docs(skills): add use-graph, the general entry-point skill - #107
Conversation
Routes to graph-project-setup, graph-plan-authoring, and graph-github-actions-setup for progressive context, and covers the everyday surface (running plans, ask/chat, catalog inspection) directly. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
|
Note graph_review_9000 · A pure documentation PR adding a new entry-point skill file that summarizes and routes to existing docs/skills, with no code changes. At a glance
🧵 0 new · 0 open · 0 resolved this run · 0 declined 🟡 Needs verification / intentional (1)Stated Intent — The PR body's technical claims (command syntax, exit codes, skill cross-references) cannot be verified against the actual CLI/docs source since those files weren't provided in the diff; the described content matches the added file itself but not necessarily current CLI behavior. ⚪ N/A · ✅ PassTest Coverage — Docs-only change; no behavioral code to test. Pipeline/Control-Step Invariants — No pipeline or control-step code touched. Template/Shape Engine — No template or shape-inference code touched. Rust/Async/Storage Footguns — No Rust code in this diff. Plan Composition/Model Routing — No plan composition, routing, or provider code touched. Repo Conventions/Docs Parity — New skill file follows the sibling skill format (frontmatter, GitHub-URL cross-references) as claimed; this is itself a docs addition so no separate docs-parity gap applies. CLI Surface/Copy — No CLI code, help text, or exit-code behavior changed — only prose describing existing behavior. |
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Summary
Adds
skills/use-graph/SKILL.md— a general entry-point skill that orients an agent on graph's everyday surface and routes deeper work to the existing skills for progressive context.graph-project-setup, plan creation/editing →graph-plan-authoring, CI →graph-github-actions-setup; docs named canonical over all skills.concepts.mdx: plans as chained tool calls, the single namespaced catalog, the shape cache, human-authored plans never replan, the streams/exit-code contract.plan list/show/validate/runwith input forms and exit-code reading,ask/chat+ threads as prototyping that seeds the shape cache,tools/shapes/mcpcatalog inspection.graph-plan-authoring.ask/chatis the agent loop.Test plan
docs/reference/cli.mdxanddocs/getting-started/concepts.mdx🤖 Generated with Claude Code