Inspects which common CLI tools are installed and configured, and switches contexts for kubectl, docker, gh, glab, argocd and kargo
Repo oldwinter/all-cli, read from GitHub 1 min ago. oldwinter has not joined Take My Repo.
oldwinter/all-cli is a public repo on GitHub by oldwinter. Inspects which common CLI tools are installed and configured, and switches contexts for kubectl, docker, gh, glab, argocd and kargo. It declares no license on GitHub. Public does not mean open source: without a license, nobody is granted the right to reuse it. It is written mainly in Go, has 1 stars and last changed on 2026-11-16. Nobody has reviewed it on Take My Repo yet, which says nothing about its quality. Read from GitHub on 2026-11-22.
In the maintainer's words. Shown from GitHub; Take My Repo does not keep a copy.
all-cli is a small CLI tool that inspects whether common CLI tools are installed/configured, and (in v0.1) manages “contexts” for:
kubectl (context + namespace)docker (docker contexts)gh (active account per host)glab (GitLab host context)argocd (Argo CD CLI contexts)kargo (default project)It also detects “current context” for other common tools in all-cli status (read-only), including aws, aliyun, wrangler, vercel, railway, netlify, argocd, kargo, mise, and k9s.
The inventory is intentionally broader than the context-switching set. It now also tracks common local CLIs grouped by category, including navigation (fd, rg, fzf, zoxide), shell/data helpers (eza, bat, yq), task/runtime tools (uv, just, mise), cloud/deployment CLIs (aws, aliyun, wrangler, vercel, railway, netlify), web automation (opencli), Kubernetes helpers (kubectx, kubens, kubecolor, krew), AI terminals (claude, codex, openclaw, opencode, gemini, grok, ccusage, litellm-proxy), and a few workflow tools such as linear and simplex-cli.
It defaults to human-friendly output and supports --json for stable machine-readable output (e.g. a future SwiftUI macOS app).
brew install oldwinter/tap/all-cli all-cli version
Prebuilt Linux binaries are published on
GitHub Releases as
all-cli_<version>_<os>_<arch>.tar.gz (for example
all-cli_0.0.0-33.1.71ef29f_linux_amd64.tar.gz). A checksums.txt file is
published alongside the archives when present.
# linux_amd64 example. Replace VERSION with the release version # (the git tag is typically vVERSION; archive names omit the leading v). VERSION=0.0.0-33.1.71ef29f curl -fsSL -O "https://github.com/oldwinter/all-cli/releases/download/v${VERSION}/all-cli_${VERSION}_linux_amd64.tar.gz" curl -fsSL -O "https://github.com/oldwinter/all-cli/releases/download/v${VERSION}/checksums.txt" sha256sum --ignore-missing -c checksums.txt mkdir -p "$HOME/.local/bin" tar -xzf "all-cli_${VERSION}_linux_amd64.tar.gz" all-cli install -m 0755 all-cli "$HOME/.local/bin/all-cli" export PATH="$HOME/.local/bin:$PATH" all-cli version
gh release download --repo oldwinter/all-cli --pattern '*linux_amd64.tar.gz'
also works if GitHub CLI is already installed. Put the extracted all-cli
binary on PATH before running all-cli version.
go build ./cmd/all-cli ./all-cli version
go install writes the binary to $(go env GOBIN) when GOBIN is set,
otherwise $(go env GOPATH)/bin. On a clean Linux install that directory is
often not already on PATH, so all-cli version fails with command not found
even though install succeeded. Add it to PATH before running the binary:
go install github.com/oldwinter/all-cli/cmd/all-cli@latest # Put GOPATH/bin (or GOBIN) on PATH so `all-cli` is found bin_dir="$(go env GOBIN)" if [ -z "$bin_dir" ]; then bin_dir="$(go env GOPATH)/bin" fi export PATH="$bin_dir:$PATH" all-cli version
To persist this, add the export PATH=... line to your shell profile.
The repository includes a pinned Go development container with GitHub CLI,
just, and pre-commit. In VS Code or another Dev Container client, choose
Reopen in Container. The container downloads Go modules and installs the
repository hooks automatically.
For a host-based setup:
brew install just golangci-lint pipx pipx install pre-commit pre-commit install
go mod tidy
git diff --exit-code -- go.mod go.sum
go test ./...brew install just
just help
just ci
just check
just build
just status --tools kubectl,docker --group-by noneRecommended day-to-day flow:
# run CI-equivalent checks just ci # run stricter local checks (ci + race + format check) just check # build and run quick smoke checks just smoke # exercise every commit-time hook just pre-commit
High-frequency recipes:
just ci: CI policy (tidy + format + repository policy + vet + tests + 80% coverage + lint)just check: stronger local gate (ci + race tests + three-pass stability tests)just policy: check source size, issue-linked debt markers, and AGENTS.md freshnessjust lint: enforce cyclomatic-complexity and duplicate-code thresholdsjust pre-commit: run every configured commit hook against all filesjust fmt / just fmt-check: format code or enforce formattingjust test-cover: run coverage and print per-package function coveragejust coverage-html: write coverage.html for detailed coverage browsingjust build / just build-release: normal build or release-style buildjust run ...: run from source (for example just run status --json)just status ... / just status-json ...: run status from the local built binaryjust smoke: quick binary sanity check (version + status)just release-check / just release-snapshot: validate or dry-run GoReleaserjust release: publish release artifacts (with confirm prompt)just tag vX.Y.Z: create and push annotated tag (with confirm prompt)Note about Go toolchain mismatch:
justfile now runs all Go commands through env -u GOROOT -u GOTOOLDIR go, which avoids common stale shell variable mismatches.just go-env.Command map (same grouping as all-cli --help):
| Group | Commands |
|---|---|
| Primary | status, report, current, catalog, describe, diagnose, doctor, fix, snapshot, diff, schema |
| Cloud platforms | aws, aliyun, wrangler |
| Tool integrations | mise, k9s, kubectl, docker, gh, glab, argocd, kargo |
| Other | version, options, completion |
Start with all-cli catalog to browse every tracked tool without running
external commands, then all-cli describe <tool> for one tool or
all-cli status for live checks.
all-cli version keeps its compact human-readable output. Add --json when a
script needs the version and build metadata as separate fields:
all-cli version all-cli version --json
Use all-cli options --json to inspect the effective global flags in a
script-friendly shape:
all-cli options all-cli options --json
all-cli status defaults to grouping by category and sorting tools by tool (A-Z).
all-cli status all-cli status --json all-cli status --ids all-cli status --tools kubectl,docker all-cli status --categories ai,cloud all-cli status --categories ai --missing-only all-cli status --missing-only --ids all-cli status --unconfigured-only --ids all-cli status --group-by none all-cli status --sort tool-desc all-cli status --sort category-desc all-cli status --sort attention all-cli status --timeout 10s all-cli status --no-progress all-cli status --summary-only all-cli status --quiet
Use --ids to print one checked tool ID per line after applying the status filters and
sort order. This mode suppresses the progress spinner, so stdout contains only IDs. If
you also pass --json, the full JSON report takes precedence.
Use --summary-only to print just the Summary: ... line without the tables or the
warning and error details, after applying the status filters. It is handy as a
prompt-friendly one-liner or a quick health check (for example with --tools,
--categories, or --missing-only). This mode also suppresses the progress spinner,
and is mutually exclusive with --ids. If you also pass --json, the full JSON
report takes precedence.
Use the global --no-progress flag to suppress progress indicators in human-readable
commands. status --quiet also suppresses the status spinner while showing only tools
with issues; JSON output never emits progress indicators.
Use --categories to check one or more registry categories at once. When combined with
--tools, both filters apply, so the result contains only tools matching both selections.
With shell completion loaded, comma-separated category values complete in place: for
example, --categories cloud,k<TAB> keeps cloud and offers k8s.
Use --missing-only to turn the inventory into a focused installation checklist. It can
be combined with --categories or --tools, and is mutually exclusive with
--installed-only.
Use --sort attention for a quick morning triage: tools that need attention (not
installed, unconfigured, or carrying warnings/errors) float to the top in tool order
while everything healthy stays below. It pairs with the summary counts printed under
every status table.
Use --unconfigured-only for the complementary setup checklist: installed tools that
report no configuration (the ones waiting for a login or init step). It composes with
--ids for scripting and is mutually exclusive with --missing-only.
all-cli current shows the active accounts, clusters, projects, and environments
reported by every installed context-aware tool in one compact view.
all-cli current all-cli current --tools kubectl,docker all-cli current --categories cloud,k8s all-cli current --unresolved-only all-cli current --ids all-cli current --summary-only all-cli current --json
Use --tools or --categories to check only the contexts you need and avoid invoking
unrelated CLIs. When combined, both filters must match.
Use --ids to print one installed context-aware tool ID per line for a picker or
shell loop. JSON output takes precedence when both flags are supplied.
Use --unresolved-only to focus on installed context-aware tools for which
all-cli could not read any active context. It is useful for quickly spotting a
missing login or local configuration; it can be combined with --tools,
--categories, --ids, or --summary-only. JSON output remains the complete
report and takes precedence.
Use --summary-only for a one-line, prompt-friendly count of installed
context-aware tools and the ones with a readable active context. JSON output
takes precedence when both flags are supplied.
Example text output:
TOOL CURRENT
aws profile=work region=us-west-2
docker context=desktop-linux
railway none
all-cli diagnose turns the same status facts into agent-readable diagnostic items with severity, evidence, suggested actions, autofix safety, and related tool IDs.
all-cli diagnose all-cli diagnose --json all-cli diagnose --tools kubectl,docker,gh --json all-cli diagnose --profile ci --json
all-cli doctor is the health-check entrypoint for humans or automation. Without --fix it is read-only and returns the same diagnostic report shape as diagnose.
all-cli doctor all-cli doctor --tools kubectl,docker,gh --json
Add --fix to install missing tools that have a known brew, npm, pipx, or go
recipe. Preview the commands with --dry-run first, and use --tools to limit
which tools are installed. --installer auto (the default) picks the first
supported installer found in PATH; pass --installer brew|npm|pipx|go to force
one. Each install command runs with a 10-minute timeout, and any failed install
makes doctor exit non-zero. With --json, the output follows
doctor-fix-report-v0.1.
all-cli doctor --fix --dry-run all-cli doctor --fix --tools gh,kubectl all-cli doctor --fix --dry-run --installer npm --json
all-cli fix is dry-run only in this release. It builds a fix plan from diagnostics and explicitly reports which items are blocked because automatic mutation is not allowlisted.
all-cli fix --dry-run all-cli fix --dry-run --tools aws,gh --json
Snapshots can be saved and compared later:
all-cli snapshot --json > before.json all-cli snapshot --json > after.json all-cli diff before.json after.json --json
Add --details to see the old and new values of each changed field directly in
text output, without opening the snapshot files or parsing JSON:
all-cli diff before.json after.json --details
Snapshot diff: added=0 removed=0 changed=1
- kubectl changed fields=current
current: {"context":"dev"} -> {"context":"staging"}
Values use compact JSON so spaces, quotes, and newlines remain unambiguous;
null means the field was absent. Added and removed tools keep their summary
lines. This flag works with --tools, standard input, and --exit-code.
--json keeps its existing full before/after summaries and ignores --details.
Add --exit-code when a script or CI job should return status 1 if any tool was
added, removed, or changed. The complete text or JSON report is still printed:
all-cli diff before.json after.json --json --exit-code
Use --tools to focus a comparison on selected tracked tool IDs without
recapturing or editing full snapshots. The summary and --exit-code reflect
only those tools. A selected tool absent from both snapshots has no changes.
The filter works with text, JSON, and standard input:
all-cli diff before.json after.json --tools kubectl,docker --exit-code
all-cli snapshot --json | all-cli diff before.json - --tools kubectl --jsonPrint the matching JSON Schema directly from the installed binary when another tool needs to validate these reports offline:
all-cli schema status > status.schema.json all-cli schema diagnostic > diagnostic.schema.json all-cli schema doctor-fix > doctor-fix.schema.json
Use - for either diff input to compare a saved snapshot with a live pipeline
without creating another file. Standard input snapshots are limited to 1 MiB:
all-cli snapshot --json | all-cli diff before.json - --jsonTurn a captured JSON snapshot into Markdown for an issue or pull request without probing local tools again. The report keeps the original timestamp, tool facts, warnings, and errors:
all-cli report --from before.json > report.md all-cli report --from before.json --tools kubectl,docker > cluster-report.md all-cli report --from before.json --categories cloud,k8s > platform-report.md all-cli report --from before.json --tools aws,kubectl --categories cloud,k8s > filtered-report.md all-cli snapshot --tools kubectl,docker --json | all-cli report --from -
--from - reads standard input, limited to 1 MiB. Add --tools to select tracked
tool IDs, or --categories to select tools by their captured category, from either
a file or standard input without editing or recapturing the snapshot. Category names
must be current built-in registry categories, but saved rows match their captured
category values. When both filters are present, a row must match both. Selected tools
keep their original order and captured facts; tools absent from the snapshot are
omitted. An empty value leaves that filter unrestricted. Add --json to emit the
selected status with diagnostics derived only from those facts instead of Markdown.
Without --from, the same filters limit which local tools are checked.
Machine-readable shape for status --json is also summarized as JSON Schema (schema_version v0.1).
all-cli status --json now includes additive English metadata for machine callers:
legend: explains shared fields such as installed, configured_state, capabilities, warnings, errors, and the meaning of metadata fields.metadata: explains purpose, configured_when, known keys inside current, suggested agent_actions, and short interpretation notes.diagnostics: structured diagnostic items derived from the tool summaries.all-cli diagnose --json and all-cli doctor --json emit diagnostic-report-v0.1, which includes:
summary: counts by info, warning, and error.tools: the source status summaries used for diagnosis.diagnostics: severity, problem, evidence, suggested_actions, safe_to_autofix, and related_tool.Example shape:
{
"schema_version": "v0.1",
"legend": {
"configured_state": {
"yes": "all-cli found enough local state to treat the tool as configured"
},
"metadata_fields": {
"purpose": "Short English description of what the tool is for"
}
},
"tools": [
{
"id": "kubectl",
"current": {
"context": "example-cluster"
},
"metadata": {
"purpose": "Kubernetes CLI used to inspect and operate clusters through kubeconfig contexts.",
"configured_when": "At least one kubeconfig context is available and readable.",
"current_field_descriptions": {
"context": "The active kubeconfig context name."
},
"agent_actions": [
"inspect_status",
"show_current",
"list_contexts",
"switch_context"
]
}
}
]
}Use all-cli catalog to browse every tracked tool without running any external
commands. catalog list, catalog ls, and catalog listing are aliases for
the full catalog; each alias can take an optional search term. Add an optional
search term to match tool IDs, names,
categories, binary names, and purposes; human table output shows the active
query. Use --categories to browse one or more exact registry categories;
category filters and search terms can be combined. Add --match-all to match
every whitespace-separated word anywhere in the metadata, instead of requiring
one contiguous phrase. Matching is case-insensitive; words may match different
fields, and an empty query still lists the full catalog:
all-cli catalog all-cli catalog list all-cli catalog list k8s all-cli catalog kubernetes all-cli catalog --categories ai,cloud all-cli catalog kubernetes --categories k8s,cloud all-cli catalog cloud --json all-cli catalog "tui kubernetes" --match-all all-cli catalog list "cloud account" --match-all --ids
Use --summary instead to see an alphabetical category breakdown and total
tool count without the individual rows. Search, --match-all, and category
filters apply before counting. No matches produces a total of zero:
all-cli catalog --summary
all-cli catalog "cloud account" --match-all --summary
all-cli catalog --categories ai,cloud --summary--summary and --ids are mutually exclusive. --json keeps the full catalog
report unchanged and takes precedence over either text mode.
Use --ids to print one matching tool ID per line for a picker or a shell loop:
all-cli catalog kubernetes --ids
all-cli catalog --ids | fzfIDs keep the catalog's category-then-ID order. A search with no matches prints
nothing in this mode. If you also pass --json, the full JSON report takes precedence.
Use all-cli describe <tool> to inspect the built-in purpose, configuration
criteria, context capabilities, and agent actions without running the tool:
all-cli describe kubectl all-cli describe kubectl --json
When the tool has dedicated all-cli subcommands, describe lists them using the
same command tree the CLI runs, so discovery and the next command stay in sync:
all-cli commands:
all-cli kubectl current Show current kubectl context and namespace
all-cli kubectl list List kubeconfig contexts
all-cli kubectl namespace <namespace> Set default namespace for the current kubectl context
all-cli kubectl status Show kubectl status and current context
all-cli kubectl use <context> Switch current kubectl context (and optionally set namespace for it)
If a tool ID is misspelled, describe and every --tools filter suggest a
nearby tracked ID when there is a clear match:
unknown tool ID "kubctl"; did you mean "kubectl"?
all-cli kubectl status all-cli kubectl current all-cli kubectl list all-cli kubectl use <context> --namespace <ns> all-cli kubectl namespace <ns>
all-cli docker status all-cli docker current all-cli docker list all-cli docker use <context>
all-cli gh status all-cli gh list --json all-cli gh use --hostname github.com --user <login>
all-cli glab status all-cli glab list all-cli glab use <host>
all-cli aws status all-cli aws current all-cli aws list
all-cli aliyun status all-cli aliyun current all-cli aliyun list
all-cli wrangler status all-cli wrangler current
all-cli mise status all-cli mise current
all-cli k9s status all-cli k9s current
all-cli argocd status all-cli argocd current all-cli argocd list all-cli argocd use <context>
all-cli kargo status all-cli kargo current all-cli kargo use <project> all-cli kargo use --unset
all-cli does not print or read plaintext tokens/secrets.--show-token flags and only uses official CLI outputs to determine “configured” state.Risky or operational behavior can be rolled out through the typed feature-flag registry documented in docs/feature-flags.md.
CLI observability is disabled by default. Explicitly enabling
ALL_CLI_FEATURES=telemetry-v1 activates only the sinks configured by
environment variables:
ALL_CLI_LOG_PATH for structured JSON slog events,ALL_CLI_METRICS_PATH for atomic Prometheus textfile metrics,ALL_CLI_TRACEPARENT for W3C trace correlation,SENTRY_DSN for contextualized failed-command events,POSTHOG_API_KEY for privacy-minimized command-use analytics.Telemetry omits arguments, environment contents, external command output, identity, and configuration values. See docs/observability.md for the data contract, dashboards, setup, and opt-out, and docs/runbooks for alert, release, rollback, and privacy procedures.
oldwinter/all-cli is a public repo on GitHub by oldwinter. Inspects which common CLI tools are installed and configured, and switches contexts for kubectl, docker, gh, glab, argocd and kargo.
It declares no license on GitHub. Public does not mean open source: without a license, nobody is granted the right to reuse it.
Its maker tagged it for macOS, Windows and Linux, or as Docker or cross-platform. That is declared by the maker; Take My Repo has not tested it.
Open "Take it" on its page: Take My Repo writes an install plan for your machine and hands it to your AI agent (Claude Code, Codex or Cursor), which asks you before each step. Take My Repo has not tested it on any machine.
Its last change on GitHub was on 2026-11-16. GitHub counts 10 open issues and pull requests. Take My Repo does not judge from that whether it is maintained.
Nobody has reviewed it on Take My Repo yet, which says nothing about its quality.
A series of DAGs/Workflows to help maintain the operation of Airflow
#1 Self hosted CMMS web & mobile application that allows you to manage enterprise maintenance for free - Computerized Maintenance Management System
Tools and living docs 🧬 for Debian-based servers and Web Applications
The open standard for Home Assistant instance health monitoring.
:construction: :hammer_and_wrench: shows a 503 error page when maintenance-mode is on.

我的第二大脑 second brain,我的数字花园 digital garden,用obsidian双链笔记软件写作而成

个人在用的 agent skills 集合,含自动触发与手动调用两类及 loop 维护套装,供 coding agent 用户按需单独安装
Extracted built-in skills and add-ins from the Factory Droid CLI
基于 Herdr 的本地优先多 harness 工作流控制面:确定性 coordinator 向 Droid、Codex、Claude Code 等交互式 agent 派发任务,保留 durable queue、lease、重试、去重和收据
This page is also available as structured data, with the same sources and dates.
/api/v1/resources/gh/oldwinter/all-cliTake My Repo has not tested this on your machine. Requirements come from the project's own documentation.