Template and conventions for authoring cross-machine sync scenarios in internal/harness/scenarios_test.go. Load when a bug or feature involves two or more machines, pushes/pulls, merge behavior, or profile state. Paired with ccsync-repro-first.
Run ccsync against a sandboxed $HOME and a local bare git repo for manual TUI verification. Use when a change affects what the user sees on screen (new screen, reshuffled flow, messaging) and unit tests alone can't confirm "the feature works" matches "the tests pass". Paired with the /ccsync-isolated command.
Reference map of non-obvious internal APIs in ccsync. Covers sync.Run inputs/outputs, manifest.Action semantics, three-way merge, first-sync (baseCommit=="") policy, PreserveLocalExcludes, PromotePath, profile extends, gitx init-vs-clone, secrets keying, and state policies. Load when editing internal/sync, internal/manifest, internal/merge, internal/profile, or internal/state.
Release checklist for ccsync. Covers version bump, full verify pipeline, release notes, commit + tag, and user-machine recovery instructions. Load when cutting any new release (patch, minor, or major).
Repro-first bug workflow for ccsync. When the user reports a bug, use this before editing any code — it produces a failing test that pins the bug, then a fix, then a re-introduction to confirm the test is load-bearing. Invoke on every bug report.
Bubbletea unit-test playbook for internal/tui screens. Covers minimal AppContext scaffolding, driving tea.KeyMsg, extracting tea.Sequence sub-commands via reflection, asserting on switchScreenMsg, and the autoJoinMsg pattern for auto-advancing screens. Load when adding or fixing TUI behavior.
Query patterns for the academic citation graph — Semantic Scholar, OpenAlex, Crossref, and DBLP. Reference for research-academic-graph and citation-graph-builder subagents. Covers endpoints, rate limits, response shapes, and jq snippets for filtering.
How to query the arXiv API effectively for CS topics. Reference for the research-arxiv subagent. Covers category codes, date filters, sort orders, Atom XML parsing, ID extraction, and the HTML view URL pattern used by paper-reader.