skill-maintenance-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
No arguments | |||
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
} |
| logging | {} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| skill_backupA | 改前备份技能目录到 ~/.agents/skill-backups/<今天>/（skill-evolution 铁律: 修订前必备份）。 |
| skill_scan_corruptionA | 技能文件损坏扫描（铁律: 每次写入/修订后必跑）。检测 formfeed(\x0c) / 真 tab / 行中 CR——JSON 序列化转义被解释的隐性损坏。 |
| skill_log_decisionA | 向技能的 references/decision-log.md 追加决策记录（四字段: 诊断/修订/证据/结果——记「为什么改」, 未来 agent 不重新踩坑）。 |
| skill_read_decisionsB | 读取技能 decision-log 概要（条目标题列表），修订前先看避免重复踩坑。 |
| skill_validateC | 技能体检: frontmatter 必填（name/description/version）+ 损坏扫描 + 孤儿 reference（未挂链=不可发现）+ decision-log 概要。 |
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 5 tools
Each tool targets a distinct maintenance action: backup, corruption scan, decision logging, decision reading, and aggregate validation. However, skill_validate explicitly includes corruption scanning and a decision-log summary, so an agent may wonder when to call it versus skill_scan_corruption or skill_read_decisions.
All names use a consistent skill_ prefix and snake_case, which is predictable. Minor deviations exist: skill_backup and skill_validate are verb-only, while others use verb_noun, and log_decision/read_decisions differ in number.
Five tools is well-scoped for skill maintenance. Each tool covers a specific, non-redundant part of the backup-scan-validate-log workflow.
The core maintenance loop is present, but backup lacks a corresponding restore or list-backups operation, and decision-log reading only returns a summary. These are notable gaps for a server explicitly concerned with safe skill revision.