trellis-spec-bootstrap

Generates project-specific Trellis coding specs from real codebase analysis.

Updated Jun 8, 2026
One-click install
npx skills add https://github.com/SilentFlower/flower-trellis --skill trellis-spec-bootstrap-silentflower
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: trellis-spec-bootstrap
Source: https://github.com/SilentFlower/flower-trellis/tree/main/.agents/skills/trellis-spec-bootstrap
Command: npx skills add https://github.com/SilentFlower/flower-trellis --skill trellis-spec-bootstrap-silentflower

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? New Trellis projects ship with generic spec templates full of placeholder text that give agents no real guidance. This Skill analyzes the actual repository and rewrites .trellis/spec/ into concrete, source-backed coding guidelines. ## Core Features & Use Cases - Repository Analysis: Uses GitNexus code graphs, ABCoder AST parsing, and direct source reads to map package boundaries, runtime layers, and core abstractions. - Spec Decomposition & Writing: Splits spec work by package or layer only when the codebase justifies it, then writes rules backed by real file paths, examples, and anti-patterns. - Placeholder Verification: Runs a final pass to remove template boilerplate, fix index files, and confirm every claim traces to source or tests. - Use Case: After running trellis init on a monorepo, invoke this Skill to replace the default spec templates with package-specific guidance derived from the actual TypeScript sources and tests. ## Quick Start Analyze this repository and rewrite the .trellis/spec directory with concrete, source-backed coding guidelines for each package.

Frequently Asked Questions about trellis-spec-bootstrap

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I create project-specific Trellis spec files?▼

Run the spec bootstrap workflow after Trellis initialization: analyze the repository with GitNexus or ABCoder, decompose work by real package or layer boundaries, then write spec files citing concrete source paths. Finish by grepping for leftover placeholder text and fixing index files.

What tools analyze a codebase for writing coding guidelines?▼

GitNexus builds a code knowledge graph for module boundaries, execution flows, and impact analysis via `npx gitnexus analyze`. ABCoder parses code into UniAST for exact signatures and type shapes. Direct source and test reads verify findings before they become spec rules.

Does this workflow require a specific AI agent platform?▼

No, the workflow is platform-neutral and uses a single-agent model by default. GitNexus and ABCoder are configured through whatever MCP mechanism the host provides, and optional helper subagents are an implementation detail, not a requirement.

When should Trellis spec work be split into multiple tasks?▼

Split only along real ownership boundaries: one task per package with distinct conventions, per layer when frontend and backend rules differ, or per cross-cutting pattern. Small libraries need a single focused spec pass rather than artificial decomposition.

Why do Trellis spec templates need to be rewritten?▼

Default templates contain generic advice and placeholder prose that give agents no actionable guidance. Effective specs must cite real source files, tests, and repeated local patterns, with non-applicable template sections deleted entirely.