Official library of CLIs generated by the CLI Printing Press. Endorsed, tested, and community-contributed.
Created 2026-03-28 — roughly 45 days old at appraisal date. Despite youth, the repo ships 82 CLIs across 16 categories with per-CLI release tags (agent-capture-current, ticketmaster-current, etc.) and a commit on the appraisal date itself. The registry auto-generation tooling (tools/generate-registry) and the machine-readable registry.json catalog signal deliberate architecture rather than a dump of scripts. The missing license is a meaningful liability that caps the maturity score regardless of velocity. Calling it a well-structured beta with strong momentum.
README covers the full install surface: one-command starter pack, individual installs, bundle mixing, focused skill installs, and the complementary slash-command interface. Three highlighted use-case examples (ESPN, flight-goat, sentry-pp-cli) with concrete natural-language queries demonstrate real ergonomics rather than hypotheticals. Each CLI has its own directory and SKILL.md. An external browseable catalog at printingpress.dev supplements the in-repo table. The one gap: no explicit troubleshooting section and the private-catalog / GITHUB_TOKEN requirement is mentioned only in passing at the end of the install block.
Go is the correct choice for distributable CLIs — single static binary, fast, cross-platform, no runtime dependency on the end host. Registry auto-generation via tools/generate-registry and the dual-install orchestrator (npm thin wrapper → go install + skill fetch) show structural discipline. However: no license, no dependency manifest surfaced, no CI badges or test infrastructure referenced in the README, and code quality across 82 community-contributed CLIs is inherently variable. The SQLite mirror pattern in sentry-pp-cli (local store for compound cross-project queries) is a notable engineering decision worth calling out positively.
Last commit on the appraisal date (2026-05-12), coinciding with the latest release (ticketmaster-current). 16 open issues across 82 CLIs is a healthy ratio. Multiple named contributors visible in the catalog (cathrynlavery, tmchow, hnshah, rderwin, rderwin). Companion project (cli-printing-press) is actively linked. No signs of drift.
917 stars and 161 forks in ~45 days represents strong organic traction — equivalent growth rates outpace most CLI projects at 6 months. Community-contributed CLIs across at least five distinct GitHub handles indicate the contribution loop is functioning. Companion npm package and external site (printingpress.dev) extend discoverability beyond GitHub. Downstream dependents not directly measurable from available data.
Overall: 3.7/5
Category: Agent-Optimized CLI Catalog Known alternatives in vault: None. openclaw--gogcli (Google Workspace CLI) is the nearest analogue in the vault but is a single-purpose CLI, not a catalog or orchestration layer. No prior appraisals exist in this category. Differentiation: Three structural differentiators set this apart from anything in the vault: (1) Each CLI ships paired with a focused SKILL.md and slash-command interface — agents get a pre-baked, doc-free interface rather than raw binary flags. (2) Breadth is exceptional: 82 CLIs spanning travel, commerce, media, cloud, developer tools, marketing, food, and more in under 45 days. (3) The SQLite mirror pattern (exemplified by sentry-pp-cli) enables compound queries that the upstream APIs structurally cannot answer — this is an architectural decision, not a convenience wrapper. The npm orchestrator resolving bundles to Go binaries + skills in one shot is also distinctive. No alternative in the vault approaches any of these three dimensions. Gap or crowd: Significant gap. No vault repo serves as a multi-CLI catalog, skill library, or agent-tool distribution layer. Adding this fills a foundational tooling role, not an incremental one.
Score: 5/5
Harvestable: (1) The CLI + paired skill pattern (Go binary + SKILL.md + slash command per tool) is directly portable to any PAI skill system. (2) The registry.json catalog design — machine-readable, auto-generated, installable by name or bundle — is a reusable architecture for any tool catalog the PAI system might maintain. (3) The --cli-only / --skill-only flag convention is a useful design pattern for dual-artifact tooling. (4) The SQLite local-mirror pattern from sentry-pp-cli (sync remote data locally, run SQL the upstream API never built) is high-value for PAI data pipeline design. (5) Specific CLIs immediately usable as PAI tools: espn (live sports), sentry-pp-cli (compound developer observability queries), flight-goat (multi-source travel search), archive-is (paywall bypass + archival), allrecipes (structured recipe retrieval), airbnb (multi-source accommodation pricing).
Integration path: Near-trivial for consumption. npx -y @mvanhorn/printing-press install starter-pack delivers four binaries and four skills in one command. Individual skills install via npx skills add mvanhorn/printing-press-library/cli-skills/pp-<name> -g. The main friction is the private catalog requiring GITHUB_TOKEN or GH_TOKEN — a one-time environment setup. No rebuild needed; CLIs are ready-to-use Go binaries. Skill files drop directly into an existing PAI skill directory.
Overlap with existing: No vault repos overlap in purpose. openclaw--gogcli covers Google Workspace specifically and may be superseded in practice if this library includes a comparable Google Workspace CLI among its 82 entries (not confirmed from the partial README excerpt). The catalog role itself has zero overlap with anything currently vaulted.
Adoption cost: Trivial for consuming individual CLIs and skills. Moderate if integrating the registry.json catalog management layer into a custom PAI orchestration pipeline (requires reading the orchestrator source and mapping to the PAI tool-management layer).
The README's framing — "muscle memory for an agent: no hunting through docs, no wrong turns, no wasted tokens" — maps precisely to PAI infrastructure concerns and signals the author understands the agent-tool UX problem at a design level, not just an implementation level. The missing license is the single largest risk for any integration requiring legal clarity; this should be tracked and the repo re-evaluated if a license is added. The private catalog caveat (GITHUB_TOKEN required for live installs) adds a minor operational dependency that should be accounted for in automated PAI bootstrap scripts. Given the 45-day growth trajectory and functioning community contribution model, catalog breadth will expand materially; periodic reappraisal is warranted at 90-day intervals. The companion repo (mvanhorn/cli-printing-press) is the generation toolchain and may be worth a separate appraisal if PAI custom CLI generation becomes a use case.