docs(skills): Paradev naming in skill text; keep superset skill identifiers

Merged
KI
10h ago
+19-2

What changed

  • .agents/skills/superset-settings/SKILL.md: the description now says "Paradev (formerly Superset)" and names the paradev alias. The body says paradev is the same binary as superset.
  • docs/agent-tooling.md: a new section, "Skill names after the Paradev rename". It lists the skill and provisioning names that keep superset and gives the reason.

Inventory (.agents/skills/**, docs/agent-tooling.md, managed-skill provisioning)

Display text that still said Superset: only the superset-settings description. That is fixed here. Every other match is an identifier that something looks up by name:

WhereStringKind
.agents/skills/superset-settingsskill name superset-settingsidentifier (/superset-settings, $superset-settings)
most skills, docs/agent-tooling.mdsuperset <command>CLI binary (paradev alias already exists, T-cli)
skills, docsSUPERSET_HOME_DIR, SUPERSET_WORKSPACE_ID, SUPERSET_API_KEY, …env vars (T-env)
incident-triage, dev-doctor*.superset.sh, Sentry org superset-shdomains / external IDs
mobile-* skillsSuperset.xcworkspace, scheme Superset, sh.superset.mobile, superset://, /data/local/tmp/superset-mobile-owner, R2 bucket superset-private-devbuild output names / bundle ID / URL scheme / bucket
db-migrations, cdp-verification@superset/*package names
trpc-compat, mobile-demo-recordx-superset-client, x-superset-organization-idHTTP headers
incident-triage, apps/docs/.../skills.mdxsuperset:doctor, superset:<skill>plugin slash namespace (T-plugin)
docs/agent-tooling.md.superset-profile.json, superset-dev (dev-skills sync)file / plugin names
packages/agent-setup/src/managed-skills.tsBUNDLED_SOURCE_NAME = "superset", MANAGED_COMMAND_NAMESPACE = "superset", <!-- superset-managed-skill v1 -->, .superset-managed, disabled keys superset/<skill>provisioning identifiers
.agents/commands/task*.md, create-pr.mdmcp__superset__*, superset-homeMCP server name / GitHub user

The desktop Skills list (SkillsList.tsx) shows plain skill names (10x, automate, …) and already says Paradev. It needs no change.

What stays on the old name and why

No identifier is renamed. Agent CLIs have no skill alias:

  • A second skill dir (for example paradev-settings beside superset-settings, or paradev-<skill> beside superset-<skill>) shows the same skill two times in every agent's list.
  • A rename in place breaks saved prompts and habits that call the old name. For managed skills, a changed BUNDLED_SOURCE_NAME also makes the reaper delete every marker-bearing superset-<skill> dir that was provisioned before. Disabled-skill keys saved as superset/<skill> would also stop matching, so disabled skills would come back.

The task allows a rename only with an old name that keeps resolving and no duplicate. That is not possible here, so these names stay and the new docs section lists them.

How old installs behave

They behave the same as before. No provisioned path, marker or key changes.

Validation

  • bun test src/managed-skills.test.ts in packages/agent-setup: 23 pass (no code change; sanity check only)
  • bunx biome check on the changed files: biome does not process markdown, so it checked nothing
  • No typecheck, check:i18n or check:plugins: no code, UI strings or plugins changed

Summary by cubic

Updates the AGENTS skill text to reflect the Paradev rename and documents which names intentionally keep superset.

  • The superset-settings skill description now says "Paradev (formerly Superset)" and notes that paradev is the same binary as superset.
  • docs/agent-tooling.md lists the identifiers that stay on superset (CLI command, env vars, skill and provisioning names) and explains why: agents have no skill alias, so a second name duplicates the skill and a rename breaks saved prompts and managed-skill provisioning.