Glossary

Glossary

The vocabulary here is strict on purpose, because two axes never mix. Structure is what a thing physically is: a component, a plugin, a workspace, a marketplace. Quality is how good a plugin is once graded: the Bronze, Silver or Gold designation it earns. A skill library is a grade, not a separate artifact.

Every term below is defined in one sentence, and the sections run roughly in the order you meet these words elsewhere. Where you want the reasoning behind a rule rather than its meaning, the explanation pages carry that; where you want a check’s history, universal-checks.md carries that.

Structure - what a thing physically is

Component - the unit of reuse: one building block (a skill, command, subagent, hook, workflow, chain contract, or MCP server) that works standalone, droppable into any repo without adopting the rest.

Skill - the atomic capability: one SKILL.md directory packaging procedural knowledge an agent loads on demand.

Plugin - the unit of release: a package of components with a manifest and exactly one version, the installable artifact that carries the version.

Workspace - a directory holding several plugins developed together (for example product-on-purpose/), not itself installable.

Marketplace - a catalog that lists plugins for discovery and install, kept separate from any plugin it lists.

Quality - how good a plugin is

Skill library - the grade a plugin earns by conforming to the Standard, a designation rather than an artifact you install.

Tier - a named rung of conformance the gate verifies and reports, each rung including every requirement of the rung below it.

Bronze (Universal) - the start line: portable, agent-agnostic files (valid skills, a minimal library.json, an AGENTS.md) that run unchanged on any agentskills.io-compliant agent, backed by checks U1-U9 and U11-U18.

Silver (Convergent) - the multi-agent rung: the machinery (subagents, commands, workflows, chain contracts, per-target emission, semver governance) emitted in the right format for every target agent, backed by S1-S8 on top of Bronze.

Gold (Advanced) - the self-proving summit: deep lifecycle plus self-hosting CI (documented hooks, regression-covered chains, drift-checked generated files, release notes, a deprecation policy), backed by G1-G10 on top of Silver.

The grading machinery

The gate - the deterministic core that decides pass or fail: npx agent-skills-toolkit runs the checks and exits with a real status code, so a plugin proves itself in CI rather than relying on an opinion.

Check - one validation rule implemented as a portable module under scripts/checks/, emitting error or warn findings, where the gate fails on any error and never on a warn alone.

reqId - the stable identifier a check backs, prefixed by tier (U Universal, S Convergent, G Advanced), so a finding points at exactly one requirement.

Spine - the 35 checks the toolkit ships: U1-U9 and U11-U18 (Universal), S1-S8 (Convergent), and G1-G10 (Advanced), the gap at U10 being a check retired in Standard v0.11 whose number was never reused.

Burndown - the actionable list the tier report hands back naming exactly what blocks the next tier, keyed to reqIds, so the climb reads as a worklist rather than a guess.

Conformance - the property of meeting a tier’s requirements as the gate verifies them, with any claim above what is actually met flagged.

Eval (judgment-based evaluation) - the behavioral and qualitative review the askit-evaluate skill can run beside the gate as opt-in evidence, which informs a decision but never makes one.

Eval-set - a skill’s collection of cases proving it behaves and triggers correctly: golden examples, at least one anti-example, and a triggering set of {query, should_trigger} pairs.

Emission and the canonical files

library.json - the authored single source of truth for a plugin’s cross-agent metadata (name, version, tier, agent targets, prefix, component index), from which the native per-agent manifests are generated.

Emission - generating each component in the correct format for each target agent from that one library.json, so Claude Code and Codex stay in lockstep.

Per-target manifest - an agent’s native manifest (Claude’s .claude-plugin/plugin.json, Codex’s .codex-plugin/plugin.json) generated from library.json and drift-checked, so a hand-edited one is an error.

Chain contract - the agents/_chain-permitted.yaml file declaring, per component, which other components it may invoke, required only when chaining is actually used.

agentskills.io spec - the open Agent Skills specification that the Standard is a strict superset of at the Universal tier, so a Bronze plugin’s skills are portable across the broader ecosystem.

See also

• Conformance and tiers - how the checks roll up into the tier a plugin satisfies.
• Universal checks - every Bronze check, with its source file and how each one is discharged.
• STANDARD.md - the normative (RFC-2119) definitions these terms summarize.
• README.md - the toolkit overview and the catalog of skills, subagents, and commands.