meowary

Canonical context-gathering patterns for workflows -- QMD semantic search + web search in parallel, query construction, result handling, and resource anchoring. Load when implementing workflow Step 0.5 (Clarify) context-gathering logic. See also qmd skill for CLI mechanics.

Inbox capture and processing — file naming, capture format, source-note format, and processing rules. Load when creating a capture or source note, routing inbox items during evening or weekly processing, or distilling a source note into resource articles.

Read Jira issues — query assigned issues, sprint boards, and epics; extract facts for resource articles. Load when pulling Jira context for resource enrichment or when looking up a Jira issue key.

Jenkins CLI — query jobs, trigger builds, view logs, download artifacts, manage credentials and nodes. Load when working with Jenkins CI, checking build status, or troubleshooting pipelines.

Project dashboard format, lifecycle, and Step 0 state-reading protocol — tasks, dev log, status, companion file structure, cross-linking, and archiving. Load when creating a project, updating a dashboard, reading project state at the start of any lifecycle workflow, or archiving a completed project.

Multi-source retrieval strategy — 5-tier source priority (QMD → direct read → Confluence → Jira → web), citation format, confidence tags, staleness rules, and gap reporting. Load when answering any question that requires searching the knowledge base, citing sources, or reporting what is unknown.

Knowledge graph philosophy, tag conventions, health scripts, operation log, and maintenance rules for resource articles. Load when creating, enriching, merging, splitting, or deleting a resource article, or at the start of any /r workflow.

PR/MR lifecycle, CI monitoring, and issue management via gh/glab CLIs — title and commit conventions, authentication, and platform-specific subskill routing. Load when creating or reviewing a PR/MR, monitoring CI, managing issues, or pushing code changes; then also load the platform-specific subskill.

Area dashboard format and lifecycle — ongoing responsibilities, task states, Focus section, log, and archiving. Load when creating a new area, updating an area's tasks or focus, appending a log entry, or archiving a completed responsibility.

Codebase context file format, lifecycle, and enrichment rules — architecture, tech stack, build commands, test setup, coding conventions, CI, and key decisions. Load when working in any external codebase, reading or writing a codebases/<name>.md file, or when codebase-specific conventions, build commands, or architecture context are needed.

Read Confluence pages and maintain Confluence tracking — search, fetch, record page IDs in sync registry and article frontmatter, and transform page content into resource facts. Load when fetching a Confluence page, enriching a resource from Confluence, or updating meta/confluence-sync.json.

Journal layer philosophy and sub-skill index — append-only principles, front matter rules, reading list management, waiting-for list management, and routing to daily, weekly, and meeting sub-skills. Load when creating or updating any journal file, adding items to the reading list or waiting-for list, or routing to a sub-skill for format details.

Pack a repository or file subset into a single AI-friendly file using repomix. Load when preparing an external codebase for analysis, review, or planning — before scout, research, or peer-review workflows on large external repos.

Manage git worktrees via wt — create, switch, list, merge, and remove worktrees; check out MR/PR branches in isolation. Load when working on multiple branches simultaneously, checking out a PR/MR branch for review, or managing worktree lifecycle.

QMD CLI mechanics -- query types (lex/vec/hyde), query syntax, multi-get, and index maintenance. Load when issuing a qmd query, choosing query types for better recall, or re-indexing after bulk changes.

Prose quality rules — active voice, concision, formatting conventions, and editor checklist. Load when writing or editing any human-readable content: resource articles, notes, ADRs, RFCs, proposals, READMEs, or any document.