==================================================================
  GearScope sandbox test
  skill:    agents-best-practices
  variant:  fresh
  script:   /Users/openclaw/gearscope/sandbox/skills/agents-best-practices/fresh.sh
  sandbox:  gs-agents-best-practices-fresh-20260516-234700
  started:  2026-05-16T22:47:04Z
  sbx:      Client Version:  v0.29.0 7055fecde6b84aeb963d1680879e5620af15c119
unknown
==================================================================
[run-test] creating sandbox...
c70a7d044afb: Already exists
e07454cc05d8: Already exists
81438aaf4f82: Already exists
Digest: sha256:c70a7d044afbb8b6fc0ab6a41e0cd3c704df9c61c68906e6bfef68e49e4215fb
Status: Image is up to date for docker/sandbox-templates:shell-docker
INFO: Configuring Docker
✓ Created sandbox 'gs-agents-best-practices-fresh-20260516-234700'
  Workspace: /Users/openclaw/gearscope (direct mount)
  Agent: shell

To connect to this sandbox, run:
  sbx run gs-agents-best-practices-fresh-20260516-234700
[run-test] executing test script in sandbox...
INFO: Starting Docker daemon
--- Cloning agents-best-practices ---
Cloning into 'agents-best-practices'...

=== 1. Repo structure check ===
  [OK] SKILL.md exists
  [OK] README.md exists
  [OK] LICENSE exists
  [OK] references exists
  Reference markdown files: 15

=== 2. SKILL.md frontmatter validation ===
  [OK] SKILL.md has 'name' field
  [OK] SKILL.md has 'description' field
  [OK] Version: 1.2.0
  [WARN] Description does NOT start with 'Use when': Use this skill when designing, generating an MVP blueprint for, auditing, refact...
  [OK] Frontmatter size: 602 bytes (within 1024 limit)
  Total frontmatter errors: 0

=== 3. Reference file existence ===
  [OK] references/mvp-agent-blueprint.md (572 lines)
  [OK] references/architecture.md (283 lines)
  [OK] references/agentic-loop.md (245 lines)
  [OK] references/tools-and-permissions.md (297 lines)
  [OK] references/planning-and-goals.md (234 lines)
  [OK] references/context-memory-compaction.md (314 lines)
  [OK] references/prompt-caching-and-cost.md (247 lines)
  [OK] references/skills-and-connectors.md (281 lines)
  [OK] references/system-prompts-instructions.md (165 lines)
  [OK] references/provider-api-patterns.md (240 lines)
  [OK] references/security-evals-observability.md (212 lines)
  [OK] references/agent-legibility-feedback-loops.md (185 lines)
  [OK] references/checklists.md (199 lines)
  [OK] references/coverage-audit.md (61 lines)
  [OK] references/source-links.md (54 lines)
  Missing references: 0

=== 4. Cross-reference validity ===
  [OK] SKILL.md -> references/agent-legibility-feedback-loops.md (exists)
  [OK] SKILL.md -> references/agentic-loop.md (exists)
  [OK] SKILL.md -> references/architecture.md (exists)
  [OK] SKILL.md -> references/checklists.md (exists)
  [OK] SKILL.md -> references/context-memory-compaction.md (exists)
  [OK] SKILL.md -> references/coverage-audit.md (exists)
  [OK] SKILL.md -> references/mvp-agent-blueprint.md (exists)
  [OK] SKILL.md -> references/planning-and-goals.md (exists)
  [OK] SKILL.md -> references/prompt-caching-and-cost.md (exists)
  [OK] SKILL.md -> references/provider-api-patterns.md (exists)
  [OK] SKILL.md -> references/security-evals-observability.md (exists)
  [OK] SKILL.md -> references/skills-and-connectors.md (exists)
  [OK] SKILL.md -> references/source-links.md (exists)
  [OK] SKILL.md -> references/system-prompts-instructions.md (exists)
  [OK] SKILL.md -> references/tools-and-permissions.md (exists)
  [OK] README.md -> references/agentic-loop.md (exists)
  [OK] README.md -> references/checklists.md (exists)
  [OK] README.md -> references/context-memory-compaction.md (exists)
  [OK] README.md -> references/mvp-agent-blueprint.md (exists)
  [OK] README.md -> references/planning-and-goals.md (exists)
  [OK] README.md -> references/prompt-caching-and-cost.md (exists)
  [OK] README.md -> references/provider-api-patterns.md (exists)
  [OK] README.md -> references/security-evals-observability.md (exists)
  [OK] README.md -> references/skills-and-connectors.md (exists)
  [OK] README.md -> references/source-links.md (exists)
  [OK] README.md -> references/tools-and-permissions.md (exists)
  Cross-reference errors: 0

=== 5. Content depth: reference file word counts ===
  agent-legibility-feedback-loops.md: 907 words
  agentic-loop.md: 787 words
  architecture.md: 1190 words
  checklists.md: 1377 words
  context-memory-compaction.md: 1062 words
  coverage-audit.md: 549 words
  mvp-agent-blueprint.md: 2022 words
  planning-and-goals.md: 738 words
  prompt-caching-and-cost.md: 1033 words
  provider-api-patterns.md: 874 words
  security-evals-observability.md: 666 words
  skills-and-connectors.md: 986 words
  source-links.md: 282 words
  system-prompts-instructions.md: 647 words
  tools-and-permissions.md: 767 words
  TOTAL reference content: 13887 words

=== 6. SKILL.md content quality ===
  SKILL.md: 204 lines, 1580 words
  [OK] Has 'Core stance' section
  [OK] Has 'When to activate' section
  [OK] Has 'How to use' section
  [OK] Has 'Reference map' section
  [OK] Has 'Gotchas' section
  [OK] Has 'Non-negotiable' section
  Code block markers: 4 (expecting 4+ for a well-documented skill)

=== 7. License check ===
  License: MIT License

=== 8. README install instructions ===
  [OK] README mentions 'npx skills add'
  [OK] README mentions 'git clone'
  [OK] README mentions 'Codex'
  [OK] README mentions 'Claude Code'

=== 9. Coverage audit reference ===
  [OK] coverage-audit.md exists
  Coverage sections: 4

=== DONE ===

==================================================================
  finished: 2026-05-16T22:47:25Z
  exit:     0
==================================================================