CLI to install and manage agent skills
September 12, 2026 · View on GitHub
CLI to install and manage agent skills
agent-skill-manager (asm) is a scriptable CLI built for AI agents and automation — install, search, audit, and organize skills across Claude Code, Codex, Cursor, and 18 more tools. Every command supports --json and --yes for non-interactive use. An optional TUI (asm) is available for local browsing.
Get Started · Browse 6,000+ skills · Full docs
Problems asm solves
| Pain | Without asm | With asm |
|---|---|---|
| Scattered installs | Same skill copied into ~/.claude/skills/, ~/.codex/skills/, ~/.cursor/rules/ — different versions, no single view | One asm list across all 21 providers and scopes |
| No inventory | ls through hidden dirs; no idea what is installed, duplicated, or outdated | asm search, asm inspect, asm stats, asm audit |
| Invisible context cost | Every installed skill's description is resident in the agent's prompt on every message, whether or not it ever fires | asm stats --tokens, asm audit residency |
| Install just to read once | Installing is the only way to get a skill in front of an agent — and it stays resident forever afterwards | asm get <skill> prints the body and installs nothing |
| Risky manual installs | Clone repos, copy folders, hope SKILL.md is valid | asm install validates frontmatter, scans security, pins registry commits |
| Agent-unfriendly output | Human-only copy-paste workflows | Structured JSON via --json; skip prompts with --yes |
| New agent, new chore | Every tool adds another skill directory to track | Add or disable providers in one config file |
At a glance
| Catalog | 6,082 skills from 77 repos — browse online |
| Providers | 21 providers (Agents, Claude Code, Pi, OpenCode, Codex, …) |
| Agent-ready CLI | --json, --yes, --machine on list, search, install, audit, eval |
| Security | Pre-install scan for shell exec, network access, credential exposure |
| License | MIT — no accounts, no telemetry |
How it works
graph LR
A[AI agent / script] --> B["asm command --json"]
B --> C[Discover skill dirs]
B --> D[Install / audit / search]
D --> E{Security scan}
E --> F[Validate SKILL.md]
F --> G[Agent provider paths]
H[GitHub / Registry] --> D
- Install
asmonce (npm or curl). - Run commands from your agent, shell, or CI — no prompts when you pass
--yes. - Parse JSON output to decide what to install, remove, or audit next.
Screenshots
The optional TUI (asm with no args) — filter, inspect, and audit skills across every provider from one dashboard.
Features
| Feature | What you get |
|---|---|
| Cross-provider inventory | asm list --json — every skill, every agent, one response |
| Local skill tags | asm tag add|remove; asm list --tag / asm search --tag (AND) |
| One-command install | asm install github:user/repo or asm install skill-name |
| Agent-parseable output | --json on list, search, inspect, install, audit |
| Duplicate audit | asm audit --yes removes redundant skills non-interactively |
| Semantic overlap audit | asm audit overlap finds different-name skills doing the same job |
| Attention budget | asm stats --tokens shows resident vs body token cost per tool |
| Residency audit | asm audit residency ranks skills to demote out of resident context |
| Reference tier | asm get <skill> delivers a body once, at zero residency |
| Security scan | asm audit security before install |
| Authoring pipeline | asm init, asm link, asm eval, asm publish |
| Bundles | asm bundle install — curated sets in one pass |
| Local library | asm install --library — install once, asm activate per provider |
| Cross-tool linking | Reinstall or symlink when a skill already exists in another tool |
Getting started
Install (npm):
npm install -g agent-skill-manager
List installed skills (JSON for agents):
asm list --json
Install a skill:
asm install github:anthropics/skills --yes
Search installed and catalog skills:
asm search "code review" --json
Alternative install: curl -sSL https://raw.githubusercontent.com/luongnv89/asm/main/install.sh | bash
Node.js 22+ required. Optional TUI: run asm with no arguments.
Common tasks
| Task | Command |
|---|---|
| Machine-readable inventory | asm list --json |
| Skill metadata for agents | asm inspect my-skill --json |
| Non-interactive install | asm install code-review -p claude --yes --json |
| Remove duplicates | asm audit --yes |
| Find same-job near-duplicates | asm audit overlap |
| See your context cost | asm stats --tokens |
| Find skills to demote | asm audit residency |
| Use a skill without residency | asm get code-review |
| Scan before install | asm audit security github:user/repo |
| Scaffold a new skill | asm init my-skill -p claude |
| Live dev via symlink | asm link ./my-skill -p claude |
| Publish to registry | asm publish ./my-skill --yes |
| Install a bundle | asm bundle install frontend-dev --yes |
| Tag an installed skill | asm tag add code-review testing |
| Filter inventory by tags | asm list --tag cli,testing |
Full command reference, flags, and examples: CLI Commands. Catalog UI and bundles: luongnv.com/asm (use Tags / TagFilter, AND match, ?tag=).
Attention budget
Disk is cheap; context is not. Every installed skill's frontmatter
description sits in the agent's system prompt on every message, whether or
not the skill ever fires. The full SKILL.md body only costs anything when the
skill actually fires.
asm reports the two separately — token counts are estimates and always render
with a leading ~:
asm stats --tokens # resident vs body tokens, per tool and per scope
asm stats --tokens --json # same data for scripts and agents
Attention Budget
----------------------------------------
Resident = frontmatter descriptions, paid on every message.
Bodies = full SKILL.md, paid only when a skill fires.
Tool Skills Resident Bodies
Claude Code 74 ~5k tokens ~281k tokens
Codex 10 ~736 tokens ~76k tokens
---------------------------------------------
Total 84 ~5.7k tokens ~357k tokens
By Scope
global 71 ~4.9k tokens ~300k tokens
project 13 ~836 tokens ~57k tokens
Heaviest resident descriptions
claude-api ~1.2k tokens (claude, global)
docx ~233 tokens (claude, global)
pptx ~217 tokens (claude, global)
code-review ~198 tokens (claude, global)
frontend-design ~176 tokens (claude, project)
dataviz ~164 tokens (claude, global)
skill-creator ~151 tokens (claude, global)
xlsx ~143 tokens (claude, global)
pdf ~138 tokens (claude, global)
release-manager ~129 tokens (codex, global)
Median resident cost: ~48 tokens
Run asm audit residency for demotion advice
Residency audit
asm audit residency ranks the installed skills that are not earning their
resident context and pairs each one with a command that works for how that
skill is actually installed. It only reports — nothing is deactivated,
disabled, or removed unless you run one of the suggested commands yourself.
asm audit residency
asm audit residency --json
The advice assumes this demotion ladder:
| Tier | asm mechanism | Residency |
|---|---|---|
| Installed (auto-triggers) | provider directory / asm activate | description resident, always |
| Saved (adapt, no trigger) | asm install --library, asm deactivate | none until asm activate |
| Disabled (kept on disk) | asm disable (reverse with asm enable) | none |
| Reference (read on use) | default asm get — no retained skill copy | none |
Which command a candidate gets depends on how it is installed. asm deactivate
only works on a live symlink into the asm library, so it is suggested only
there; everything else gets asm disable, which is universal and reversible.
The asm deactivate path additionally gets asm get <skill> as a second
destination: deactivating removes the provider symlink but leaves the library
copy, so the body is still one command away. The asm disable path does not,
because disabling renames the canonical SKILL.md — for a library-linked skill
that is the library copy itself — leaving nothing for asm get to read.
Skills provided by plugin marketplaces are counted in the totals but never
listed as candidates — no asm command demotes them.
Two of the signals the report would like to use — trigger collision and
last-used counts — have no data source in asm yet. They are listed as
unavailable rather than silently dropped, so the report still works today.
No network calls, accounts, or telemetry are involved: both reports run entirely against your local filesystem.
Reference tier: use a skill without installing it
Demoting a skill does not mean losing it. asm get <skill> resolves a skill and
writes its SKILL.md body to stdout — no provider directory, no library
entry, no residency. Default get retains no skill copy. With --path, it
instead borrows a full directory on disk until you explicitly clean it up.
asm get code-review # body to stdout
asm get code-review > /tmp/SKILL.md # save it without installing
asm get github:owner/repo:skills/review # any `asm install` shorthand
asm get code-review --json # includes optional dependencies and exact content
That makes asm usable as delivery infrastructure, not only as an installer.
An agent can pull a skill in for a single task and drop it again:
asm get code-review | your-agent --system-prompt-file -
<skill> walks a resolution ladder, and the rung that answered is reported as
the tier so provenance is always visible:
| Order | Tier | Resolves |
|---|---|---|
| 1 | installed | a skill already installed in one of your agents |
| 2 | library | a library skill, including deactivated ones |
| 3 | index | an indexed catalog skill, matched by exact name |
| 4 | registry | the ASM registry — same route as asm install <name> |
An explicit github:owner/repo[#ref][:path] shorthand, a GitHub URL, or a local
path skips the ladder. A name that matches more than one indexed repo or
registry author is never guessed: asm get exits non-zero and lists the
qualified candidates.
The catalog stores metadata, not bodies, so an index, registry or remote
asm get costs a shallow clone into a temp directory, which is deleted before
the command returns. Those fetches run the same pre-install security scan
asm install runs; the verdict goes to stderr and into --json under
security. It reports rather than blocks — asm get never installs or executes the skill — and
--audit prints the full asm audit security report for the fetched skill.
Output discipline: stdout carries the body (or, with --json, a single JSON
object) and nothing else, so piping and redirecting are safe. Provenance,
progress, and the security verdict go to stderr.
Borrow the full skill directory
When a skill needs supporting scripts, templates, references or binary assets:
borrowed_path="$(asm get --path code-review)" # also: asm get code-review --path
# Read "$borrowed_path/SKILL.md" and its supporting files after get exits.
asm cleanup "$borrowed_path" # caller cleans up when finished
Every source tier is copied, including local paths, installed skills and
library entries. The original is never changed or deleted. The copy lives under
ASM's config directory in get-borrows/, independently of dependency sessions;
the remote staging clone is still deleted before get returns. There is no
exit-triggered borrow deletion and no --keep requirement. asm install remains
the permanent installation workflow.
Plain --path stdout is one absolute directory path plus a newline; provenance,
security and cleanup guidance stay on stderr. --path --json (or --machine)
returns path, sorted relative files, existing provenance/security fields,
and cleanup: {command: "asm", args: ["cleanup", path]} instead of content.
Git internals and ASM ownership metadata are omitted from the file list; .git
is not copied. Executable modes and binary files are preserved. Installed root
symlinks are resolved before copying; nested symlinks are refused.
asm cleanup <borrowed-path> accepts only an exact registered borrow root, not
an original source, parent, child, or symlink alias. Unknown and repeated cleanup
are safe no-ops. Missing or mismatched ownership proof refuses deletion and may
quarantine suspicious replacement content for inspection. Cleanup also supports
--json and --machine.
Just-in-time optional dependencies
A skill can declare optional skill dependencies without installing them:
dependencies:
- code-review
- github:owner/repo:skills/helper
Installing the parent does not install this list. The calling agent owns the
run lifecycle: discover the list, acquire only a dependency that the current
branch actually reaches, use the returned skillMdPath immediately, and
release the caller-supplied session in its own finally or shutdown handler.
asm deps discover parent-skill --json
asm deps acquire code-review --session run-123 --json
# Read the returned skillMdPath now; no provider catalog rescan is required.
asm deps release --session run-123 --json
Acquisition records whether the resolved target existed before the run. Remote temporary copies live under ASM's config directory and are lease-owned; installed, library, and explicit local targets are recorded as pre-existing. Release is idempotent, removes only artifacts with matching ASM ownership proof, and preserves every pre-existing target.
ASM is a support tool, not a process supervisor: it does not launch the agent,
run its task, or detect arbitrary caller failure. Normal cleanup therefore
belongs in the caller's finally/shutdown handling. An uncatchable termination
(for example, power loss or SIGKILL) can leave persistent lease state. On a
later invocation, the caller may classify and recover leases older than an
explicit, conservatively chosen cutoff:
asm deps cleanup --stale-before 2026-09-08T00:00:00Z --dry-run --json
asm deps cleanup --stale-before 2026-09-08T00:00:00Z --json
Age is a caller-selected recovery policy, not proof that an agent failed. Pick
a cutoff older than the longest expected live run and inspect --dry-run
output first. ASM returns a neutral canonical path and does not promise that a
foreign provider will rescan its skill catalog in the middle of a session.
FAQ
Is it free? MIT licensed. No accounts or paywalls.
Is this built for AI agents?
Yes. Commands return structured JSON (--json), accept non-interactive flags (--yes, --machine), and map to discrete actions an agent can chain — list inventory, search catalog, install, audit, uninstall.
Which agents are supported?
21 providers: a generic Agents provider (supported by most harnesses except Claude Code), Claude Code, Pi, OpenCode, Codex, Oh My Pi, Grok CLI, Hermes, OpenClaw, Cursor, GitHub Copilot, Windsurf, Google Antigravity, Gemini CLI, Cline, Roo Code, Continue, Aider, Zed, Augment, and Amp. Disable any via asm config edit.
What about the TUI?
Run asm with no args for an optional local browser. The CLI is the primary interface for scripts, CI, and agent tool calls.
Private repos?
asm install github:user/repo --transport ssh --yes
SKILL.md format?
A directory with SKILL.md (YAML frontmatter + markdown body). Run asm init my-skill to scaffold one.
Get started
npm install -g agent-skill-manager
asm list --json
Browse catalog · Documentation · Contributing · MIT Licensed
Build, test, and ship skills
Scaffold
asm init my-skill
asm init my-skill -p claude
asm init my-skill --path ./skills
Develop with live reload
asm link ./my-skill -p claude
asm link ./skill-a ./skill-b ./skill-c -p claude
asm link ./my-skills-folder -p claude
asm link ./my-skill --name my-alias -p claude --force
Audit and inspect
asm audit security my-skill
asm audit security ./path/to/my-skill
asm audit security --all
asm inspect my-skill
asm inspect my-skill --json
Test install flow
asm install github:you/awesome-skill
asm install github:you/awesome-skill -p claude --force --yes --json
Publish to ASM Registry
asm publish ./my-skill
asm publish --dry-run ./my-skill
Requires gh CLI authenticated with gh auth login.
Typical workflow
asm init awesome-skill -p claude- Edit
SKILL.md asm link ./awesome-skill -p claude- Test with your agent
asm audit security awesome-skillasm inspect awesome-skillasm eval ./awesome-skill- Push to GitHub
asm install github:you/awesome-skillasm publish ./awesome-skill
Skill verification and quality eval
Indexed skills are evaluated automatically. Skills passing all criteria get a verified badge in the catalog.
Verification criteria (all required):
- Valid frontmatter with
nameanddescription - Body content with at least 20 characters of instruction text
- No malicious patterns (
atob(), suspicious base64, hex escapes, hardcoded credentials) - Directory contains a readable
SKILL.md
asm index ingest github:your-user/your-repo
asm index search "your-skill" --json
Quality scoring (asm eval) runs a rubric over structure, clarity, safety, and naming:
asm eval ./my-skill
asm eval ./my-skill --machine
asm eval ./my-skill --fix
asm eval-providers list
See docs/eval-providers.md for the provider model.
ASM Registry
The ASM Registry lists community-published skills. Install by name — no GitHub URL needed.
asm install code-review
asm install luongnv89/code-review
asm install code-review --no-cache
asm publish ./my-skill
| Flag | Description |
|---|---|
--dry-run | Preview manifest without creating a PR |
--force | Override warning-level security findings |
--yes | Skip confirmation |
--machine | JSON output |
Resolution flow: fetch index (1-hour cache) → find manifest with pinned commit → clone at exact commit → install.
Open-source skill collections
Over 2,800 skills across curated repositories. Use asm search <term> to discover, then asm install github:owner/repo.
Last updated: 2026-07-07
| Repository | Description | Stars | Skills |
|---|---|---|---|
| anthropic-skills | Official Agent Skills from Anthropic | 95,957 | 18 |
| superpowers | Agentic skills framework | 89,816 | 14 |
| everything-claude-code | Performance optimization for Claude Code, Codex | 81,392 | 183 |
| agency-agents | Specialized expert agents | 50,749 | — |
| ui-ux-pro-max-skill | Design intelligence for UI/UX | 43,112 | 7 |
| antigravity-awesome-skills | 1,000+ skills for Claude Code, Cursor | 25,047 | 1,322 |
| marketingskills | Marketing — CRO, SEO, growth | 14,099 | 33 |
| agentskills | Agent Skills specification | 13,342 | — |
| claude-skills | 192 skills across engineering, marketing | 7,434 | 451 |
| taste-skill | Stops generic AI output | 3,389 | 7 |
| affiliate-skills | Affiliate marketing funnel | 99 | 48 |
| skills | Reusable agent skills | 1 | 35 |
asm install github:anthropics/skills
asm install github:anthropics/skills --all
Supported agent tools
21 built-in providers, all enabled by default. Disable via asm config edit.
| Tool | Global Path | Project Path |
|---|---|---|
| Agents (most harnesses except Claude Code) | ~/.agents/skills/ | .agents/skills/ |
| Claude Code | ~/.claude/skills/ | .claude/skills/ |
| Pi | ~/.pi/skills/ | .pi/skills/ |
| OpenCode | ~/.config/opencode/skills/ | .opencode/skills/ |
| Codex | ~/.codex/skills/ | .codex/skills/ |
| Oh My Pi | ~/.omp/agent/skills/ | .omp/skills/ |
| Grok CLI | ~/.grok/skills/ | .grok/skills/ |
| Hermes | ~/.hermes/skills/ | .hermes/skills/ |
| OpenClaw | ~/.openclaw/skills/ | .openclaw/skills/ |
| Cursor | ~/.cursor/rules/ | .cursor/rules/ |
| GitHub Copilot | ~/.github/instructions/ | .github/instructions/ |
| Windsurf | ~/.windsurf/rules/ | .windsurf/rules/ |
| Google Antigravity | ~/.antigravity/skills/ | .antigravity/skills/ |
| Gemini CLI | ~/.gemini/skills/ | .gemini/skills/ |
| Cline | ~/Documents/Cline/Rules/ | .clinerules/ |
| Roo Code | ~/.roo/rules/ | .roo/rules/ |
| Continue | ~/.continue/rules/ | .continue/rules/ |
| Aider | ~/.aider/skills/ | .aider/skills/ |
| Zed | ~/.config/zed/prompt_overrides/ | .zed/rules/ |
| Augment | ~/.augment/rules/ | .augment/rules/ |
| Amp | ~/.amp/skills/ | .amp/skills/ |
Add custom providers in config.
Troubleshooting: shadowed installs
Multiple asm binaries on PATH can shadow a fresh upgrade.
Diagnose: asm --version warns about multiple binaries. Run asm doctor for a full report.
Fix: Remove the stale install with your package manager, then confirm with asm --version.
CLI Commands
Commands
| Command | Description |
| ------------------------------- | ----------------------------------------- | ----------------------------------------- |
| asm list | List all discovered skills |
| asm search <query> | Search by name/description/provider |
| asm tag add\|remove | Edit local tags on an installed skill |
| asm inspect <skill-name> | Show detailed info for a skill |
| asm get <skill> | Print a skill's body, install nothing — --path borrows the full directory |
| asm cleanup <path> | Remove a directory borrowed by asm get --path |
| asm deps discover\|acquire\|release\|cleanup | Manage caller-owned temporary dependency leases |
| asm install <source> | Install from GitHub or registry |
| asm publish [path] | Publish to ASM Registry |
| asm uninstall <skill-name> | Remove a skill |
| asm init <name> | Scaffold a new skill |
| asm link <path> [<path2> ...] | Symlink skills for live dev |
| asm audit | Detect duplicate skills |
| asm audit security <name> | Security audit |
| asm audit residency | Rank skills that do not earn residency |
| asm eval <skill> | Score skill quality |
| asm eval-providers list | List eval providers |
| asm stats | Aggregate installed skill metrics |
| asm stats repo <repo> | Per-repo indexed skill stats |
| asm stats author <owner> | Per-author indexed skill stats |
| asm stats index | Indexed catalog stats summary |
| asm activate <skill> | Link a library skill into a provider |
| asm deactivate <skill> | Remove a library activation symlink |
| asm library | Manage centrally installed library skills |
| asm library list | List local library skills |
| asm library update <skill> | Update one local library skill |
| asm library update --all | Update all local library skills |
| asm export | Export inventory as JSON |
| asm import <file> | Import skills from exported manifest |
| asm outdated | Show outdated skills |
| asm update [name...] | Update outdated skills |
| asm disable <target> | Disable skill(s) without uninstalling |
| asm enable <target> | Re-enable disabled skill(s) |
| asm index ingest <repo> | Index a skill repo |
| asm index search <query> | Search indexed skills |
| asm index list | List indexed repos |
| asm index remove <owner/repo> | Remove repo from index |
| asm index overlap | Find semantically overlapping skills in the index |
| asm bundle list | List bundles (--predefined) |
| asm bundle install <name> | Install every skill in a bundle |
| asm bundle create <name> | Create bundle from installed skills |
| asm bundle show <name> | Show bundle details |
| asm bundle modify <name> | Add/remove skills |
| asm bundle export <name> | Export bundle to JSON |
| asm bundle remove <name> | Remove saved bundle |
| asm config show | Print config |
| asm config path | Print config path |
| asm config reset | Reset to defaults |
| asm config edit | Open config in $EDITOR |
| asm doctor | Run environment health checks and diagnostics |
Global options
-h, --help Show help
-v, --version Print version
--json JSON output
--machine Stable machine-readable JSON envelope (v1)
-s, --scope <scope> global, project, or both
-p, --tool <name> Filter by tool (list, search)
--tag <tag[,tag]> Filter by all tags; repeatable (list, search)
--sort <field> name, version, or location
--flat Show one row per tool instance (list, search)
--model-invocable Only skills the model can invoke
--user-invocable Only skills the user can invoke
-y, --yes Skip confirmations
--no-color Disable ANSI colors
-V, --verbose Show debug output
Examples
asm list --scope global --sort location
asm list --summary
asm list --compact --group-by tool --limit 20
asm list --model-invocable --user-invocable
Listings and asm inspect show invocability as model, user, or both (never collapsed). --model-invocable and --user-invocable are independent; both flags keep skills that match both (typically both).
asm tag add code-review testing cli
asm tag remove code-review automation
asm list --tag cli --tag testing
asm search code --tag cli,testing
Tags are local overlays (not written into SKILL.md). --tag requires every listed tag (AND). The catalog UI TagFilter uses the same AND matching and stores selections in ?tag=.
asm search "code review" --json
asm audit --yes
asm audit security github:user/repo
asm audit security --all
asm stats --tokens
asm audit residency --json
asm eval ./my-skill
asm init my-skill -p claude
asm link ./my-skill -p claude
asm uninstall old-skill --yes
asm index ingest github:anthropics/skills
asm index search "frontend design" --json
Bundles — pre-defined sets for common workflows. Browse at luongnv.com/asm/#/bundles:
asm bundle list --predefined
asm bundle install frontend-dev
asm bundle install ./my-bundle.json
In a terminal, asm bundle install prompts for the tool(s), then which bundle
skills to install, then the install scope; Esc on any picker aborts. Pass a
prompt's own flag to skip it — -p/--tool for the tool picker,
-s/--scope global|project for the scope picker, -y for the skill and scope
pickers. Outside a terminal there are no pickers, and -p/--tool is required
unless exactly one tool is enabled:
asm bundle install frontend-dev --tool claude --scope project --yes
asm bundle create my-workflow
asm bundle export my-workflow ./my-workflow.json
iOS/Swift catalog: ASM indexes public Swift/Apple skill repos. Use asm bundle install ios-release or search for swift, swiftui, swift testing, uikit, swiftdata, app store connect.
Pre-defined bundles: frontend-dev, devops, content-writing, ios-release, marketing, software-dev, game-dev — full list via asm bundle list --predefined.
Installing skills from GitHub
Single-skill repo:
asm install github:user/my-skill
asm install github:user/my-skill#v1.0.0 -p claude
Multi-skill repo:
asm install github:user/skills --path skills/code-review
asm install github:user/skills --all -p claude -y
Subfolder URL:
asm install https://github.com/user/skills/tree/main/skills/agent-config
asm install github:user/skills#main:skills/agent-config
Private repos:
asm install github:user/private-skill --transport ssh
asm install github:user/private-skill -t auto
Vercel skills CLI:
asm install github:user/skills --method vercel --skill my-skill
| Flag | Description |
|---|---|
-p, --tool <name> | Target agent |
--name <name> | Override directory name |
--path <subdir> | Install from subdirectory |
--all | Install all skills in repo |
-m, --method <method> | default or vercel |
-t, --transport <mode> | https, ssh, or auto |
--library | Install into the local library |
-f, --force | Overwrite existing |
-y, --yes | Skip confirmation |
--json | JSON output |
Multi-skill repos: scans up to 5 levels deep. Root SKILL.md installs by default; use --all for every skill. Requires git on PATH.
Optional TUI (run asm with no args)
Local browser for skills. The CLI above is the primary interface for agents and automation.
Keyboard shortcuts
| Key | Action |
|---|---|
↑/↓ or j/k | Navigate list |
Enter | View details |
d | Uninstall |
/ | Search / filter |
Esc | Back / clear |
Tab | Cycle scope |
s | Cycle sort |
r | Refresh |
c | Configuration |
a | Audit duplicates |
q | Quit |
? | Help |
Configuration
On first run, config is created at ~/.config/agent-skill-manager/config.json:
{
"version": 1,
"providers": [
{
"name": "claude",
"label": "Claude Code",
"global": "~/.claude/skills",
"project": ".claude/skills",
"enabled": true
},
{
"name": "codex",
"label": "Codex",
"global": "~/.codex/skills",
"project": ".codex/skills",
"enabled": true
},
{
"name": "opencode",
"label": "OpenCode",
"global": "~/.config/opencode/skills",
"project": ".opencode/skills",
"enabled": true
},
{
"name": "pi",
"label": "Pi",
"global": "~/.pi/skills",
"project": ".pi/skills",
"enabled": true
},
{
"name": "hermes",
"label": "Hermes",
"global": "~/.hermes/skills",
"project": ".hermes/skills",
"enabled": true
},
{
"name": "openclaw",
"label": "OpenClaw",
"global": "~/.openclaw/skills",
"project": ".openclaw/skills",
"enabled": true
},
{
"name": "agents",
"label": "Agents",
"global": "~/.agents/skills",
"project": ".agents/skills",
"enabled": true
},
{
"name": "cursor",
"label": "Cursor",
"global": "~/.cursor/rules",
"project": ".cursor/rules",
"enabled": true
},
{
"name": "copilot",
"label": "GitHub Copilot",
"global": "~/.github/instructions",
"project": ".github/instructions",
"enabled": true
},
{
"name": "windsurf",
"label": "Windsurf",
"global": "~/.windsurf/rules",
"project": ".windsurf/rules",
"enabled": true
},
{
"name": "antigravity",
"label": "Google Antigravity",
"global": "~/.antigravity/skills",
"project": ".antigravity/skills",
"enabled": true
},
{
"name": "gemini",
"label": "Gemini CLI",
"global": "~/.gemini/skills",
"project": ".gemini/skills",
"enabled": true
},
{
"name": "cline",
"label": "Cline",
"global": "~/Documents/Cline/Rules",
"project": ".clinerules",
"enabled": true
},
{
"name": "roocode",
"label": "Roo Code",
"global": "~/.roo/rules",
"project": ".roo/rules",
"enabled": true
},
{
"name": "continue",
"label": "Continue",
"global": "~/.continue/rules",
"project": ".continue/rules",
"enabled": true
},
{
"name": "aider",
"label": "Aider",
"global": "~/.aider/skills",
"project": ".aider/skills",
"enabled": true
},
{
"name": "zed",
"label": "Zed",
"global": "~/.config/zed/prompt_overrides",
"project": ".zed/rules",
"enabled": true
},
{
"name": "augment",
"label": "Augment",
"global": "~/.augment/rules",
"project": ".augment/rules",
"enabled": true
},
{
"name": "amp",
"label": "Amp",
"global": "~/.amp/skills",
"project": ".amp/skills",
"enabled": true
}
],
"customPaths": [],
"preferences": {
"defaultScope": "both",
"defaultSort": "name"
}
}
- All 21 providers are enabled by default
- Set
"enabled": falseto skip a provider you don't use - Add arbitrary directories via
customPaths - Manage via
asm config show|path|reset|editor presscin the TUI
SKILL.md format
Every skill is a directory with a SKILL.md file — YAML frontmatter plus markdown instructions.
---
name: my-skill
description: "What this skill does"
license: "MIT"
compatibility: "Claude Code, Codex"
allowed-tools: Bash Read Grep Glob WebFetch
effort: medium
metadata:
version: 1.0.0
creator: "Your Name <you@example.com>"
---
| Field | Required | Description |
|---|---|---|
name | yes | Unique identifier |
description | yes | One-line summary |
license | no | SPDX identifier |
compatibility | no | Compatible agents |
allowed-tools | no | Tools the skill uses |
effort | no | low, medium, high, max |
metadata.version | no | Semver (default 0.0.0) |
metadata.creator | no | Author |
asm init my-skill
asm init my-skill -p claude
From source
git clone https://github.com/luongnv89/asm.git
cd asm
npm install
npm run build
npm start
Inspect install script before running:
curl -sSL https://raw.githubusercontent.com/luongnv89/asm/main/install.sh -o install.sh
less install.sh
bash install.sh
Project structure
asm/
├── bin/ # CLI entry point
├── dist/ # Built bundle (npm ships this)
├── scripts/ # Build, preindex, catalog
├── src/
│ ├── index.tsx # TUI (ink)
│ ├── cli.ts # CLI dispatcher
│ ├── scanner.ts # Skill discovery
│ ├── installer.ts # GitHub install pipeline
│ └── views/ # TUI views
├── docs/ # Extended documentation
├── assets/ # Logos and screenshots
├── install.sh # curl installer
└── package.json
Tech stack
Documentation
| Document | Description |
|---|---|
| Architecture | System design and data flow |
| Eval Providers | Pluggable eval framework |
| Development | Local setup and debugging |
| Deployment | Publishing and CI |
| Changelog | Version history |
| Brand Kit | Logo, colors, typography |
| Contributing | How to contribute |
| Security | Vulnerability reporting |
| Code of Conduct | Community guidelines |
Landing page: luongnv.com/asm — catalog, bundles, author/repo stats, filtered search, and catalog TagFilter (AND, ?tag=).
Acknowledgements
Contributors
| Contributor | PRs |
|---|---|
| @luongnv89 | 38 merged PRs |
| @Mordris | #111 |
Dependencies
| Library | Description |
|---|---|
| ink | React renderer for the TUI |
| @inkjs/ui | Prebuilt ink components |
| react | UI for TUI and web catalog |
| react-dom | DOM renderer for catalog |
| react-window | List virtualization for catalog |
| yaml | SKILL.md frontmatter parser |
Roadmap
Contributing
See CONTRIBUTING.md.