Thin Orchestrator Skills
User-invoked entry-point skills can be a line or two long, delegating to a model-invoked primitive by explicitly instructing the agent to call the Skill tool, so behaviour lives in one place and is composed rather than copied.
Evidence: moderate
Matt Pocock's most popular skill, grill-me, has a body of one sentence: call the Skill tool with "grilling". grill-with-docs is the same, calling the Skill tool twice, for "grilling" and "domain-modeling". The user-invoked skill is a named entry point; the behaviour lives in a model-invoked primitive that several entry points share (grilling, described in grilling the design tree, sits behind grill-me, grill-with-docs, triage, wayfinder and improve-codebase-architecture).
The wiring convention is specific. Dependencies are written as an operative instruction to call the Skill tool with a bare skill name, not as a /skill mention left for the model to interpret and not as a cross-folder file link. Naming the tool is reported to fire more reliably, and dropping the slash keeps the instruction harness-neutral. One skill per call: a step needing two skills says "call the Skill tool twice", because "call it with X and Y" reads as one call. A user-invoked precondition (such as running setup) is phrased as an instruction to tell the human, since no skill can fire it.
The payoff is a single source of truth for each discipline and very small skills: across the promoted set, entry points run 7 to 16 lines while the primitives carry the weight.
Boundary: this only works when the target is model-invoked, as described in model-invoked versus user-invoked skills.
Sources
- https://github.com/mattpocock/skills/tree/main/skills (read 2 Oct 2026, v1.2.3)
- https://github.com/mattpocock/skills/blob/main/.agents/invocation.md (read 2 Oct 2026)