Skip to main content
Glama

save_skill

Save a skill to your private registry so you can run, share, or improve it later (name, description, body, outcome, tags).

Addressing: omit skill_id to save in your own namespace, which updates your own skill of this name or creates a new one. Saving by name never edits a skill someone else shared with you. To edit a skill you own or have edit access to, pass its skill_id (and a matching name). If a skill of this name is shared with you at edit access and you save by name alone, the reference is ambiguous and the call asks you to address it by id.

Optional source ('manual' | 'teach' | 'optimization' | 'description_optimization' | 'audit') attributes the version to a manual save, a teach_skill session, a skill optimization session, a description optimization session, or an audit_skill session; omit to leave NULL.

Returns skill_id and version. Skills are private to the caller; share via publish_template_version.

files is a full snapshot: any path you do not include is deleted. Only pass files when you hold the complete tree. To change specific files in a skill that already exists, use update_skill_files, which keeps the paths you do not name. When a save does drop paths, the response lists them under deleted, so check that key and restore anything you did not mean to remove.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
payloadYesPublish-skill payload aligned with the Claude Code skills convention. Required top-level fields mirror a Claude Code ``SKILL.md`` frontmatter: ``name`` (kebab-case handle) and ``description`` (one-sentence summary). ``outcome`` and ``tags`` are top-level discovery facets surfaced by ``list_skills``; everything else lives in the ``body`` and is opaque to the registry. Skills are always private to the caller. Public sharing is the ``templates`` surface (``publish_template_version`` / ``fork_template``); there is no ``visibility`` knob on a skill row. Unknown fields are silently dropped (``extra="ignore"``). Fields such as ``manifest``, ``visibility``, and ``kpi`` do not cause validation errors and are simply ignored. The server accepts both ``slug`` and ``name`` as input for the skill handle. Responses return both ``name`` and ``slug`` fields with identical values.

Output Schema

TableJSON Schema
NameRequiredDescriptionDefault

No arguments

Schema Changelog

Changes observed during successful MCP inspections.

  1. Changed3 schema fields changed
    • changedInput schema / properties / payload / properties / files / description
      Previous value: -"Sibling files bundled with the skill. Include a 'demo/README.md' writeup with images referenced by relative path inside 'demo/' to render a visual demo on the public template page once the skill is published as a template."New value: +"Sibling files bundled with the skill. Include a 'demo/README.md' writeup with images referenced by relative path inside 'demo/' to render a visual demo on the public template page once the skill is published as a template. A supplied list is a full snapshot: any path you do not include is deleted. Omitting the field entirely preserves the existing tree unchanged; to change specific files without resending the rest, use `update_skill_files` instead."
    • changedInput schema / properties / payload / properties / outcome / description
      Previous value: -"The business outcome this skill is meant to drive, surfaced as a discovery facet."New value: +"The business outcome this skill is meant to drive, surfaced as a discovery facet. When omitted on an update, the existing outcome is preserved; send null to clear it."
    • changedInput schema / properties / payload / properties / tags / description
      Previous value: -"Discovery tags surfaced when listing and searching skills."New value: +"Discovery tags surfaced when listing and searching skills. When omitted on an update, the existing tags are preserved; send an empty list to clear them."
  2. Added

TDQS

A4.8/5.0
Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Adds significant behavioral context beyond annotations: skills are private (shared via publish_template_version), files is a full snapshot (deleting omitted paths), response includes deleted files for recheck, and saving by name never edits shared skills. No contradiction with annotations; clarifies conditional destructiveness not captured by destructiveHint.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Well-structured with clear paragraphs: first sets purpose, then addressing behavior, source, return/sharing, and files. Every sentence adds value. Slightly lengthy but justified by complexity; could potentially combine some points without losing clarity.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given the tool's complexity (nested schema, file snapshot semantics, sharing nuances) and presence of output schema, the description is highly complete. Covers edge cases (shared skill ambiguity, concurrency token), relationship with siblings, and notes what's not included (visibility knob). Fulfills all informational needs.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 100%, but description adds extra meaning: explains relationship between skill_id and name, files snapshot versus null preservation, source attribution, expected_version_token concurrency, and the fact that unknown fields are silently dropped. Also notes server acceptance of both slug and name.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the tool saves a skill to a private registry for later run, share, or improvement. It highlights key fields (name, description, body, outcome, tags) and distinguishes between create and update via skill_id, with explicit mentions of sibling tools like update_skill_files.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Provides explicit when/when-not guidance: omit skill_id to save in own namespace (create or update own skill); pass skill_id to edit a specific skill; warns about ambiguous name sharing for edit-access skills. Also explains files snapshot behavior versus incremental update using update_skill_files, and advises to check the deleted key in response.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Try in Browser

Glama MCP Gateway

Add one secure layer between your agents and this server.