Schema.org structured data. JSON-LD format preferred. Use when user says "schema", "structured data", "rich results", "JSON-LD", or "markup".

community