SKILLEMALL.ai
GitHub ↗ RU Sign in

DF self-evolving-agent

AI 비서가 자기 로그를 분석해서 AGENTS.md 개선안을 제안하는 자동화. v5.0: 시맨틱 임베딩(Ollama nomic-embed-text, FP ~8%), 실시간 스트리밍 모니터(<30초 알림), 플릿 분석(다중 인스턴스). v4.3: 대화형 승인, 멀티포맷 리포트, GitHub Issues. v4.2: Ollama LLM. v4.1: 멀티플랫폼 배달. v4.0: 4단계 파이프라인. 제안만 함, 직접 수정 절대 없음.

Not recommendedlow grade D
modbender/skill-library-mcp Agent Skills author: modbender MIT 80 files · 22 scripts body ≈ 5 476 tokens Open the sourcegithub.com↗ analyzed 2 d ago

AI 비서가 자기 로그를 분석해서 AGENTS.md 개선안을 제안하는 자동화. v5.0: 시맨틱 임베딩(Ollama nomic-embed-text, FP ~8%), 실시간 스트리밍 모니터(<30초 알림), 플릿 분석(다중 인스턴스). v4.3: 대화형 승인, 멀티포맷 리포트…

As a process F 31/100 · Will not run — References files that are not bundled: scripts/v5/embedding-analyze.sh, scripts/v5/stream-monitor.sh, scripts/v5/fleet-analyzer.sh

IntegrationGitHubSlackDiscordTelegramAI and agentstype and topics are labelled automatically from the skill text
Runs in: OpenClaw Claude / Claude Code ChatGPT / Codex Gemini CLI Cursor GitHub Copilot / VS Code Qwen Code Kimi Code CLI Z.ai (GLM) DeepSeek (Deep Code) Mistral Vibe LangChain (deepagents) Goose OpenCode
JSON
Technical rating
D
55/100
safety, quality, tests
Safety 60%
56
Quality 40%
53
Run on models
none yet
Process rating
F
31/100
Will not run
References files that are not bundled: scripts/v5/embedding-analyze.sh, scripts/v5/stream-monitor.sh, scripts/v5/fleet-analyzer.sh
Tools and files w 18
0
Result and completion w 14
0
Inputs and preconditions w 11
0
the three weakest of ten parameters · all ten

What is at stake

Medium-severity findings: the skill is probably honest, but read what alarmed the scanner.

Dangerous commands medium severity

Below is the worst case for this category. The finding here is medium: the guard saw a sign, not a proof.

If you install

The skill contains commands that delete files, rewrite disks or execute code fetched from the network. The agent may run them without asking if it believes the instructions require it.

For the author

Replace destructive commands with safe equivalents that ask for confirmation, scope them to one folder, and stop piping curl into a shell: pin a version and a checksum.

Exfiltration medium severity

Below is the worst case for this category. The finding here is medium: the guard saw a sign, not a proof.

If you install

The instructions or scripts send data out: environment variables, keys, file contents, chat history. You may never notice, because the agent performs the upload as "part of the task".

For the author

If the upload is not needed for the task, remove it: catalogs flag such skills and delist them. If it is needed, name the destination explicitly, say what leaves the machine, and give the user a switch.

How to improve

  1. Say in the description WHEN to use the skill ("use when…", example requests): that is the agent's main cue.
  2. The SKILL.md body is over 5,000 tokens: move reference detail into references/ and load it when needed.
  3. The text references files that are not there: add them or drop the references.
For the model run — optional
  • Your own cases (evals/evals.json, 4–6 real requests with expected answers): the full check would then run those instead of a model-drafted suite.
  • A spec.yaml with trigger phrases and assertions — a behaviour contract for CI; `skilltest init` writes a template.

Guard findings · 16

✓ No critical or high findings

Medium and low: 16
  • medium Dangerous commands cmd-persistence data/fleet/main-analysis.json:76
    Persistence mechanism (cron / launchd / scheduled task / autorun registry) (string literal in code, not executed)
    "launchctl bootout gui/$(id -u) ~/Library/LaunchAgents/ai.openclaw.watchdog.plist",
    code literal
  • medium Exfiltration net-credential-use docs/feature-roadmap-v4.1.md:450
    Credential used in a network call (verify the destination is the intended service)
    curl -sf "https://api.telegram.org/bot${TELEGRAM_BOT_TOKEN}/sendMessage" \
  • medium Dangerous commands cmd-persistence install/install.sh:59
    Persistence mechanism (cron / launchd / scheduled task / autorun registry) (string literal in code, not executed)
    local plist_dst="$HOME/Library/LaunchAgents/com.sea.monitor.plist"
    code literal
  • medium Dangerous commands cmd-persistence install/install.sh:97
    Persistence mechanism (cron / launchd / scheduled task / autorun registry) (string literal in code, not executed)
    run "launchctl load '$plist_dst'"
    code literal
  • medium Dangerous commands cmd-persistence install/install.sh:170
    Persistence mechanism (cron / launchd / scheduled task / autorun registry) (string literal in code, not executed)
    echo "  macOS: ~/Library/LaunchAgents/com.sea.monitor.plist"
    code literal
  • medium Exfiltration exfil-webhook-url scripts/setup-wizard.sh:311
    Webhook / callback URL commonly used for exfiltration (verify the destination) (detector / deny-list definition)
    if echo "$wh" | grep -q "^https://hooks.slack.com/"; then
    detector
  • medium Exfiltration exfil-webhook-url scripts/setup-wizard.sh:318
    Webhook / callback URL commonly used for exfiltration (verify the destination) (detector / deny-list definition)
    if [ -z "$wh" ] || ! echo "$wh" | grep -q "^https://hooks.slack.com/"; then
    detector
  • low Secrets in code secret-high-entropy-token data/alerts/2026-02-18.json:5
    High-entropy token-like string (may be an id, hash or a credential) (quoted — discussed, not commanded)
    "id": "2026…_ko",
    quoted
  • low Exfiltration exfil-webhook-url docs/feature-roadmap-v4.1.md:450
    Webhook / callback URL commonly used for exfiltration (verify the destination) (placeholder value)
    curl -sf "https://api.telegram.org/bot${TELEGRAM_BOT_TOKEN}/sendMessage" \
    placeholder
  • low Exfiltration exfil-webhook-url README.md:541
    Webhook / callback URL commonly used for exfiltration (verify the destination) (placeholder value)
    webhook_url: "https://hooks.slack.com/services/T.../B.../..."
    placeholder
  • low Exfiltration exfil-webhook-url scripts/setup-wizard.sh:99
    Webhook / callback URL commonly used for exfiltration (verify the destination) (placeholder value)
    echo "  bash $0 --platform slack --webhook https://hooks.slack.com/..."
    placeholder
  • low Exfiltration exfil-webhook-url scripts/setup-wizard.sh:314
    Webhook / callback URL commonly used for exfiltration (verify the destination) (placeholder value)
    echo "   ${RED}❌ https://hooks.slack.com/... 형식이어야 합니다${RESET}"
    placeholder
  • low Exfiltration exfil-webhook-url scripts/setup-wizard.sh:319
    Webhook / callback URL commonly used for exfiltration (verify the destination) (placeholder value)
    echo "${RED}❌ 유효한 Slack Webhook URL 필요: --webhook https://hooks.slack.com/...${RESET}" >&2
    placeholder
  • low Exfiltration exfil-webhook-url scripts/v4/deliver.sh:121
    Webhook / callback URL commonly used for exfiltration (verify the destination) (placeholder value)
    "https://api.telegram.org/bot${TG_TOKEN}/sendMessage" \
    placeholder
  • low Exfiltration exfil-webhook-url scripts/v4/interactive-approve.sh:259
    Webhook / callback URL commonly used for exfiltration (verify the destination) (placeholder value)
    "https://api.telegram.org/bot${tg_token}/sendMessage" \
    placeholder
  • low Exfiltration exfil-webhook-url SKILL.md:615
    Webhook / callback URL commonly used for exfiltration (verify the destination) (placeholder value)
    webhook_url: "https://hooks.slack.com/services/T.../B.../..."
    placeholder

Files scanned: 80. Evidence is masked. Grey chips explain why severity was lowered.

Against the Agent Skills spec

  • warning description-no-when description does not say WHEN to use the skill (no "use when")
  • warning body-long SKILL.md body ≈ 5476 tokens (recommended < 5000); move details to references/
  • warning missing-ref reference to a missing file: scripts/v5/embedding-analyze.sh
  • warning missing-ref reference to a missing file: scripts/v5/stream-monitor.sh
  • warning missing-ref reference to a missing file: scripts/v5/fleet-analyzer.sh
  • warning missing-ref reference to a missing file: scripts/v5/trend-analyzer.sh
  • warning missing-ref reference to a missing file: scripts/v5/orchestrator.sh

Process rating: all ten parameters 31/100

Will not run. References files that are not bundled: scripts/v5/embedding-analyze.sh, scripts/v5/stream-monitor.sh, scripts/v5/fleet-analyzer.sh
  • 0Tools and files. 5 referenced file(s) missing: scripts/v5/embedding-analyze.sh, scripts/v5/stream-monitor.sh, scripts/v5/fleet-analyzer.sh
  • 0Result and completion. Does not say what the result is
  • 0Inputs and preconditions. Does not say what the process needs to start
  • 0Failures and branches. Linear process with no failure handling
  • 0Progress reporting. Says nothing while it works
  • 20When it triggers. No condition that starts the skill
  • 30Running it twice. 2 mutating operations with no state check
  • 70Execution cost. Instruction body is 5476 tokens
  • 100Steps. 98 steps
  • 100Consistency. Name and required fields are in place
  • low 21 top-level sections: this looks like several domains in one skill

Everything here is measured from the skill text rather than judged by a model, so the numbers are checkable. A parameter weighs more when it is a more common reason for the process to stall.

Quality signals

  • +5Description has no quoted example phrases that should trigger the skill
  • +4Description does not say when NOT to use the skill (false activations)
  • +3Output format is not stated: the model decides each time
  • -2localhost URLs: will not work for another user
  • -294 emoji in the instructions: noise for the model
  • +1No license
  • +2Single-language instructions
  • +3Description length 240: enough signal without eating the budget
  • +4Structure: 45 headings
  • +3Step-by-step instructions: 98 items
  • +4Has examples (33 code blocks)
  • +4Reference files are cited in the instructions (8 of 24)
  • +3All 4 scripts are documented

Quality base 70; lint remarks subtract, signals add up to 100. Result: 53.