AutoVault
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| GITHUB_TOKEN | No | Optional GitHub API rate-limit headroom. | |
| AUTOVAULT_MODE | No | Runtime mode: local for stdio/library, remote for HTTP MCP. | local |
| AUTOVAULT_DB_PATH | No | SQLite capability index path. | $AUTOVAULT_STORAGE_PATH/autovault.sqlite |
| AUTOVAULT_HTTP_PORT | No | HTTP port when PORT is not injected by the platform. | 3000 |
| AUTOVAULT_LOG_LEVEL | No | Log level: debug, info, warn, or error. | info |
| AUTOVAULT_PUBLIC_URL | No | Public origin for OAuth metadata and callbacks (required in remote mode). | |
| AUTOVAULT_ADMIN_EMAIL | No | First remote owner email (required until owner exists). | |
| AUTOVAULT_SEARCH_MODE | No | Search backend (text currently implemented). | text |
| AUTOVAULT_STORAGE_PATH | No | Root path for installed skills. | ~/.autovault |
| AUTOVAULT_PROFILE_LINKS | No | Comma-separated agent=/skills/root links for profile sync. | |
| AUTOVAULT_SKILL_INSTALL | No | Vendor routing: prefer-autovault, both, native, native-only, or off. | prefer-autovault |
| AUTOVAULT_ADMIN_PASSWORD | No | First remote owner password, at least 12 characters (required until owner exists). | |
| AUTOVAULT_ALLOWED_ORIGINS | No | Optional CORS allowlist for remote mode. | |
| AUTOVAULT_LOG_DIAGNOSTICS | No | Set to 1 to let structured diagnostic logs pass through public CLI output suppression. | |
| AUTOVAULT_SECURITY_STRICT | No | Block denylist hits when true; warn when false. | true |
| AUTOVAULT_AGENTSKILLS_BASE | No | Override the agentskills API base. | https://agentskills.io/api/v1 |
| AUTOVAULT_PROFILE_CONFIG_PATH | No | Optional named profile config path. | $AUTOVAULT_STORAGE_PATH/profiles.config.json |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": true
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| get_skillA | Find and load installed skills. Pass |
| add_skillA | Add a known skill from a source. For GitHub, agentskills, and URL sources, pass |
| propose_skillA | Submit a newly authored SKILL.md to AutoVault. Runs validation, security scan, capability cross-check, and three-tier deduplication (exact content hash → near-exact similarity → functional overlap). Use when the user asks to save a conversationally created skill, or after you've drafted one in response to a workflow the user wants reused. Always prefer this over writing skill files directly to disk. |
| bulk_importB | Import every immediate child directory under |
| update_skillA | Update an installed skill. With only |
| delete_skillA | Delete an installed skill from the vault and refresh generated profiles. Existing host skill roots are discovered by default so managed consumer links are pruned; pass discover_profile_roots: false to opt out. This also removes vault-local transforms for that skill. |
| check_updatesA | Detect upstream drift for skills installed from |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 7 tools
Each tool targets a distinct lifecycle action—get, add, propose, update, delete, bulk import, and update checking—with clear boundaries. Even the similar add_skill and propose_skill are explicitly separated by whether the skill is sourced from an existing location or newly authored.
Most tools follow a consistent verb_noun pattern with snake_case: get_skill, add_skill, update_skill, delete_skill, propose_skill. bulk_import and check_updates deviate slightly from the *_skill suffix but still use the same imperative style and are immediately understandable.
Seven tools is well-scoped for a skill vault server. Each tool covers a meaningful operation without redundancy, and the count is neither bloated nor too thin.
The toolset covers the full skill lifecycle: create/add, read/search, update, delete, and upstream change detection. A minor gap is the lack of a dedicated list/export tool, though get_skill with a query and alternates partially covers discovery.