hodlbod
5b8fef5b23
NIP fixes: - RelayMembers (13534): use NIP-43 `member` tags (not `p`) and set the required NIP-70 `-` protected tag. - Profile (kind 0): remove display-name support entirely (getter, setter, display() fallback, and the search weight). - Comment (1111): A/a tags now carry a real address, not the event id. - BlossomServerList (10063): normalize server URLs with normalizeUrl (HTTP), not normalizeRelayUrl (which forced wss://). - HandlerRecommendation (31989): fix inverted removeRecommendation filter; add setSupportedKind()/supportedKind() for the NIP-89 d-tag. - Report (1984): place the report-type string on the e tag (note reports) or p tag (profile reports); always emit the p tag. Docs/skills: - Add @welshman/domain docs (docs/domain/) and the welshman-domain skill. - Prune @welshman/util docs/skill of the moved Profile/List/Handler/Encryptable helpers; register domain in the sidebar, index, and skills README. - Apply accuracy fixes to the @welshman/app docs/skill. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01BsMjvv7krpZeHK1Njeneru
2.0 KiB
Welshman Agent Skills
Agent skills for working with the welshman nostr library in AI coding tools.
What are these skills?
Agent skills are Markdown instruction sets that are auto-loaded by AI coding tools (Claude Code, Cursor, Cline, etc.) when relevant to your current task. Each skill contains focused documentation about a specific part of the welshman nostr library. When you ask a question that matches a skill's domain, the tool loads the full skill content automatically, giving the agent accurate, up-to-date knowledge about welshman's APIs and patterns.
Installation
Via npx (recommended)
npx skills add coracle-social/welshman
Available skills
| Skill | Description |
|---|---|
| welshman | General overview, package map, getting started |
| welshman-util | Core nostr types, events, filters, tags, NIPs (profiles/lists/handlers/rooms moved to domain) |
| welshman-domain | Nostr event kinds as Reader/Builder classes: profiles, lists, rooms, handlers, zaps |
| welshman-lib | Utilities: LRU, emitter, deferred, task queue |
| welshman-net | Relay connections, request/publish, auth |
| welshman-router | Relay selection strategies |
| welshman-store | Svelte stores and Repository pattern |
| welshman-signer | Signing, login methods, encrypted events |
| welshman-feeds | Dynamic feed construction |
| welshman-app | Instance-based client: plugins, sessions, publishing, requests |
| welshman-content | Note content parsing and rendering |
| welshman-editor | Svelte rich-text editor component |
How skills activate
At startup, the AI reads each skill's description to understand what it covers. When you ask a question that matches a skill's domain — for example, asking about relay connections or feed construction — the full skill content is loaded automatically to inform the response.
You can also invoke a skill manually using its slash command:
/welshman
/welshman-net
/welshman-store
This is useful when you want to prime the agent with a specific skill before starting a task.