Repository Guidelines
## User Requirements
- Always use Context7 MCP for library/API documentation, code generation, setup steps, and configuration guidance without requiring an explicit user prompt.
## Icons (desktop renderer)
- Every icon
resources). The primary language is Go; HCL appears in acceptance test configurations and website documentation.
## Agent Registry
This project uses specialized personas for different tasks.
### Available Personas
- **`@contributor`**: [Contributor Persona
more details on the types of tests and how to run them.
### 4.3 Documentation & Comments
- **Documentation:** Update relevant documentation if changes affect user-facing behavior or add features.
- **Comments:** Include
showcase test server; the stale-build hazard | Model-invoked, or `/test-servers` |
Longer-form human documentation lives in [`docs/`](./docs) — see the table in the
[README](./README.md#documentation).
## Project Structure
```
inspector
insufficient, and compatibility impact. Ask if those
choices are unclear.
- Update relevant help, schemas, documentation, and tests in the same change.
Describe the public CLI change in the pull request
Documentation instructions
## TypeScript documentation
- Make all TypeScript documentation changes under `docs/v2/typescript/`.
- Add new TypeScript pages to the v2 TypeScript navigation in `docs/docs.json`.
- Update existing v2 TypeScript pages instead of their
Copilot SDK docs style guide
This style guide applies to all documentation in the `docs/` directory. These docs are synced to `github/docs-internal` via a normalization pipeline, so they must follow
src/sdk/java` or exported
as `java`. Keep API usage examples in [README.md](README.md), the
[SDK documentation](../docs/getting-started.md), and public Javadoc rather than
duplicating them in these contributor instructions.
## Layout and compatibility
public struct can break callers'
struct literals; preserve compatibility with existing construction patterns.
- Document public APIs with rustdoc. Keep cross-language parity notes in
[README.md](README.md), rather than repeating comparisons
runs `load_dotenv()`
before the submodule imports (hence the `# noqa: E402`).
## Data handling
- **No documents, no ingestion.** "Data" here is the **expense event payload**
moving between graph nodes, shaped
changing the repo rather than lifting from it.
> **The code is the documentation.** Each pattern below points at the actual production
> function that implements it — not a parallel toy example
clone-and-study RAG agent grounded on **Agent Platform Search** (Discovery
Engine). Drop documents in a GCS bucket and a managed Data Connector ingests,
chunks, embeds, and indexes them
Vector Search 2.0**, with a
Kubeflow Pipelines (KFP) pipeline that loads, chunks, and ingests documents into
a Collection that auto-generates embeddings server-side. The hard/interesting
parts are **Terraform** (Collection
tree,
and redact private data before uploading.
Commit media only when it serves maintained documentation, product assets, or
test baselines, not merely to obtain a public demo URL. If attachment
most paths; usually just adding a row to its provider matrix is enough.
### 7. Documentation
- `README.md`: add the provider to the "Supported Providers" list if it's user-visible.
- `docs/providers.md
account provides a singleton read session + a management UI, for authoring collections of context documents that agents read as observations. Collections have one of two visibilities: **private** (owned