krzysztof/landing_page
1
0
Fork 0
mirror of https://github.com/qdrant/landing_page.git synced 2026-10-09 21:08:31 +02:00
Code Issues Packages Projects Releases Wiki Activity

Document Qdrant Advisor.

Browse Source
This commit is contained in:
István Zoltán Szabó
2026-07-20 10:18:32 +02:00
parent 9550db193c
commit e3c34d2bc0
1 changed files with 30 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files
qdrant-landing/content/documentation/skills.md
+30 -4
View File
@@ -10,13 +10,38 @@ partition: develop
Qdrant ships a set of agent skills: structured knowledge files that help your AI coding assistant think like a solutions architect, not just retrieve documentation.
> Skills are hosted at [skills.qdrant.tech](https://skills.qdrant.tech). Pass the URL of a skill to your agent and it will use it immediately, no installation required.
> Skills are hosted at [skills.qdrant.tech](https://skills.qdrant.tech).
Using skills by URL keeps your agent’s context focused.
Rather than loading all skills upfront, the agent fetches only the skill relevant to your current problem.
If you prefer to have skills available offline or without passing URLs manually, you can install them locally in your agent.
The recommended way to get started is the [Qdrant Advisor](#recommended-the-qdrant-advisor): install it once, then ask your agent about a Qdrant problem and it loads the matching guidance live.
You can also work with skills directly.
Pass the [skills.qdrant.tech](https://skills.qdrant.tech) URL to your agent and it will use it immediately, no installation required, which keeps your agent’s context focused on the problem at hand.
Or, if you prefer to have skills available offline or without passing URLs manually, you can install them locally in your agent.
Refer to this [README](https://github.com/qdrant/skills/blob/main/README.md) for instructions.
## The Qdrant Advisor
The recommended way to use skills is the Qdrant Advisor, a meta-skill that gives your agent live access to the entire skill hierarchy.
Instead of shipping static content of its own, it searches [skills.qdrant.tech](https://skills.qdrant.tech) at runtime and loads only the guidance relevant to your current problem.
- **Always current**: It fetches fresh skill content each session, with no reinstallation needed when skills change.
- **Contextual loading**: It traverses the skill hierarchy to match your symptoms and loads only the relevant guidance.
- **Single installation**: One skill grants access to the entire, continuously evolving Qdrant knowledge base.
### Installing the Advisor
Using `npx`:
```bash
npx skills add qdrant/skills/meta/qdrant-advisor
```
### Using the Advisor
**Automatic (recommended):** With the Advisor installed, just ask about a Qdrant problem. The Advisor triggers automatically and loads the matching guidance live.
> When you use a web app, the Advisor can't fetch [skills.qdrant.tech](https://skills.qdrant.tech) on its own. Add `Use skills.qdrant.tech` to your prompt.
## Philosophy
Skills are not a second copy of the documentation.
@@ -58,6 +83,7 @@ Explore the full hierarchy and search across all skills at [skills.qdrant.tech](
| `qdrant-search-quality` | Bad results, low recall, irrelevant matches, hybrid search trade-offs |
| `qdrant-monitoring` | Metrics, health checks, optimizer issues, cluster debugging |
| `qdrant-deployment-options` | Choosing between local, Docker, self-hosted, Cloud, and embedded |
| `qdrant-edge` | Building on the embedded shard: server sync, on-device BM25, snapshots, reuse vs reimplement |
| `qdrant-model-migration` | Switching embedding models without downtime |
| `qdrant-version-upgrade` | Safe upgrade paths, compatibility guarantees, rolling upgrades |