Set up
Skills are installed by placing their folder where your agent loads skills.
The download comes from github.com/Sertxito/boost_backend itself, not from our server, so what you get is exactly what its author published.
Install instructions come from the README
README
synced Nov 1, 2026boost-backend
Official backend boost for GitHub Copilot. Standalone: agents, skills, prompts and instructions that Copilot discovers automatically.
Overview
boost-backend ships reusable backend reasoning assets, including:
- architecture guidance
- API design patterns
- refactoring playbooks
- performance heuristics
- security review criteria
- testing standards
- data access guidance
- backend DevOps recommendations
Prerequisites
- Node.js 20 or later
- npm 10 or later
Quick Start
This boost is self-contained: it does not need any runtime, plugin or MCP server. One command copies its assets to the places where GitHub Copilot / VS Code discover them automatically.
# From the root of YOUR project
npx github:Sertxito/boost_backend
# or, once published on npm
npx boost-backend
| Asset | Installed into |
|---|---|
| Agents | .github/agents/*.agent.md |
| Skills | .github/skills/<name>/SKILL.md and .agents/skills/<name>/SKILL.md |
| Prompts | .github/prompts/*.prompt.md |
| Instructions | .github/instructions/*.instructions.md |
| Global instructions | .github/copilot-instructions.md (only if the project has none) |
Options: --target <dir>, --force (overwrite), --dry-run, --no-agents-dir (skip .agents/skills), --check (verify install), --uninstall.
Existing files are never overwritten without --force, so several boosts can coexist in the same project.
Verify: run npx boost-backend --check, reload VS Code (Developer: Reload Window), then type @ in Copilot Chat (agents), / (prompts) and open the Configure Tools / Skills picker.
Package Layout
agents/ -> who acts
specs/ -> universal truths and non-negotiable guidance
skills/ -> operational procedures that can evolve
prompts/ -> runtime prompt composition per capability
instructions/ -> domain instructions reused by agents and skills
evals/ -> evaluation criteria and test cases
examples/ -> examples of high-quality outputs
docs/ -> package documentation and contribution guides
Mental Model
Agent = who I am
Spec = what I know
Skill = how I work
Prompt = how I compose
Eval = how I measure quality
Example = what good looks like
Local Customization
Installed files are plain files in your project (.github/). Edit them freely; re-run with --force to restore the originals.
Quality Checks (Maintainers)
Run all repository checks before publishing:
npm run quality
This runs structure validation, references validation, eval checks, artifact alignment checks, and placeholder checks.
Naming and collisions
Installed assets are prefixed with backend- (e.g. .github/agents/backend-<name>.agent.md, .github/skills/backend-<name>/) so several boosts can live in the same project, and internal references are rewritten to match.
- Nothing that already exists in the target is overwritten: each ignored file is reported with a
[WARN] ... already existed ... IGNOREDmessage. Use--forceto overwrite. - An existing
.github/copilot-instructions.mdis kept; the boost's global guidance is installed asbackend-global.instructions.mdinstead. - Re-running the installer is safe;
--checkand--uninstallwork from the manifest.