Add manifest v2 fields: experimental, description, min_cli_version#35
Merged
fjakobs merged 3 commits intodatabricks:mainfrom Mar 26, 2026
Merged
Conversation
Co-authored-by: Isaac
This was referenced Mar 22, 2026
…metadata Co-authored-by: Isaac
arsenyinfo
approved these changes
Mar 23, 2026
Co-authored-by: Isaac
github-merge-queue bot
pushed a commit
to databricks/cli
that referenced
this pull request
Mar 26, 2026
…ude-experimental flags (#4813) ## PR Stack 1. [1/5] State + release discovery + directory rename (#4810) 2. [2/5] Install writes state + interactive agent selection (#4811) 3. [3/5] Update + uninstall + version commands (#4812) 4. **[4/5] List improvements + command restructuring + flags** (this PR) 5. [5/5] Project scope (--project/--global) (#4814) Manifest v2 PR: databricks/databricks-agent-skills#35 **Base**: `simonfaltum/aitools-pr3-lifecycle` (PR 3) ## Why Commands are nested under `skills` which burns namespace for future component types (hooks, etc.). There's no way to install specific skills, target specific agents from the CLI, or see installed status in the list output. ## Changes **New `list` command**: Table output with skill names, versions, installed status, `[experimental]` tags. Sorted alphabetically. Uses `tabwriter` for alignment. **Flat command structure**: `aitools install/update/uninstall/list/version` at the top level. Hidden `skills install` and `skills list` aliases for backward compat. **Flags**: - `--skills` (string, comma-separated) on install, update, uninstall: operate on specific skills - `--skills` (bool) on list, version: show detailed skills view - `--agents` (string, comma-separated) on install: target specific agents, validates against registry, skips interactive prompt - `--include-experimental` on install: include experimental skills **Selective uninstall**: `--skills` on uninstall removes only named skills, preserves state file with remaining. ## Test plan - [x] `install --skills`, `--agents`, `--include-experimental` flags work - [x] `--agents` validates against registry, errors on unknown - [x] `uninstall --skills` selective removal, state preserved - [x] `update --skills` selective update - [x] `list` shows detailed table - [x] `skills install` and `skills list` hidden aliases work (Execute path tested) - [x] `cmd/apps/init.go` integration preserved - [x] All lint checks pass
github-merge-queue bot
pushed a commit
to databricks/cli
that referenced
this pull request
Mar 26, 2026
## PR Stack 1. [1/5] State + release discovery + directory rename (#4810) 2. [2/5] Install writes state + interactive agent selection (#4811) 3. [3/5] Update + uninstall + version commands (#4812) 4. [4/5] List improvements + command restructuring + flags (#4813) 5. **[5/5] Project scope (--project/--global)** (this PR) Manifest v2 PR: databricks/databricks-agent-skills#35 **Base**: `simonfaltum/aitools-pr4-restructure` (PR 4) ## Why Skills are currently global-only. Teams working on the same project can't share a curated set of skills via their repo. Project-scoped installation puts skills alongside the code, so they can be committed to git and shared. ## Changes Before: All skills install to `~/.databricks/aitools/skills/` (global only). Now: - `--project` flag on install, update, uninstall: operates on `<cwd>/.databricks/aitools/skills/` - `--global` flag: explicit global scope (default behavior) - `--global` + `--project` -> error - Interactive scope prompt on install (default: global). Uses `huh.NewSelect` in cmd layer. - Non-interactive defaults to global. - `SupportsProjectScope` field on `Agent` struct. Only Claude Code and Cursor support project scope. - Incompatible agents get a warning, not an error: "Skipped <agent>: does not support project-scoped skills." - Version shows both scopes when both exist. - State includes `scope: "project"` field for project installs. ## Test plan - [x] `--project` installs to cwd-relative directory - [x] `--global` + `--project` -> error - [x] No flag, interactive -> scope prompt shown, default is global - [x] No flag, non-interactive -> global (no prompt) - [x] Incompatible agents warned, not errored - [x] Symlinks only created for capable agents in project scope - [x] Version shows both scopes when both exist - [x] `SupportsProjectScope` set correctly for all 6 agents - [x] All lint checks pass
fjakobs
approved these changes
Mar 26, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Why
The Databricks CLI is adding support for skill filtering by experimental status and CLI version compatibility. The manifest needs new fields to support this.
Changes
Before: Manifest v1 with
version,updated_at, andfilesper skill.Now: Manifest v2 adds three new fields to each skill entry:
experimental(bool): Whether the skill is experimental. All current skills arefalse.description(string): One-line description of the skill.min_cli_version(string): Minimum CLI version required. Empty string means no minimum.The CLI handles both v1 and v2 manifests gracefully. Missing v2 fields use zero-value defaults (non-experimental, no description, no version requirement).
Also updated
scripts/generate_manifest.pyto emit v2 fields with aSKILL_METADATAdict for per-skill configuration.Test plan
python3 scripts/generate_manifest.py validatepasses