Install
$ agentstack add skill-xfurti-leogriel-leogriel ✓ scanned · ✓ verified — works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
Verified badge
Passed review? Show it. Paste this badge into your README — it links to the public security report.
Reliability & compatibility
Declared compatibility
Compatibility is declared by the source manifest. End-to-end runtime verification is coming — see below.
We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps — measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.
How agent discovery & health will work →About
Leogriel
Operational playbook for managing Agent Skills with the leogriel CLI. Run real commands; do not simulate output or copy SKILL.md files between agent directories by hand.
Golden rules
- Choose the intended scope — project skills live in
.leogriel/skills//; explicit personal skills added with-glive in~/.leogriel/skills//. - Commit project state — track
agent-skills.json,agent-skills.lock, and vendored.leogriel/skills/content required by local/imported entries. - Use portable specifiers — prefer
github:,npm:,skills.sh/, or project-relativefile:./; never commit absolute machine paths. - Install before sync —
installmaterializes project dependencies;synconly refreshes agent targets. - Verify after changes — run
leogriel doctorandleogriel audit; useaudit --strictin CI. - Import rather than copy by hand — plain
leogriel importdiscovers agent directories, deduplicates identical skills, and vendors selected content into the project store. - Plan maintenance before writing — use
outdatedorupdate --dry-run; use--latest --save --yesonly when intentionally changing an npm constraint. - Never replace unmanaged targets implicitly — replacement requires exact skill, agent, scope, confirmation, and a backup.
- Treat behavioral tests as untrusted code — pin a model when comparing over time, keep network denied unless required, and review every command assertion before using
--trust-tests.
Decision tree
| Situation | Command | |-----------|---------| | New project, no manifest | leogriel init or leogriel init --with-skill | | Add a project dependency | leogriel add then leogriel install | | Add a personal skill outside a project | leogriel add -g | | Skills already in .claude/skills, .codex/skills, etc. | leogriel import --dry-run then leogriel import | | Migrating from npx skills | leogriel import from-npx | | Re-link project targets | leogriel sync --project | | Re-link personal targets | leogriel sync --global | | Re-fetch project dependencies | leogriel update or leogriel update | | Discover or inspect a skill | leogriel search / leogriel info | | Review pending updates | leogriel outdated / leogriel update --dry-run | | CI reproducible install | leogriel install --frozen then leogriel audit --strict | | Diagnose personal installation | leogriel doctor -g | | Compare behavior with and without a skill | leogriel test --runs 3 --model |
Quick recipes
# Bootstrap a project
leogriel init --with-skill
leogriel add github:vercel-labs/agent-skills@main#skills/web-design-guidelines
leogriel install
# Vendor a project-local skill
leogriel add file:./my-skill
leogriel install
leogriel sync --project
# Import existing agent directories
leogriel import --dry-run
leogriel import
# Personal/global skill
leogriel add -g file:./my-personal-skill
leogriel list -g
leogriel doctor -g
# Inspect a project
leogriel list
leogriel doctor
leogriel audit --json
# Discovery and maintenance
leogriel search typescript
leogriel info skills.sh/vercel-labs/skills/find-skills
leogriel outdated
leogriel update --dry-run
After a local add/import, expect the manifest and lock to use file:./.leogriel/skills/ and the lock canonicalPath to use .leogriel/skills/. A global add uses ~/.leogriel/skills/ and stores its state under ~/.leogriel/.
Reproducibility (0.6+)
- Immutable remote lock: GitHub and skills.sh resolve to a full commit SHA; npm resolves to an exact version and tarball integrity.
- Project-portable: remote specifiers and
file:./.leogriel/skills/withcanonicalPath: .leogriel/skills/. - Explicitly global:
add -g,list -g,doctor -g, andremove -g; do not use global scope for team dependencies. - Legacy:
local:imported/, absolute local paths, and project entries withcanonicalPath: ~/.leogriel/skills/should be re-imported or re-added, then committed. install --frozenrestores missing remote content from immutable lock entries without changing the lock.updateis the normal operation that intentionally changes an existing remote resolution.skills.sh///locks the repository SHA plus a skill selector; frozen install does not query the catalog.
Scoped sync and JSON
- No sync scope flags means project and global targets.
- Use
--projector--global, plus repeatable/comma-separated--agentfilters. --pruneis opt-in and only removes verified leogriel-managed links/copies; combine it with--dry-runfirst.--replace-unmanagedrequires--skill,--agent, one scope, and confirmation; it backs up the original content before replacement.- First-party commands with
--jsonemit one envelope. Exit codes: 0 success, 1 warning/partial result, 2 fatal/validation failure.
Plugins, SARIF, and completion (0.7+)
- Plugins are experimental, integrity-locked, and execute Node.js with the user's permissions; they are not sandboxed.
- Prefer npm plugins. Local plugins require
plugin add --allow-local. audit --format sarif --output results.sarifwrites GitHub-compatible SARIF while the default audit remains offline.completion bash|zsh|powershellprints a script and never edits a shell profile automatically.
Failure modes
| Symptom | Action | |---------|--------| | Project canonical path missing | Run leogriel install; if the lock still uses the legacy global path, re-add/re-import the skill | | Global canonical path missing | Run leogriel doctor -g, then re-add the personal skill if needed | | Integrity mismatch | leogriel update for remote content, or re-add/re-import changed local content | | Symlink fails on Windows | Set config defaultMode: copy, then run leogriel doctor --fix | | No project agents linked | leogriel sync --project; verify enabled agents in ~/.leogriel/config.json | | Frozen install failed | Fix manifest/lock drift; legacy mutable entries require leogriel update | | mutable-resolution | Run leogriel update once to upgrade a legacy remote entry | | E_LOCK_TIMEOUT | Wait for the other leogriel process or inspect stale locks with doctor |
References
- [commands.md](references/commands.md) — full command cheat sheet
- [specifiers.md](references/specifiers.md) — specifier grammar
- [manifest-lock.md](references/manifest-lock.md) — manifest and lock semantics
- [workflows.md](references/workflows.md) — team onboarding, migration, CI
- [troubleshooting.md](references/troubleshooting.md) — doctor, Windows, coexistence
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: xFurti
- Source: xFurti/leogriel
- License: MIT
- Homepage: https://xfurti.github.io/leogriel/
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet — be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.