agentsclimarketplace

Shared system init shape parity

Skill ychampion/cskill-agents/agents/claude-code/skills/shared-system-init-shape-parity

Agent skills for coding CLIs, multi-agent runtimes, context engines, MCP extensions, and terminal tooling. Instead of using claude code's source code, give your agent skills to create your own!

Install
npx -y skills add ychampion/cskill-agents --skill shared-system-init-shape-parity

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

What its author says it does

Copied from the file, not written here

Centralize system init or capability payload construction in one builder so multiple transports emit identical schemas and only vary inputs or delivery.

SKILL.md

3.1 KB, 567 tokens by cl100k_base, as published. Nobody here has run it

SKILL: Shared System Init Shape Parity

Domain: transport-architecture Trigger: Use when the same system init or capability-announcement payload is emitted from multiple entrypoints or transports and downstream clients depend on one stable schema. Source Pattern: Distilled from reviewed remote-control, bridge transport, and capability-advertising implementations.

Core Method

Define one canonical builder for the init payload and make every entrypoint call it. Put field names, defaults, derived state, feature-gated fields, and final key shaping inside that builder, not in the callers. Each transport path should do only path-local work: gather inputs, prefilter or redact collections when policy requires it, and choose when to send the message. This keeps the wire schema identical across stream, bridge, headless, or future transports so clients and tests do not drift between supposedly equivalent init events.

Key Rules

  • Make the shared builder own the full schema contract, including optional and feature-gated fields plus derived values such as session identifiers or fast-mode state.
  • Let callers vary only inputs and delivery mechanics. They may filter commands, redact sensitive inventories, or load metadata asynchronously, but they should not hand-assemble sibling init objects.
  • If two paths describe the same conceptual event, route both through the same builder even when one yields into a stream and another pushes via a bridge writer.
  • Add new init fields once in the builder, then update callers only enough to supply the new inputs; never patch a field into one transport and plan to mirror it later.
  • Keep transport-specific policy outside the builder when it truly differs by caller. The builder should serialize the canonical shape from supplied inputs, not decide privacy, bridge safety, or connection timing on its own.
  • Review and test parity at the schema level, not just field-by-field intent, because drift usually appears as omitted, renamed, or differently defaulted keys.

Example Application

A CLI has a normal query stream and a reconnecting remote bridge. The query engine emits the first init event by calling the shared init-payload builder. On bridge connect, the bridge path filters unsafe commands, redacts local integration inventories, and calls that same builder before sending its own init event. Both consumers receive the same key set and layout even though the collection contents differ.

Anti-Patterns (What NOT to do)

  • Do not duplicate system init object literals in separate transports; new fields will land in one path and silently disappear from another.
  • Do not hide caller-specific policy inside the canonical builder if some transports need full metadata while others need filtered subsets.
  • Do not justify a near-copy serializer just because one path is "only for bridge" or "only for headless"; once clients depend on it, it is part of the public wire contract.

Gives 0 of the 12 instructions most project setup skills give in 567 tokens

Counted across 999 of the 1,637 authors here whose files we hold, read 2026-08-06

  • ask one question at a timein 29 of 999, across 28 files
  • detect the package manager from lockfilesin 28 of 999, across 9 files
  • present findings to the userin 25 of 999, across 4 files
  • explore current repo statein 24 of 999, across 3 files
  • update the agent skills block in place if it existsin 24 of 999, across 3 files
  • install husky lint-staged and prettierin 23 of 999, across 4 files
  • create the lintstagedrc filein 22 of 999, across 3 files
  • commit all changed filesin 22 of 999, across 3 files
  • run lint-staged to verify it worksin 22 of 999, across 3 files
  • initialize huskyin 21 of 999, across 2 files
  • create the husky pre-commit filein 21 of 999, across 2 files
  • create a prettierrc file if missingin 21 of 999, across 2 files

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.