Repository navigation
v0.1.1 — multiple agents & accounts, peek, sharper init/check/doctor - #8
Merged
Merged
Conversation
Deepen the readiness checklist via data, not Go: a `readiness:` block in each adapter manifest declares extra best-practice checks (file-exists / dir-exists / git-tracked) cited to the provider's own docs, so criteria are attributable, not sloop's opinion. Optional checks render as advisory info (•) so power-user features don't nag a basic setup. Also adds a workspace-level "AGENTS.md committed to git" check (a best practice every provider echoes), a "learn more" line citing each provider's source, and seeds claude.yaml with the .claude/agents check from Claude Code's best-practices page. Documents the new field in ADAPTERS.md.
`sloop init` no longer enables every AI CLI on the machine. It enables your primary tool by default; `--tools claude,cursor` or the interactive prompt add more per workspace. Downstream (sync/check/hooks) read that list, so a Claude-only folder isn't littered with GEMINI.md or .cursor. Add needsCanonical: sloop's portable-context layer (AGENTS.md + pointer files) is only set up when there's something to be portable across — two or more tools, or a native-mode tool. A lone pointer-mode tool keeps its own context file (e.g. CLAUDE.md from `claude` /init); sloop stays out of the way and manages the fleet and skills. init/sync/check all honor this. check now respects a context file the tool already owns (treated as ready, not a "repair" nag) and only asks for AGENTS.md in a canonical workspace. Tests updated to cover both the minimal single-tool and canonical multi-tool shapes.
Two init UX fixes from real-world testing: - Tool selection now asks about EVERY detected tool, including the primary — it isn't always Claude. The primary is defaulted on (a bare Enter keeps the simple case), but you can decline it; if nothing is picked, the primary is enabled as a fallback. - Only prompt for work that isn't already done: skip "Pre-fill AGENTS.md" when AGENTS.md exists, "Create standard folders" when they already exist, and "Install status hooks" when they're already installed (scaffoldNeeded / hooksNeeded gates). - Fix a latent bug where init opened two bufio readers on stdin (racing over buffered input); the whole interactive flow now shares one reader.
Three init polish fixes from testing: - The scaffold prompt now lists exactly the folders it will create, e.g. "Create standard folders (.claude/agents, .cursor/rules)?" — and only the missing ones, via missingScaffold. - init no longer claims to have scaffolded a folder that already existed: the scaffold loop skips dirs that are present, so the summary lists only what it actually created. - Ask about the primary tool first (primaryFirst), so Claude (or whatever the default is) leads the per-tool prompts instead of alphabetical order.
Tools create their own standard folders (and sloop already creates the skills dir it needs via the symlink), so scaffolding empty .claude/agents / .cursor/rules with a .gitkeep is unnecessary intervention. Drop the interactive scaffold prompt; `--scaffold` stays as an explicit opt-in for anyone who wants the layout.
Match the check command's readability: green ✓ / grey ✗ marks, and three clear
groups — AI provider CLIs, Multiplexer (powers ps/run/attach), and Mode. A
missing CLI is grey, not red, since the tools are optional. The Mode line now
explains itself ("ask — asks before actions" / "auto — runs without prompts").
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
v0.1.1 — run multiple agents & accounts, peek, sharper onboarding
Builds on v0.1.0. Three themes:
Run multiple agents & accounts
sloop run claude@review/-n/--nameruns a second agent of the same tool in one repo (session<repo>__tool__instance).-N/--newauto-names the next free slot (claude·2,claude·3…). Defaultsloop run claudestill re-attaches, unchanged.sloop profile add|ls|rmsaves a tool + env in global~/.sloop/config.yaml; launch it assloop run @<name>(e.g. a different account viaCLAUDE_CONFIG_DIR).--env KEY=VALinjects env one-off without a profile (~/$VARexpanded).ps, picker, status line) shows instances astool·instance.Peek — overlay a waiting agent
sloop peekfloats a waiting agent's live pane over your current screen so you can answer it and drop back withoutswitch-clientswapping your whole screen;sloop peek setupbinds a key. Needs tmux ≥ 3.2.⏳ N waitingbadge.Sharper onboarding (init / check / doctor)
sloop initis provider-respecting: asks about the tools you actually use (Claude-first), prompts honestly and skips already-done work, selects tools per workspace, and no longer offers to scaffold a provider's folders.sloop checkgains more readiness criteria, sourced from each tool's adapter manifest.sloop doctorgroups + colors output and explains themodeline.Notes
docs/design/profiles.md,docs/design/peek.md. Docs updated: README command menu,docs/guide/USAGE.md,docs/reference/CONFIG.md,CHANGELOG.md.ws__toolnames and no-flagsloop runbehave exactly as before; the session-name parser is tool-aware so older sessions keep parsing correctly.go build ./... && go vet ./... && go test ./...pass.After merge: tag
v0.1.1onmainto trigger the GoReleaser release + Homebrew formula update.