[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M799WQRF7SXSKM5YGSDF9E4G":3,"skill-same:01M799WQRF7SXSKM5YGSDF9E4G":121},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":26,"commitSha":27,"skillDirectory":28,"forksCount":14,"repositoryHtmlUrl":29,"repositoryHomepageUrl":30,"ownerHtmlUrl":31,"ownerBlogUrl":32,"ownerDescription":33,"updatedAt":34,"install":35,"resources":48},"01M799WQRF7SXSKM5YGSDF9E4G","data-store",null,"Govern provider-agnostic data reads and state transitions through declared data-source operations, not model-authored raw queries.","ZachDreamZ\u002Frunx","ZachDreamZ","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F22769595?v=4","01M799NTP9N5VYWB4JY50MC7Y3","NexusCore","01M799NTQ1AYT8TZC0XTSEZXDJ",0,"MIT","https:\u002F\u002Fgithub.com\u002FZachDreamZ\u002Frunx\u002Fblob\u002Fmain\u002Fskills\u002Fdata-store\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: data-store\ndescription: Govern provider-agnostic data reads and state transitions through declared data-source operations, not model-authored raw queries.\nrunx:\n  category: data\n---\n\n# Data Store\n\nOperate a data source through a governed adapter contract. This skill gives an\nagent enough context to read, append, or project state without learning provider\nsecrets, inventing SQL, or depending on one storage backend.\n\nThe storage backend can be Postgres, SQLite, D1, Redis, DynamoDB, S3, a ledger,\nor a product API. The runx boundary is the same: a declared data source exposes\ntyped operations; the graph supplies bounded params; the adapter executes the\noperation; the receipt records the resource, authority, idempotency, version,\ndigest, and redaction evidence.\n\n## Adapter selection\n\nThe operator chooses a data source at run time. The skill receives\n`data_source_ref` and operation inputs; project or hosted configuration binds\nthat ref to the concrete adapter. A local development ref might be\n`local:\u002F\u002Frunx-data-store\u002Fdev-board`. A production ref might be\n`tenant:\u002F\u002Facme\u002Fboard` bound to `data.postgres`, `data.d1`, `data.redis`, or a\nproduct-owned HTTP adapter.\n\nDo not put provider logic in the domain skill. Messageboard, CRM, support, and\nbusiness-ops skills should ask for durable facts to be read or written; the data\nsource binding decides whether those facts live in local JSON, SQL, Redis, D1,\nobject storage, or a product API. Switching providers is a binding change, not a\nrewrite of the skill.\n\nThe bundled OSS profile calls `data.source`. Unbound `local:\u002F\u002F...` refs default\nto durable local SQLite under `.runx\u002Fdata\u002Flocal-sources\u002F`, with one source-scoped\ndatabase file per logical ref, so stateful skills can be dogfooded without\nstanding up hosted infrastructure. Pass `store_id` only when a fixture\nintentionally wants the deterministic `data.local` JSON store. The graph inputs\nstay the same when a project later binds the source to Postgres, Redis, D1,\nobject storage, or a product API.\n\nAdapter preference is operator configuration, not model choice. To choose Redis,\nSQLite, or a hosted provider, bind the same `data_source_ref` through\n`RUNX_DATA_SOURCES` or `.runx\u002Fdata-sources.json`; do not add provider branches to\nthe domain skill.\n\n## What this skill does\n\n- Reads data through named queries or read operations declared by a data-source\n  adapter.\n- Appends state transitions with idempotency keys and expected versions.\n- Reads projections, event streams, or bounded latest-stream-head pages so\n  loops can resume from explicit state without exporting full history.\n- Produces receipt-bound evidence for data source, resource, operation, params,\n  row\u002Fevent limits, versions, and output digests.\n- Keeps product semantics outside the data layer. Messageboards, CRMs, billing\n  ledgers, and support desks define their own events and reducers.\n- Ships a fixture adapter (`data.local`), durable local SQLite adapter\n  (`data.sqlite`), and Redis adapter (`data.redis`) behind the same operation\n  envelope.\n\n## When to use this skill\n\n- A graph needs durable state between turns, such as queue position, board\n  state, sync cursor, review status, or approval inbox state.\n- A skill must query a bounded slice of product data before deciding the next\n  action.\n- A workflow needs to append an auditable event or effect transition with\n  optimistic concurrency.\n- An operator wants one provider-agnostic shape that can later move from local\n  JSON or SQLite to Postgres, Redis, D1, Supabase, Turso, DynamoDB, or another\n  store.\n\n## When not to use this skill\n\n- To let a model write arbitrary SQL, Redis commands, or database migrations.\n- To export broad data sets, secrets, raw PII, or unrestricted tables.\n- To hide product decisions in storage code. Domain skills still own state\n  machines, acceptance criteria, and business rules.\n- To treat a projection as independent truth when the event stream or receipt\n  chain is available and required for review.\n- To bypass payment, send, deploy, moderation, or human approval gates.\n\n## Procedure\n\n1. Identify the domain skill and transition first. The data store is a carrier,\n   not the policy owner.\n2. Select the logical data source. Use `data_source_ref` to name the project or\n   tenant source; let the project binding choose the adapter. Do not put raw\n   database URLs, provider credentials, or SQL in the skill input.\n3. Select a declared operation: named read query, append event, read events,\n   read projection, or list stream heads. Do not synthesize raw provider\n   commands.\n4. Check authority. Reads need the narrow resource\u002Fquery scope; writes need the\n   transition scope, idempotency key, and expected version unless the operation\n   is explicitly append-only without concurrency.\n5. Bind typed params. Enforce row\u002Fevent limits, tenant\u002Fpartition keys, and\n   redaction rules before the adapter runs.\n6. For writes, use optimistic concurrency and idempotency. A retry with the same\n   idempotency key and same payload returns the existing effect; a different\n   payload under the same key is a conflict.\n7. Return the operation result with resource refs, version movement, digests,\n   redaction notes, and stop conditions. Receipts should link this data effect\n   to the domain transition that caused it.\n\n## Edge cases and stop conditions\n\n- `needs_source`: the data source, resource, query name, tenant key, or schema\n  summary is missing.\n- `needs_input`: required operation params are incomplete, malformed, or not\n  specific enough to bind a declared data-source operation.\n- `needs_authority`: the caller lacks the declared read\u002Fwrite scope or provider\n  grant.\n- `needs_version`: a mutating operation lacks `expected_version` where the data\n  source requires optimistic concurrency.\n- `conflict`: the current version differs from `expected_version`, or an\n  idempotency key is reused with different content.\n- `too_broad`: the requested read lacks partition filters, exceeds limits, or\n  asks for raw export.\n- `redaction_required`: the operation would return secrets, private PII, or\n  fields outside the declared projection.\n- `provider_unavailable`: the adapter cannot reach the data source, times out,\n  or cannot prove whether a write committed.\n\n## Output schema\n\nAll runners return `runx.data.operation_result.v1`:\n\n```json\n{\n  \"schema\": \"runx.data.operation_result.v1\",\n  \"data_source_ref\": \"local:\u002F\u002Fexample\",\n  \"provider\": \"local-json-event-store\",\n  \"operation\": \"append_event\",\n  \"resource\": \"board_events\",\n  \"aggregate_id\": \"posting-123\",\n  \"status\": \"committed\",\n  \"before_version\": 0,\n  \"after_version\": 1,\n  \"idempotency_key\": \"posting-123:create\",\n  \"event_ref\": \"board_events:posting-123:1\",\n  \"result_digest\": \"sha256:...\",\n  \"projection_digest\": \"sha256:...\",\n  \"rows\": [],\n  \"events\": [],\n  \"redactions\": [],\n  \"stop_conditions\": []\n}\n```\n\nProvider adapters may add provider evidence under `provider_evidence`, but they\nmust not expose credentials or raw secret material.\n\nFor event streams, adapters derive a readable `event_type` in this order:\nexplicit `event.type`, explicit `event.event_type`, then\n`event.effect_family + \".\" + event.operation`. Domain skills that emit the\ngeneric `runx.effect.transition.v1` packet should include `effect_family` and\n`operation` on every event so readback projections say `messageboard.accept`,\n`business_ops.route`, or another meaningful transition instead of `data.event`.\n\n## Worked example\n\nA messageboard skill decides that `posting.claimed` is allowed. It emits a\ndomain transition packet. The graph then calls `data-store.append_event` with\nresource `board_events`, aggregate id `posting-123`, expected version `2`, and\nidempotency key `posting-123:claim:agent-9`. The data adapter appends the event\nonly if the stream is still at version `2`. The receipt proves the decision,\nthe data operation, and the new version. A later loop turn calls\n`data-store.read_events` or `read_projection` to resume from the explicit board\nstate.\n\n## Inputs\n\n- `data_source_ref` (required): stable logical ref for the data source. The\n  project or hosted binding maps this ref to the concrete adapter and provider\n  profile.\n- `resource` (required): declared resource, stream, table, keyspace, or\n  projection name.\n- `operation` (required for tool-level use): `append_event`, `read_events`,\n  `read_projection`, or `list_stream_heads`.\n- `aggregate_id` (required for event operations): stream or partition key.\n- `event` (required for `append_event`): domain event or transition packet.\n- `idempotency_key` (required for writes): stable retry key.\n- `expected_version` (required when the source enforces concurrency): current\n  stream\u002Fresource version expected by the caller.\n- `limit` (optional): maximum rows or events to return.\n- `after_version` (optional for `read_events`): return an ascending page whose\n  event versions are strictly greater than this value. Omit it to retain the\n  existing latest-tail read. Compare the last returned event version with\n  `after_version` in the result envelope to know whether another page remains.\n- `event_types` (optional for `list_stream_heads`): at most 20 exact latest\n  event types. No pattern or arbitrary field queries are accepted.\n- `cursor` (optional for `list_stream_heads`): opaque cursor returned by the\n  previous page. Limits are capped at 100.\n- `store_id` (local fixture adapter only): deterministic local store id that\n  opts into the bundled `data.local` proof adapter. Omit it for durable local\n  SQLite. Production adapters should ignore it.\n\n## Invocation examples\n\nDurable local dogfood with the bundled default:\n\n```bash\nrunx skill data-store append_event \\\n  -i data_source_ref=local:\u002F\u002Frunx-data-store\u002Fdev-board \\\n  -i resource=board_events \\\n  -i aggregate_id=posting-123 \\\n  --input-json expected_version=0 \\\n  -i idempotency_key=posting-123:create:v1 \\\n  --input-json event='{\"type\":\"posting.created\",\"payload\":{\"title\":\"verify a receipt link\"}}' \\\n  --json\n```\n\nFixture-only dogfood can still use `store_id` to select the JSON fixture store:\n\n```bash\nrunx skill data-store append_event \\\n  -i data_source_ref=local:\u002F\u002Frunx-data-store\u002Fdev-board \\\n  -i store_id=dev-board \\\n  -i resource=board_events \\\n  -i aggregate_id=posting-123 \\\n  --input-json expected_version=0 \\\n  -i idempotency_key=posting-123:create:v1 \\\n  --input-json event='{\"type\":\"posting.created\",\"payload\":{\"title\":\"fixture proof\"}}' \\\n  --json\n```\n\nProduction graph shape is the same at the skill boundary:\n\n```bash\nrunx skill data-store append_event \\\n  -i data_source_ref=tenant:\u002F\u002Facme\u002Fboard \\\n  -i resource=board_events \\\n  -i aggregate_id=posting-123 \\\n  --input-json expected_version=2 \\\n  -i idempotency_key=posting-123:claim:agent-9 \\\n  --input-json event='{\"type\":\"posting.claimed\",\"payload\":{\"actor\":\"agent-9\"}}' \\\n  --json\n```\n\nThe second command only works once `tenant:\u002F\u002Facme\u002Fboard` is bound to an\ninstalled provider adapter. That binding is operator configuration and may name a\ncredential profile or hosted grant; it must not carry raw secrets.\n\nProject-specific SQLite uses the same command shape after binding the source:\n\n```json\n{\n  \"data_sources\": {\n    \"tenant:\u002F\u002Facme\u002Fboard\": {\n      \"adapter\": \"data.sqlite\",\n      \"database_path\": \".runx\u002Fdata\u002Facme-board.sqlite\",\n      \"resources\": {\n        \"board_events\": {\n          \"kind\": \"event_stream\",\n          \"partition_key\": \"aggregate_id\"\n        }\n      }\n    }\n  }\n}\n```\n\nPass that document through `RUNX_DATA_SOURCES` or `.runx\u002Fdata-sources.json`.\n\nRedis uses the same skill and graph inputs. Only the binding changes:\n\n```json\n{\n  \"data_sources\": {\n    \"tenant:\u002F\u002Facme\u002Fboard\": {\n      \"adapter\": \"data.redis\",\n      \"endpoint\": \"redis:\u002F\u002F127.0.0.1:6379\u002F0\",\n      \"key_prefix\": \"runx:{acme-board}\",\n      \"resources\": {\n        \"board_events\": {\n          \"kind\": \"event_stream\",\n          \"partition_key\": \"aggregate_id\"\n        }\n      }\n    }\n  }\n}\n```\n\nThe Redis endpoint must not embed credentials. Use local unauthenticated Redis\nfor OSS dogfood, or put production secrets behind a runx credential profile or\nhosted grant. For Redis Cluster, the binding's `key_prefix` must contain one\nsafe hash tag, such as `{acme-board}`, so the stream, idempotency, and head keys\ntouched by an append share one slot and update atomically. Stream-head pages use\nstable keyset cursors rather than mutable offsets. Durable events and\ndispositions must not receive TTLs; production Redis should enable persistence\nand use a non-evicting policy.\n","# Data Store\n\nOperate a data source through a governed adapter contract. This skill gives an\nagent enough context to read, append, or project state without learning provider\nsecrets, inventing SQL, or depending on one storage backend.\n\nThe storage backend can be Postgres, SQLite, D1, Redis, DynamoDB, S3, a ledger,\nor a product API. The runx boundary is the same: a declared data source exposes\ntyped operations; the graph supplies bounded params; the adapter executes the\noperation; the receipt records the resource, authority, idempotency, version,\ndigest, and redaction evidence.\n\n## Adapter selection\n\nThe operator chooses a data source at run time. The skill receives\n`data_source_ref` and operation inputs; project or hosted configuration binds\nthat ref to the concrete adapter. A local development ref might be\n`local:\u002F\u002Frunx-data-store\u002Fdev-board`. A production ref might be\n`tenant:\u002F\u002Facme\u002Fboard` bound to `data.postgres`, `data.d1`, `data.redis`, or a\nproduct-owned HTTP adapter.\n\nDo not put provider logic in the domain skill. Messageboard, CRM, support, and\nbusiness-ops skills should ask for durable facts to be read or written; the data\nsource binding decides whether those facts live in local JSON, SQL, Redis, D1,\nobject storage, or a product API. Switching providers is a binding change, not a\nrewrite of the skill.\n\nThe bundled OSS profile calls `data.source`. Unbound `local:\u002F\u002F...` refs default\nto durable local SQLite under `.runx\u002Fdata\u002Flocal-sources\u002F`, with one source-scoped\ndatabase file per logical ref, so stateful skills can be dogfooded without\nstanding up hosted infrastructure. Pass `store_id` only when a fixture\nintentionally wants the deterministic `data.local` JSON store. The graph inputs\nstay the same when a project later binds the source to Postgres, Redis, D1,\nobject storage, or a product API.\n\nAdapter preference is operator configuration, not model choice. To choose Redis,\nSQLite, or a hosted provider, bind the same `data_source_ref` through\n`RUNX_DATA_SOURCES` or `.runx\u002Fdata-sources.json`; do not add provider branches to\nthe domain skill.\n\n## What this skill does\n\n- Reads data through named queries or read operations declared by a data-source\n  adapter.\n- Appends state transitions with idempotency keys and expected versions.\n- Reads projections, event streams, or bounded latest-stream-head pages so\n  loops can resume from explicit state without exporting full history.\n- Produces receipt-bound evidence for data source, resource, operation, params,\n  row\u002Fevent limits, versions, and output digests.\n- Keeps product semantics outside the data layer. Messageboards, CRMs, billing\n  ledgers, and support desks define their own events and reducers.\n- Ships a fixture adapter (`data.local`), durable local SQLite adapter\n  (`data.sqlite`), and Redis adapter (`data.redis`) behind the same operation\n  envelope.\n\n## When to use this skill\n\n- A graph needs durable state between turns, such as queue position, board\n  state, sync cursor, review status, or approval inbox state.\n- A skill must query a bounded slice of product data before deciding the next\n  action.\n- A workflow needs to append an auditable event or effect transition with\n  optimistic concurrency.\n- An operator wants one provider-agnostic shape that can later move from local\n  JSON or SQLite to Postgres, Redis, D1, Supabase, Turso, DynamoDB, or another\n  store.\n\n## When not to use this skill\n\n- To let a model write arbitrary SQL, Redis commands, or database migrations.\n- To export broad data sets, secrets, raw PII, or unrestricted tables.\n- To hide product decisions in storage code. Domain skills still own state\n  machines, acceptance criteria, and business rules.\n- To treat a projection as independent truth when the event stream or receipt\n  chain is available and required for review.\n- To bypass payment, send, deploy, moderation, or human approval gates.\n\n## Procedure\n\n1. Identify the domain skill and transition first. The data store is a carrier,\n   not the policy owner.\n2. Select the logical data source. Use `data_source_ref` to name the project or\n   tenant source; let the project binding choose the adapter. Do not put raw\n   database URLs, provider credentials, or SQL in the skill input.\n3. Select a declared operation: named read query, append event, read events,\n   read projection, or list stream heads. Do not synthesize raw provider\n   commands.\n4. Check authority. Reads need the narrow resource\u002Fquery scope; writes need the\n   transition scope, idempotency key, and expected version unless the operation\n   is explicitly append-only without concurrency.\n5. Bind typed params. Enforce row\u002Fevent limits, tenant\u002Fpartition keys, and\n   redaction rules before the adapter runs.\n6. For writes, use optimistic concurrency and idempotency. A retry with the same\n   idempotency key and same payload returns the existing effect; a different\n   payload under the same key is a conflict.\n7. Return the operation result with resource refs, version movement, digests,\n   redaction notes, and stop conditions. Receipts should link this data effect\n   to the domain transition that caused it.\n\n## Edge cases and stop conditions\n\n- `needs_source`: the data source, resource, query name, tenant key, or schema\n  summary is missing.\n- `needs_input`: required operation params are incomplete, malformed, or not\n  specific enough to bind a declared data-source operation.\n- `needs_authority`: the caller lacks the declared read\u002Fwrite scope or provider\n  grant.\n- `needs_version`: a mutating operation lacks `expected_version` where the data\n  source requires optimistic concurrency.\n- `conflict`: the current version differs from `expected_version`, or an\n  idempotency key is reused with different content.\n- `too_broad`: the requested read lacks partition filters, exceeds limits, or\n  asks for raw export.\n- `redaction_required`: the operation would return secrets, private PII, or\n  fields outside the declared projection.\n- `provider_unavailable`: the adapter cannot reach the data source, times out,\n  or cannot prove whether a write committed.\n\n## Output schema\n\nAll runners return `runx.data.operation_result.v1`:\n\n```json\n{\n  \"schema\": \"runx.data.operation_result.v1\",\n  \"data_source_ref\": \"local:\u002F\u002Fexample\",\n  \"provider\": \"local-json-event-store\",\n  \"operation\": \"append_event\",\n  \"resource\": \"board_events\",\n  \"aggregate_id\": \"posting-123\",\n  \"status\": \"committed\",\n  \"before_version\": 0,\n  \"after_version\": 1,\n  \"idempotency_key\": \"posting-123:create\",\n  \"event_ref\": \"board_events:posting-123:1\",\n  \"result_digest\": \"sha256:...\",\n  \"projection_digest\": \"sha256:...\",\n  \"rows\": [],\n  \"events\": [],\n  \"redactions\": [],\n  \"stop_conditions\": []\n}\n```\n\nProvider adapters may add provider evidence under `provider_evidence`, but they\nmust not expose credentials or raw secret material.\n\nFor event streams, adapters derive a readable `event_type` in this order:\nexplicit `event.type`, explicit `event.event_type`, then\n`event.effect_family + \".\" + event.operation`. Domain skills that emit the\ngeneric `runx.effect.transition.v1` packet should include `effect_family` and\n`operation` on every event so readback projections say `messageboard.accept`,\n`business_ops.route`, or another meaningful transition instead of `data.event`.\n\n## Worked example\n\nA messageboard skill decides that `posting.claimed` is allowed. It emits a\ndomain transition packet. The graph then calls `data-store.append_event` with\nresource `board_events`, aggregate id `posting-123`, expected version `2`, and\nidempotency key `posting-123:claim:agent-9`. The data adapter appends the event\nonly if the stream is still at version `2`. The receipt proves the decision,\nthe data operation, and the new version. A later loop turn calls\n`data-store.read_events` or `read_projection` to resume from the explicit board\nstate.\n\n## Inputs\n\n- `data_source_ref` (required): stable logical ref for the data source. The\n  project or hosted binding maps this ref to the concrete adapter and provider\n  profile.\n- `resource` (required): declared resource, stream, table, keyspace, or\n  projection name.\n- `operation` (required for tool-level use): `append_event`, `read_events`,\n  `read_projection`, or `list_stream_heads`.\n- `aggregate_id` (required for event operations): stream or partition key.\n- `event` (required for `append_event`): domain event or transition packet.\n- `idempotency_key` (required for writes): stable retry key.\n- `expected_version` (required when the source enforces concurrency): current\n  stream\u002Fresource version expected by the caller.\n- `limit` (optional): maximum rows or events to return.\n- `after_version` (optional for `read_events`): return an ascending page whose\n  event versions are strictly greater than this value. Omit it to retain the\n  existing latest-tail read. Compare the last returned event version with\n  `after_version` in the result envelope to know whether another page remains.\n- `event_types` (optional for `list_stream_heads`): at most 20 exact latest\n  event types. No pattern or arbitrary field queries are accepted.\n- `cursor` (optional for `list_stream_heads`): opaque cursor returned by the\n  previous page. Limits are capped at 100.\n- `store_id` (local fixture adapter only): deterministic local store id that\n  opts into the bundled `data.local` proof adapter. Omit it for durable local\n  SQLite. Production adapters should ignore it.\n\n## Invocation examples\n\nDurable local dogfood with the bundled default:\n\n```bash\nrunx skill data-store append_event \\\n  -i data_source_ref=local:\u002F\u002Frunx-data-store\u002Fdev-board \\\n  -i resource=board_events \\\n  -i aggregate_id=posting-123 \\\n  --input-json expected_version=0 \\\n  -i idempotency_key=posting-123:create:v1 \\\n  --input-json event='{\"type\":\"posting.created\",\"payload\":{\"title\":\"verify a receipt link\"}}' \\\n  --json\n```\n\nFixture-only dogfood can still use `store_id` to select the JSON fixture store:\n\n```bash\nrunx skill data-store append_event \\\n  -i data_source_ref=local:\u002F\u002Frunx-data-store\u002Fdev-board \\\n  -i store_id=dev-board \\\n  -i resource=board_events \\\n  -i aggregate_id=posting-123 \\\n  --input-json expected_version=0 \\\n  -i idempotency_key=posting-123:create:v1 \\\n  --input-json event='{\"type\":\"posting.created\",\"payload\":{\"title\":\"fixture proof\"}}' \\\n  --json\n```\n\nProduction graph shape is the same at the skill boundary:\n\n```bash\nrunx skill data-store append_event \\\n  -i data_source_ref=tenant:\u002F\u002Facme\u002Fboard \\\n  -i resource=board_events \\\n  -i aggregate_id=posting-123 \\\n  --input-json expected_version=2 \\\n  -i idempotency_key=posting-123:claim:agent-9 \\\n  --input-json event='{\"type\":\"posting.claimed\",\"payload\":{\"actor\":\"agent-9\"}}' \\\n  --json\n```\n\nThe second command only works once `tenant:\u002F\u002Facme\u002Fboard` is bound to an\ninstalled provider adapter. That binding is operator configuration and may name a\ncredential profile or hosted grant; it must not carry raw secrets.\n\nProject-specific SQLite uses the same command shape after binding the source:\n\n```json\n{\n  \"data_sources\": {\n    \"tenant:\u002F\u002Facme\u002Fboard\": {\n      \"adapter\": \"data.sqlite\",\n      \"database_path\": \".runx\u002Fdata\u002Facme-board.sqlite\",\n      \"resources\": {\n        \"board_events\": {\n          \"kind\": \"event_stream\",\n          \"partition_key\": \"aggregate_id\"\n        }\n      }\n    }\n  }\n}\n```\n\nPass that document through `RUNX_DATA_SOURCES` or `.runx\u002Fdata-sources.json`.\n\nRedis uses the same skill and graph inputs. Only the binding changes:\n\n```json\n{\n  \"data_sources\": {\n    \"tenant:\u002F\u002Facme\u002Fboard\": {\n      \"adapter\": \"data.redis\",\n      \"endpoint\": \"redis:\u002F\u002F127.0.0.1:6379\u002F0\",\n      \"key_prefix\": \"runx:{acme-board}\",\n      \"resources\": {\n        \"board_events\": {\n          \"kind\": \"event_stream\",\n          \"partition_key\": \"aggregate_id\"\n        }\n      }\n    }\n  }\n}\n```\n\nThe Redis endpoint must not embed credentials. Use local unauthenticated Redis\nfor OSS dogfood, or put production secrets behind a runx credential profile or\nhosted grant. For Redis Cluster, the binding's `key_prefix` must contain one\nsafe hash tag, such as `{acme-board}`, so the stream, idempotency, and head keys\ntouched by an append share one slot and update atomically. Stream-head pages use\nstable keyset cursors rather than mutable offsets. Durable events and\ndispositions must not receive TTLs; production Redis should enable persistence\nand use a non-evicting policy.\n",{"name":5,"runx":24,"description":7},{"category":25},"data","main","fe90232960df6d9653e72e8d024a9d859e1ae524","skills\u002Fdata-store","https:\u002F\u002Fgithub.com\u002FZachDreamZ\u002Frunx","https:\u002F\u002Frunx.ai","https:\u002F\u002Fgithub.com\u002FZachDreamZ","https:\u002F\u002Fagentapigw.dpdns.org","Building Agent API Gateway — structured web data for AI agents. Consistent until I make it.","2026-07-17 04:21:10",{"source":8,"skillName":5,"commands":36,"prompt":46,"zipAvailable":47},[37,40,43],{"tool":38,"command":39},"npx","npx skills add ZachDreamZ\u002Frunx --skill data-store",{"tool":41,"command":42},"bunx","bunx skills add ZachDreamZ\u002Frunx --skill data-store",{"tool":44,"command":45},"pnpm","pnpm dlx skills add ZachDreamZ\u002Frunx --skill data-store","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FZachDreamZ\u002Frunx\u002Fblob\u002Fmain\u002Fskills\u002Fdata-store\u002FSKILL.md",true,[49,55,59,63,67,71,75,79,83,87,93,97,101,105,109,113,117],{"path":50,"resourceType":51,"resourceTypeDesc":52,"sizeBytes":53,"sourceHtmlUrl":54},"fixtures\u002Fappend-local-event.yaml","REFERENCE","参考文档",588,"https:\u002F\u002Fgithub.com\u002FZachDreamZ\u002Frunx\u002Fblob\u002Fmain\u002Fskills\u002Fdata-store\u002Ffixtures\u002Fappend-local-event.yaml",{"path":56,"resourceType":51,"resourceTypeDesc":52,"sizeBytes":57,"sourceHtmlUrl":58},"fixtures\u002Fappend-read-default-sqlite-event.yaml",780,"https:\u002F\u002Fgithub.com\u002FZachDreamZ\u002Frunx\u002Fblob\u002Fmain\u002Fskills\u002Fdata-store\u002Ffixtures\u002Fappend-read-default-sqlite-event.yaml",{"path":60,"resourceType":51,"resourceTypeDesc":52,"sizeBytes":61,"sourceHtmlUrl":62},"fixtures\u002Fappend-read-local-event.yaml",727,"https:\u002F\u002Fgithub.com\u002FZachDreamZ\u002Frunx\u002Fblob\u002Fmain\u002Fskills\u002Fdata-store\u002Ffixtures\u002Fappend-read-local-event.yaml",{"path":64,"resourceType":51,"resourceTypeDesc":52,"sizeBytes":65,"sourceHtmlUrl":66},"fixtures\u002Fappend-read-sqlite-event.yaml",991,"https:\u002F\u002Fgithub.com\u002FZachDreamZ\u002Frunx\u002Fblob\u002Fmain\u002Fskills\u002Fdata-store\u002Ffixtures\u002Fappend-read-sqlite-event.yaml",{"path":68,"resourceType":51,"resourceTypeDesc":52,"sizeBytes":69,"sourceHtmlUrl":70},"fixtures\u002Fappend-version-conflict.yaml",619,"https:\u002F\u002Fgithub.com\u002FZachDreamZ\u002Frunx\u002Fblob\u002Fmain\u002Fskills\u002Fdata-store\u002Ffixtures\u002Fappend-version-conflict.yaml",{"path":72,"resourceType":51,"resourceTypeDesc":52,"sizeBytes":73,"sourceHtmlUrl":74},"fixtures\u002Flist-stream-heads-empty.yaml",512,"https:\u002F\u002Fgithub.com\u002FZachDreamZ\u002Frunx\u002Fblob\u002Fmain\u002Fskills\u002Fdata-store\u002Ffixtures\u002Flist-stream-heads-empty.yaml",{"path":76,"resourceType":51,"resourceTypeDesc":52,"sizeBytes":77,"sourceHtmlUrl":78},"fixtures\u002Fread-local-events.yaml",471,"https:\u002F\u002Fgithub.com\u002FZachDreamZ\u002Frunx\u002Fblob\u002Fmain\u002Fskills\u002Fdata-store\u002Ffixtures\u002Fread-local-events.yaml",{"path":80,"resourceType":51,"resourceTypeDesc":52,"sizeBytes":81,"sourceHtmlUrl":82},"fixtures\u002Fread-local-projection.yaml",483,"https:\u002F\u002Fgithub.com\u002FZachDreamZ\u002Frunx\u002Fblob\u002Fmain\u002Fskills\u002Fdata-store\u002Ffixtures\u002Fread-local-projection.yaml",{"path":84,"resourceType":51,"resourceTypeDesc":52,"sizeBytes":85,"sourceHtmlUrl":86},"tools\u002Fdata\u002Flocal\u002Fmanifest.json",2731,"https:\u002F\u002Fgithub.com\u002FZachDreamZ\u002Frunx\u002Fblob\u002Fmain\u002Fskills\u002Fdata-store\u002Ftools\u002Fdata\u002Flocal\u002Fmanifest.json",{"path":88,"resourceType":89,"resourceTypeDesc":90,"sizeBytes":91,"sourceHtmlUrl":92},"tools\u002Fdata\u002Flocal\u002Frun.mjs","SCRIPT","脚本",14683,"https:\u002F\u002Fgithub.com\u002FZachDreamZ\u002Frunx\u002Fblob\u002Fmain\u002Fskills\u002Fdata-store\u002Ftools\u002Fdata\u002Flocal\u002Frun.mjs",{"path":94,"resourceType":51,"resourceTypeDesc":52,"sizeBytes":95,"sourceHtmlUrl":96},"tools\u002Fdata\u002Fredis\u002Fmanifest.json",3145,"https:\u002F\u002Fgithub.com\u002FZachDreamZ\u002Frunx\u002Fblob\u002Fmain\u002Fskills\u002Fdata-store\u002Ftools\u002Fdata\u002Fredis\u002Fmanifest.json",{"path":98,"resourceType":51,"resourceTypeDesc":52,"sizeBytes":99,"sourceHtmlUrl":100},"tools\u002Fdata\u002Fredis\u002FREADME.md",1278,"https:\u002F\u002Fgithub.com\u002FZachDreamZ\u002Frunx\u002Fblob\u002Fmain\u002Fskills\u002Fdata-store\u002Ftools\u002Fdata\u002Fredis\u002FREADME.md",{"path":102,"resourceType":89,"resourceTypeDesc":90,"sizeBytes":103,"sourceHtmlUrl":104},"tools\u002Fdata\u002Fredis\u002Frun.mjs",20217,"https:\u002F\u002Fgithub.com\u002FZachDreamZ\u002Frunx\u002Fblob\u002Fmain\u002Fskills\u002Fdata-store\u002Ftools\u002Fdata\u002Fredis\u002Frun.mjs",{"path":106,"resourceType":51,"resourceTypeDesc":52,"sizeBytes":107,"sourceHtmlUrl":108},"tools\u002Fdata\u002Fsqlite\u002Fmanifest.json",2907,"https:\u002F\u002Fgithub.com\u002FZachDreamZ\u002Frunx\u002Fblob\u002Fmain\u002Fskills\u002Fdata-store\u002Ftools\u002Fdata\u002Fsqlite\u002Fmanifest.json",{"path":110,"resourceType":51,"resourceTypeDesc":52,"sizeBytes":111,"sourceHtmlUrl":112},"tools\u002Fdata\u002Fsqlite\u002FREADME.md",2133,"https:\u002F\u002Fgithub.com\u002FZachDreamZ\u002Frunx\u002Fblob\u002Fmain\u002Fskills\u002Fdata-store\u002Ftools\u002Fdata\u002Fsqlite\u002FREADME.md",{"path":114,"resourceType":89,"resourceTypeDesc":90,"sizeBytes":115,"sourceHtmlUrl":116},"tools\u002Fdata\u002Fsqlite\u002Frun.mjs",24826,"https:\u002F\u002Fgithub.com\u002FZachDreamZ\u002Frunx\u002Fblob\u002Fmain\u002Fskills\u002Fdata-store\u002Ftools\u002Fdata\u002Fsqlite\u002Frun.mjs",{"path":118,"resourceType":51,"resourceTypeDesc":52,"sizeBytes":119,"sourceHtmlUrl":120},"X.yaml",8460,"https:\u002F\u002Fgithub.com\u002FZachDreamZ\u002Frunx\u002Fblob\u002Fmain\u002Fskills\u002Fdata-store\u002FX.yaml",[122,129,136,143,150,157],{"skillNo":123,"name":124,"nameZh":6,"description":125,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":126,"categories":127,"keywords":128,"publishedAt":19},"01M79A3PCWPN0YG6XG3X2YWGY4","zapier-handoff","Validate a runx execution context and hand off a governed payload to a Zapier Catch Hook with scoped auth, idempotency, and receipt expectations.","https:\u002F\u002Fgithub.com\u002FZachDreamZ\u002Frunx\u002Fblob\u002Fmain\u002Fskills\u002Fzapier-handoff\u002FSKILL.md",[],[],{"skillNo":130,"name":131,"nameZh":6,"description":132,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":133,"categories":134,"keywords":135,"publishedAt":19},"01M79A3JPAJVDGX5AGXKHB6M58","x402-pay","Execute a governed x402 payment by delegating to the canonical spend flow with the x402 runtime path selected.","https:\u002F\u002Fgithub.com\u002FZachDreamZ\u002Frunx\u002Fblob\u002Fmain\u002Fskills\u002Fx402-pay\u002FSKILL.md",[],[],{"skillNo":137,"name":138,"nameZh":6,"description":139,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":140,"categories":141,"keywords":142,"publishedAt":19},"01M79A3H7HHAQ5B4QCA0S2PPA3","work-plan","Decompose a build objective into governed runx execution steps.","https:\u002F\u002Fgithub.com\u002FZachDreamZ\u002Frunx\u002Fblob\u002Fmain\u002Fskills\u002Fwork-plan\u002FSKILL.md",[],[],{"skillNo":144,"name":145,"nameZh":6,"description":146,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":147,"categories":148,"keywords":149,"publishedAt":19},"01M79A3ESQDZ6AQ3QASKYGYTNB","web-fetch","Fetch and extract one web source within an explicit allowlist, returning the content by digest with full provenance.","https:\u002F\u002Fgithub.com\u002FZachDreamZ\u002Frunx\u002Fblob\u002Fmain\u002Fskills\u002Fweb-fetch\u002FSKILL.md",[],[],{"skillNo":151,"name":152,"nameZh":6,"description":153,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":154,"categories":155,"keywords":156,"publishedAt":19},"01M79A3C9NE99P24Q2QGG5236D","weather-forecast","Normalize provider weather evidence into an action-safe forecast packet with provenance, uncertainty, and stop conditions for downstream agents.","https:\u002F\u002Fgithub.com\u002FZachDreamZ\u002Frunx\u002Fblob\u002Fmain\u002Fskills\u002Fweather-forecast\u002FSKILL.md",[],[],{"skillNo":158,"name":159,"nameZh":6,"description":160,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":161,"categories":162,"keywords":163,"publishedAt":19},"01M79A38RWEA5NHMYBQZPCVPQP","vuln-triage","Triage dependency or ecosystem vulnerability risk into remediation and advisory packets, grounding on verified CVE evidence when it is supplied.","https:\u002F\u002Fgithub.com\u002FZachDreamZ\u002Frunx\u002Fblob\u002Fmain\u002Fskills\u002Fvuln-triage\u002FSKILL.md",[],[]]