en un clic
project-rye
project-rye contient 9 skills collectées depuis BLTGV, avec une couverture métier par dépôt et des pages de détail sur le site.
Skills dans ce dépôt
Operate Rye data safely for LLM agents. Use when implementing or executing agent read/write flows, connecting domain tables, selecting assertion keys, enforcing provenance, and applying agent-safe SQL patterns for context retrieval and auditable writes.
Register source accounts, containers, items, context profiles, and confirmation decisions in Rye without assuming external connector semantics. Use when ingesting Slack, email, Fathom, files, API records, Composio results, MCP connector output, or any other source material where an LLM must classify purpose, allowed contexts, provenance, and when semantic connections are safe to create.
Create or revise Rye onboarding scopes, source context policies, plugin bindings, and context-gap review packets before agents promote source evidence into graph knowledge.
Read Rye safely without writing. Use when an agent needs to understand what Rye currently knows, what was true as of a time, what tasks are open or done, what candidates still need review, or how accepted knowledge traces back to source evidence. This skill is read-only: SELECT queries only, no events, assertions, candidate status changes, promotions, inserts, updates, deletes, or audit writes.
Install, migrate, verify, and test Rye PostgreSQL deployments. Use when a user asks to bootstrap Rye in a database, apply core/profile migrations, run schema verification, seed sample data, or execute conformance checks.
Add new Rye domain conventions and profile migrations. Use when introducing new node/edge/assertion/event conventions, creating new profile-specific functions/views, and extending conformance tests without breaking Rye core guarantees.
Validate Rye-tracked import and update runs before or after target writes. Use when Codex needs deterministic checks for source_row, mapped_record, change-plan, metadata, Rye audit, old-value traceability, table declarations, approvals, touched IDs, or post-write verification evidence.
Select and draft reusable Rye modeling patterns. Use when a user wants to define common node, edge, assertion, event, artifact, JSON Schema, or assertion-key contracts for a domain or customer skill without changing Rye core tables.
Extract CSV and XLSX source tables for Rye-tracked imports. Use when a user needs to inspect tabular source files, emit row-level NDJSON with lineage, map source fields into destination records, group many source rows into parent records with source_set lineage, stage extracted data, or build a domain-specific intake skill on top of generic tabular primitives.