Competitive Analysis
Compare ClawNet
Honest side-by-side pages for teams evaluating trust, identity, and distribution models for agent skills and open source workflows.
Identity and trust are separate layers. Sigma Identity is the companion decentralized identity system we use for BAP-first identity/OAuth flows, while ClawNet adds signed trust attestations and provenance.
clawnet CLI vs incur
incur is a TypeScript CLI framework from wevm (makers of wagmi and viem) that bakes agent discovery into every CLI you build — run npx my-cli skills add and the agent gets skills from that CLI. The clawnet CLI connects to a shared registry where skills are published independently, signed with BAP identities, and stored on-chain as Bitcoin SV transactions.
The core difference: incur distributes skills through a specific CLI package. clawnet publishes and installs skills from a shared, verifiable ecosystem registry.
Live note: incur was announced February 27, 2026 by wevm. The wevm/incur repo launched with 17 stars.
incur: Per-CLI Skill Bundling
Every CLI built with incur automatically gets skills add and mcp add subcommands. Skills travel as part of the CLI npm package. Agents discover skills by running the CLI directly — no separate registry lookup. The TOON output format reduces token consumption by up to 3× versus MCP alternatives.
clawnet CLI: Shared Registry with Signed Provenance
Skills published via the clawnet CLI land in a shared registry that any agent can query — not tied to a single CLI package. Authorship is BAP-signed and stored on-chain, so consumers can verify who wrote the skill and audit the full version history without trusting a centralized platform.
Scope Comparison
At-a-Glance Comparison
| Dimension | incur (wevm) | clawnet CLI |
|---|---|---|
| Primary job | Framework for building agent-aware CLIs | CLI client for the ClawNet skill registry |
| Skills scope | Bundled per CLI package — tied to one tool | Shared ecosystem registry — any author, any skill |
| Install command | npx <my-cli> skills add | clawnet skill install <slug> |
| Authorship model | npm publisher identity (no cryptographic proof) | BAP-signed BSV transactions — unforgeable |
| Token efficiency | TOON format — up to 3× fewer tokens vs MCP | Standard SKILL.md — works with all agent tools |
| MCP support | Built-in: npx <my-cli> mcp add | Separate concern — registry, not runtime |
| Bot deployment | Not in scope | clawnet bot deploy — Vercel sandbox management |
| Identity management | Not in scope | BAP identity creation, backup, P2P messaging |
| Discovery surface | Must know which CLI to run first | Browse registry by author, slug, or trust graph |
| Provenance depth | npm package version history | Timestamped BSV tx chain with AIP signatures |
| Best for | CLI authors who want great agent + human DX | Skill consumers who need verified authorship |
| Can they be combined? | Yes — incur CLI can publish skills to ClawNet | Yes — clawnet installs skills from any source |
Practical Positioning
incur is a strong choice when you want agents to discover skills through a CLI you already own — TOON output, Zod schemas, and free skills add subcommands with no extra infrastructure. It does not provide a shared registry, cross-author discovery, or cryptographic authorship proofs.
The clawnet CLI is the better fit when authorship and trust matter across the ecosystem: verifiable on-chain signatures mean any agent or developer can confirm who published a skill without trusting a platform. Bot deployment and BAP identity management extend the same CLI beyond skills alone.
The two tools are complementary: a CLI built with incur can publish its skills to ClawNet so consumers get both the token-efficient DX and verifiable, gatekeeper-free provenance.