Skills
Install Galya agent skills so your AI coding agent can integrate perception and taste correctly.
Use Galya skills to teach your AI coding agent how to integrate Galya — indexing content, capturing engagement, and searching, reranking, or scoring by a user's taste.
A skill is a portable SKILL.md module you drop into your agent runtime (for example Cursor). Once installed, the agent can follow Galya's published SDKs and REST API without inventing its own integration patterns.
Skills are maintained in a public repo, separate from this API:
github.com/GalyaIntelligence/galya-human-perception-skills
Install in Cursor
Clone the skills repo, then copy the skill folders into your project's .cursor/skills/ directory (or into ~/.cursor/skills/ for all projects):
git clone https://github.com/GalyaIntelligence/galya-human-perception-skills.git
cp -r galya-human-perception-skills/skills/* .cursor/skills/For global install:
cp -r galya-human-perception-skills/skills/* ~/.cursor/skills/After copying, restart Cursor or reload the window so the skills are picked up. Prefer the skill that matches your task (for example galya-getting-started for a first integration, galya-personalized-rerank for taste-aware ranking, or the score section in galya-rest-api / galya-agents-tools for POST /v1/score).
Other agent platforms
Each skill is a self-contained SKILL.md with YAML frontmatter (name, description). Copy the folder into your platform's skills directory, or adapt the frontmatter to that platform's skill format. Install and usage details for every skill live in the skills repo.
What the skills cover
| Area | Examples |
|---|---|
| Concepts & auth | Perception/taste model, API keys, headers |
| Indexing | Single-item and batch catalog ingest, linking to entities |
| Browser tracking | @galya/sdk, DOM signals, SPA lifecycle |
| Personalized retrieval | Search, rerank, score, recommendations, clusters |
| Agents | @galya/agents tools (galya_score included), taste loops, explain/ask |
For the full skill index and package list, see the skills repo README.
Coming soon
A /taste-align skill for taste-aligned agent responses is planned in the same repo. Until then, use the Agentic Affinity Alignment guide with POST /v1/rerank or POST /v1/score.