refactor(cli): rename 'obol dataset' to 'obol sell data'

Moves the seller-side dataset command group under 'obol sell' as 'data' (with 'dataset' kept as an alias). 'obol buy dataset' is unchanged. Updates help/error strings, the monetize-dataset guide, the dataset-anonymize skill, and the hf-surface smoke flow.

Author: bussyjd

cmd/obol/dataset.go

@@ -1,16 +1,16 @@
 package main
 
-// obol dataset — owner side of a versioned, membership-gated dataset offer.
+// obol sell data — owner side of a versioned, membership-gated dataset offer.
 //
-//	obol dataset from <bundle-dir> --name <id>   ingest a bundle as a new
-//	                                             signed version (creates v1).
-//	obol dataset version <id> --bundle <dir>     append the next signed version.
-//	obol dataset publish <id>                    host the artifact server on
-//	                                             this machine + a Cloudflare
-//	                                             tunnel; gate every byte.
-//	obol dataset approve <user-code>             admit a worker (membership).
-//	obol dataset verify <id>                     walk the signed version chain.
-//	obol dataset status <id>                     versions + members.
+//	obol sell data from <bundle-dir> --name <id>   ingest a bundle as a new
+//	                                               signed version (creates v1).
+//	obol sell data version <id> --bundle <dir>     append the next signed version.
+//	obol sell data publish <id>                    host the artifact server on
+//	                                               this machine + a Cloudflare
+//	                                               tunnel; gate every byte.
+//	obol sell data approve <user-code>             admit a worker (membership).
+//	obol sell data verify <id>                     walk the signed version chain.
+//	obol sell data status <id>                     versions + members.
 //
 // The artifact server is the host gateway (same spirit as `obol sell
 // inference` / `obol research publish`): it runs on the owner's machine, never
@@ -47,10 +47,11 @@ type datasetState struct {
 	OwnerToken string `json:"owner_token"`
 }
 
-func datasetCommand(cfg *config.Config) *cli.Command {
+func sellDataCommand(cfg *config.Config) *cli.Command {
 	return &cli.Command{
-		Name:  "dataset",
-		Usage: "Publish and sell versioned, membership-gated datasets",
+		Name:    "data",
+		Aliases: []string{"dataset"},
+		Usage:   "Publish and sell versioned, membership-gated datasets",
 		Commands: []*cli.Command{
 			datasetFromCommand(cfg),
 			datasetVersionCommand(cfg),
@@ -70,7 +71,7 @@ func datasetFromCommand(cfg *config.Config) *cli.Command {
 		Flags:     []cli.Flag{&cli.StringFlag{Name: "name", Usage: "Dataset id", Required: true}},
 		Action: func(_ context.Context, cmd *cli.Command) error {
 			if cmd.NArg() != 1 {
-				return fmt.Errorf("bundle directory required: obol dataset from <bundle-dir> --name <id>")
+				return fmt.Errorf("bundle directory required: obol sell data from <bundle-dir> --name <id>")
 			}
 			return appendDatasetVersion(cfg, cmd, strings.TrimSpace(cmd.String("name")), cmd.Args().First())
 		},
@@ -85,11 +86,11 @@ func datasetVersionCommand(cfg *config.Config) *cli.Command {
 		Flags:     []cli.Flag{&cli.StringFlag{Name: "bundle", Usage: "New bundle directory", Required: true}},
 		Action: func(_ context.Context, cmd *cli.Command) error {
 			if cmd.NArg() != 1 {
-				return fmt.Errorf("dataset id required: obol dataset version <id> --bundle <dir>")
+				return fmt.Errorf("dataset id required: obol sell data version <id> --bundle <dir>")
 			}
 			id := strings.TrimSpace(cmd.Args().First())
 			if _, err := os.Stat(datasetStorePath(cfg, id)); err != nil {
-				return fmt.Errorf("dataset %q not found — create it with 'obol dataset from'", id)
+				return fmt.Errorf("dataset %q not found — create it with 'obol sell data from'", id)
 			}
 			return appendDatasetVersion(cfg, cmd, id, cmd.String("bundle"))
 		},
@@ -144,7 +145,7 @@ func appendDatasetVersion(cfg *config.Config, cmd *cli.Command, id, bundleDir st
 	u.Infof("File hash:     %s", v.FileHash)
 	u.Infof("Size:          %d bytes", v.Size)
 	u.Infof("Owner:         %s", signer.SignerID())
-	u.Dim("Publish it with:  obol dataset publish " + id)
+	u.Dim("Publish it with:  obol sell data publish " + id)
 	return nil
 }
 
@@ -164,7 +165,7 @@ func datasetPublishCommand(cfg *config.Config) *cli.Command {
 		Action: func(ctx context.Context, cmd *cli.Command) error {
 			u := getUI(cmd)
 			if cmd.NArg() != 1 {
-				return fmt.Errorf("dataset id required: obol dataset publish <id>")
+				return fmt.Errorf("dataset id required: obol sell data publish <id>")
 			}
 			id := strings.TrimSpace(cmd.Args().First())
 
@@ -180,7 +181,7 @@ func datasetPublishCommand(cfg *config.Config) *cli.Command {
 				return err
 			}
 			if len(st.Versions) == 0 {
-				return fmt.Errorf("dataset %q has no versions — run 'obol dataset from' first", id)
+				return fmt.Errorf("dataset %q has no versions — run 'obol sell data from' first", id)
 			}
 			// Never serve a chain we cannot verify against the owner key: a
 			// tampered persisted store must fail closed, not be published.
@@ -302,7 +303,7 @@ func datasetApproveCommand(cfg *config.Config) *cli.Command {
 		Action: func(ctx context.Context, cmd *cli.Command) error {
 			u := getUI(cmd)
 			if cmd.NArg() != 1 {
-				return fmt.Errorf("user code required: obol dataset approve <user-code>")
+				return fmt.Errorf("user code required: obol sell data approve <user-code>")
 			}
 			st, err := loadDatasetState(cfg, cmd.String("dataset"))
 			if err != nil {
@@ -332,7 +333,7 @@ func datasetVerifyCommand(cfg *config.Config) *cli.Command {
 		Action: func(_ context.Context, cmd *cli.Command) error {
 			u := getUI(cmd)
 			if cmd.NArg() != 1 {
-				return fmt.Errorf("dataset id required: obol dataset verify <id>")
+				return fmt.Errorf("dataset id required: obol sell data verify <id>")
 			}
 			id := strings.TrimSpace(cmd.Args().First())
 			key, err := dataset.LoadOrCreateKey(datasetKeyPath(cfg, id))
@@ -363,7 +364,7 @@ func datasetStatusCommand(cfg *config.Config) *cli.Command {
 		Action: func(ctx context.Context, cmd *cli.Command) error {
 			u := getUI(cmd)
 			if cmd.NArg() != 1 {
-				return fmt.Errorf("dataset id required: obol dataset status <id>")
+				return fmt.Errorf("dataset id required: obol sell data status <id>")
 			}
 			id := strings.TrimSpace(cmd.Args().First())
 			st, err := dataset.NewStore(datasetStorePath(cfg, id)).Load()

cmd/obol/main.go

@@ -330,7 +330,6 @@ GLOBAL OPTIONS:{{template "visibleFlagTemplate" .}}{{end}}
 			bountyCommand(cfg),
 			smokeCommand(cfg),
 			researchCommand(cfg),
-			datasetCommand(cfg),
 			modelCommand(cfg),
 			{
 				Name:  "app",

cmd/obol/sell.go

@@ -67,6 +67,7 @@ func sellCommand(cfg *config.Config) *cli.Command {
 			sellIdentityCommand(cfg),
 			sellInfoCommand(cfg),
 			sellResumeCommand(cfg),
+			sellDataCommand(cfg),
 		},
 	}
 }

docs/guides/monetize-dataset.md

@@ -35,24 +35,24 @@ the `dataset-anonymize` skill.
 ## 2. Record a signed version
 
 ```bash
-obol dataset from my-bundle --name pi-sessions
+obol sell data from my-bundle --name pi-sessions
 ```
 
 This reads the bundle, computes the artifact's whole-file SHA-256, and appends
 a **signed** `DatasetVersion` (v1) to the dataset's version log — chained to
 its predecessor, signed by your owner key (the address buyers pin). Append a new
-snapshot later with `obol dataset version pi-sessions --bundle my-bundle-v2`.
+snapshot later with `obol sell data version pi-sessions --bundle my-bundle-v2`.
 
 Walk the chain offline at any time:
 
 ```bash
-obol dataset verify pi-sessions     # rejects any reorder/tamper/middle-removal
+obol sell data verify pi-sessions     # rejects any reorder/tamper/middle-removal
 ```
 
 ## 3. Publish (host + tunnel + gate)
 
 ```bash
-obol dataset publish pi-sessions --membership invite
+obol sell data publish pi-sessions --membership invite
 ```
 
 Starts the artifact server on your machine and a Cloudflare tunnel. **Bytes
@@ -64,7 +64,7 @@ alike.
 Two ways a caller holds a member token:
 
 - **Pre-approved worker** — joins via device-auth; you run
-  `obol dataset approve <user-code>`. Gets full (head) access.
+  `obol sell data approve <user-code>`. Gets full (head) access.
 - **Anonymous market buyer** — pays the priced offer; the edge x402 verifier
   proves the settled payment, and the server mints a token scoped to exactly
   the version paid for (`/join/paid`). Payment *is* the approval; the dataset

flows/hf-surface-smoke.sh

@@ -77,16 +77,16 @@ fi
 HASH=$(shasum -a 256 "$BUNDLE/sft.jsonl" | awk '{print $1}')
 printf '{"hash":"%s","files":["sft.jsonl"]}\n' "$HASH" > "$BUNDLE/manifest.json"
 
-if "$OBOL_BIN" dataset from "$BUNDLE" --name "$DS_ID" >/dev/null 2>&1 \
-   && "$OBOL_BIN" dataset verify "$DS_ID" 2>&1 | grep -q 'Chain valid'; then
+if "$OBOL_BIN" sell data from "$BUNDLE" --name "$DS_ID" >/dev/null 2>&1 \
+   && "$OBOL_BIN" sell data verify "$DS_ID" 2>&1 | grep -q 'Chain valid'; then
   pass "1b sign + verify — signed version chain valid"
 else
   fail "1b sign+verify" "version not recorded or chain invalid"
 fi
 
 MANIFEST_HASH=$(python3 -c "import json;print(json.load(open('$OBOL_CONFIG_DIR/dataset-serve/$DS_ID.store.json'))['versions'][0]['manifestHash'])")
 
-"$OBOL_BIN" dataset publish "$DS_ID" --membership open --port "$DS_PORT" --no-tunnel >/dev/null 2>&1 &
+"$OBOL_BIN" sell data publish "$DS_ID" --membership open --port "$DS_PORT" --no-tunnel >/dev/null 2>&1 &
 curl -sf --retry 25 --retry-connrefused --retry-delay 1 "http://127.0.0.1:$DS_PORT/healthz" >/dev/null
 OWNER=$(python3 -c "import json;print(json.load(open('$OBOL_CONFIG_DIR/dataset-serve/$DS_ID.state.json'))['owner_token'])" 2>/dev/null)
 

internal/embed/skills/dataset-anonymize/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: dataset-anonymize
-description: Anonymize a dataset's JSONL (PII detection + masking) before publishing or selling it with `obol dataset`. Pluggable detector — built-in regex redactor by default, or a BYO Hugging Face token-classification model.
+description: Anonymize a dataset's JSONL (PII detection + masking) before publishing or selling it with `obol sell data`. Pluggable detector — built-in regex redactor by default, or a BYO Hugging Face token-classification model.
 ---
 
 # dataset-anonymize
@@ -38,8 +38,8 @@ export OBOL_ANONYMIZER_MODEL="<org>/<pii-token-classification-model>"
 python3 scripts/anonymize.py input.jsonl anonymized.jsonl --report
 
 # Then ingest the anonymized bundle and publish it:
-obol dataset from <bundle-dir-with-anonymized.jsonl> --name my-dataset
-obol dataset publish my-dataset
+obol sell data from <bundle-dir-with-anonymized.jsonl> --name my-dataset
+obol sell data publish my-dataset
 ```
 
 Each input line is a JSON object; the script masks string values under