메뉴
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.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
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.
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.
| name | rye-installer |
| description | 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. |
For a new local trial, run the public bootstrap. Prefer
https://projectrye.dev/onboard; if that hosted endpoint is unavailable,
fall back to the GitHub-hosted script:
tmp="$(mktemp)"
curl -fsSL https://projectrye.dev/onboard -o "$tmp" || \
curl -fsSL https://raw.githubusercontent.com/BLTGV/project-rye/main/site/public/onboard -o "$tmp"
sh "$tmp"
rm -f "$tmp"
For a remote PostgreSQL database, run
the same bootstrap with --remote "$DATABASE_URL":
tmp="$(mktemp)"
curl -fsSL https://projectrye.dev/onboard -o "$tmp" || \
curl -fsSL https://raw.githubusercontent.com/BLTGV/project-rye/main/site/public/onboard -o "$tmp"
sh "$tmp" --remote "$DATABASE_URL"
rm -f "$tmp"
Install the agent onboarding skill with
npx skills add BLTGV/project-rye --skill rye-onboarding.
Start the first adoption unit with ./scripts/rye onboard --label ... --purpose ....
Use ./scripts/rye status to confirm the selected database and active
onboarding scope.
Use the lower-level install/conformance commands only when debugging migrations or release gates.
The first-run path writes .rye.env so later commands use the same database
without the user re-entering the connection string.
Install always applies schema migrations and syncs plugin metadata from
plugins/*/rye-plugin.json. That metadata includes contributed node, edge,
assertion, event, and artifact types plus onboarding questions, never-infer
defaults, validation hooks, and admin hooks. --seed is only for quickstart
example data.
curl -fsSL https://projectrye.dev/onboard | shcurl -fsSL https://raw.githubusercontent.com/BLTGV/project-rye/main/site/public/onboard | shcurl -fsSL https://projectrye.dev/onboard | sh -s -- --remote "$DATABASE_URL"curl -fsSL https://raw.githubusercontent.com/BLTGV/project-rye/main/site/public/onboard | sh -s -- --remote "$DATABASE_URL"npx skills add BLTGV/project-rye --skill rye-onboarding./scripts/rye local --fresh./scripts/rye remote --db-url "$DATABASE_URL"./scripts/rye plugins list./scripts/rye onboard --label "First Scope" --purpose "..."./scripts/rye status./scripts/install.sh --profiles ''./scripts/install.sh --profiles crm./scripts/install.sh --profiles pm./scripts/install.sh --profiles crm,pm./scripts/conformance.shAfter install, run SELECT rye_catalog() to confirm the instance is ready and see what's connected.
Rye can be deployed to Supabase via its MCP apply_migration or execute_sql tools:
btree_gin, pg_trgm (pgcrypto is pre-installed).apply_migration.SELECT rye.rye_catalog().Key differences from self-hosted:
postgres is not a superuser — RLS applies to all queries. Always set session vars.set_config() calls.SET app.current_role = 'admin' syntax fails through the MCP. Use set_config('app.current_role', 'admin', false) instead.link_record() to connect existing domain table rows to the graph.track_table() to attach CDC triggers for change tracking../scripts/verify.sh and security tests.schema/migrations without versioning and tests../scripts/migrate.sh.