Skip to main content

fastskill sync

Writes installed skills into the agent metadata file (Claude Code / compatible layouts). Agent and file path are auto-detected when omitted.
fastskill sync
fastskill sync --yes
fastskill sync --agent claude
fastskill sync --agents-file custom.md
OptionDescription
-y, --yesNon-interactive: include all installed skills
-a, --agent <AGENT>Target agent (optional; auto-detected)
--agents-file <PATH>Override metadata file path

fastskill marketplace

marketplace create

Scans a directory for skills (folders with SKILL.md) and writes marketplace.json.
fastskill marketplace create --path ./skills --name my-marketplace
fastskill marketplace create -p . -o .claude-plugin/marketplace.json --name "My Marketplace"
OptionDescription
-p, --path <DIR>Root to scan (default: .)
-o, --output <FILE>Output path (default under .claude-plugin/ in the scan directory)
--base-url <URL>Base URL for download links
--name <NAME>Required. Marketplace name
--owner-name, --owner-email, --description, --versionOptional metadata

fastskill analyze

Requires a running service context (installed skills and index) like other service commands.

analyze matrix

Pairwise similarity between indexed skills.
fastskill analyze matrix
fastskill analyze matrix --threshold 0.8 --limit 5
fastskill analyze matrix --full --json
OptionDescription
--format <FORMAT>table, json, grid, xml
--jsonShorthand for --format json
--threshold <0..1>Minimum similarity to show (default 0.0)
-l, --limit <N>Similar skills per skill (default 10)
--fullShow all pairs instead of condensed view

analyze cluster

Groups skills by semantic similarity.
fastskill analyze cluster
fastskill analyze cluster -k 8 --min-size 2 --json
OptionDescription
-k, --num-clusters <N>Number of clusters (default 5)
--min-size <N>Hide smaller clusters (default 1)
--format, --jsonOutput selection

analyze duplicates

Finds near-duplicate skills.
fastskill analyze duplicates
fastskill analyze duplicates --threshold 0.92 --severity critical
OptionDescription
--threshold <0..1>Minimum similarity (default 0.88)
--limit <N>Max pairs to show (default 20)
--severity <LEVEL>all, medium, high, or critical
--format, --jsonOutput selection

See also