AgentSkillsCN

aget-sanity-check

运行AGET健康检查与日常维护工作。

SKILL.md
--- frontmatter
name: aget-sanity-check
description: Run AGET health inspection and housekeeping checks
allowed-tools:
  - Bash
  - Read
  - Glob

aget-sanity-check

Run health inspection on an AGET agent. This skill performs sanity checks, validates structure, and reports agent health status.

Instructions

When this skill is invoked:

  1. •

    Run the housekeeping script:

    bash
    python3 scripts/aget_housekeeping_protocol.py
  2. •

    The script checks:

    • •.aget/ directory structure
    • •version.json validity
    • •identity.json validity (if exists)
    • •governance/ directory contents
    • •5D directory structure
    • •L-doc count and format
  3. •

    Report the health status:

    • •Healthy (exit 0): All checks passed
    • •Warnings (exit 1): Non-blocking issues found
    • •Errors (exit 2): Blocking issues require attention

Output Format

Present the health report:

code
Sanity Check: [HEALTHY/WARNINGS/ERRORS]

Checks:
- Structure: [PASS/WARN/FAIL]
- Version: [PASS/WARN/FAIL]
- Identity: [PASS/WARN/FAIL]
- Governance: [PASS/WARN/FAIL]

[If warnings/errors, list specific issues]

Recommendations:
- [Action items if any]

Options

  • •--json: Machine-readable output
  • •--fix: Attempt auto-fixes for known issues
  • •--dir /path: Run on specific agent directory

Error Handling

  • •Exit code 0: All checks passed
  • •Exit code 1: Warnings found (list them)
  • •Exit code 2: Errors found (list them, recommend fixes)
  • •Exit code 3: Script configuration error

Related

  • •L532: Skills vs Learnings Distinction
  • •CAP-SESSION-002: Sanity Check Protocol