← Back to Skills Marketplace
155
Downloads
1
Stars
2
Active Installs
9
Versions
Install in OpenClaw
/install skill-insight
Description
Understand what your AI agent's skills are actually being used for — with usage reports, success/failure tracking, and unused-skill recommendations. Part of...
Usage Guidance
This skill is a local analyzer/reporting tool; it will create and update data files (data/usage.json and data/skill_registry.json) in the skill directory (by default ~/.openclaw/workspace/skills/skill-insight/data). If you enable automatic scanning (cron or run scan_sessions.py) it will read recent OpenClaw session history via the local gateway API or the openclaw CLI to detect skill mentions. It does not transmit data off your machine and it does not require API keys, but it will read session messages in-memory and record auto-detection metadata (session identifiers, timestamps, and detection notes) into usage.json. Before installing or enabling the cron job: review and confirm the data directory location (use SKILL_USAGE_TRACKER_DIR to override), inspect the cron entry, and verify your OpenClaw gateway is bound to localhost (not exposed externally). If you prefer manual control, use the record.sh/record.py commands instead of enabling the cron/scanner. If you want extra assurance, run the scripts manually once to inspect the files they create and the exact contents written to usage.json and skill_registry.json.
Capability Analysis
Type: OpenClaw Skill
Name: skill-insight
Version: 1.0.8
The skill bundle is a utility designed to track and analyze the usage frequency and success rates of an AI agent's skills. It operates by maintaining local JSON databases (`usage.json`, `skill_registry.json`) and includes a session scanner (`scan_sessions.py`) that interacts with the local OpenClaw gateway (localhost:18789) or CLI to identify skill invocations. No evidence of data exfiltration, external network calls, or malicious prompt injection was found; the instructions in SKILL.md appropriately guide the agent to log its own activity for reporting purposes.
Capability Assessment
Purpose & Capability
The skill claims to analyze existing usage data and recommend unused skills; the repository provides registry, recording, reporting, analysis, and optional session-scanning scripts that match that purpose. The components requested/used (local files, optional cron, local gateway/CLI) are proportionate to the stated purpose.
Instruction Scope
Instructions are generally scoped to recording and analyzing local usage data and optionally scanning recent sessions. One potential clarity issue: the SKILL.md initially states "It does not automatically collect data for you," yet it documents an optional cron scan and a session-scanner that can automatically detect script-type invocations if the user enables them. The session scanner reads local session messages (via the local gateway API or openclaw CLI) in-memory to detect skill mentions; it does not persist message contents but does record session identifiers and auto-detected invocation metadata into usage.json.
Install Mechanism
There is no remote install or download step; all code is included. Risk is low because it does not fetch or execute remote archives. The scripts run locally and the recommended cron entry is a standard wrapper that invokes included scripts.
Credentials
The skill declares no required credentials. It optionally reads the user's ~/.openclaw/openclaw.json to discover the gateway port and makes HTTP calls to localhost and/or invokes the local openclaw CLI. This is coherent for session scanning but is a scope of access the user should be aware of (it must be able to read session lists/messages via the local gateway or CLI). An optional SKILL_USAGE_TRACKER_DIR environment variable can change where data is read/written.
Persistence & Privilege
The skill does not request permanent platform-wide privileges (always:false). It creates/uses a local data directory under the skill root (or ~/.openclaw/workspace/skills/skill-insight) and writes usage and registry JSON files. It does not modify other skills' configs or agent settings.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install skill-insight - After installation, invoke the skill by name or use
/skill-insight - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.8
v1.0.8 - Complete rewrite: honest scope (analysis tool, not auto-tracker), Quick Setup 3-step guide, access_type coverage table, route-type setup instructions, known limitations stated upfront. v1.0.7 - Category translated + outcome tracking. v1.0.6 - Period labels translated. v1.0.5 - Full i18n. v1.0.0 - Initial release.
v1.0.7
v1.0.7 - Category headers translated (AGENT→智能体/DEVELOPMENT→开发/etc), outcome tracking workflow documented in SKILL.md + HEARTBEAT.md. v1.0.6 - Period labels translated. v1.0.5 - Full i18n. v1.0.4 - Output Language section. v1.0.3 - Full English, remove KNOWN_SCRIPTS. v1.0.2 - Fix access_type header. v1.0.1 - Fix sys.path/closure/naming/logpath. v1.0.0 - Initial release.
v1.0.6
v1.0.6 - Fix: period labels now translated (本周/今日/本月/全部时间). v1.0.5 - Full i18n support. v1.0.4 - Add Output Language section. v1.0.3 - Full English output, remove KNOWN_SCRIPTS. v1.0.2 - Fix access_type Chinese header. v1.0.1 - Fix sys.path/closure/naming/logpath bugs. v1.0.0 - Initial release.
v1.0.5
v1.0.5 - Full i18n: --lang en/zh flag, all output labels translated, agent auto-passes language. v1.0.4 - Add Output Language section. v1.0.3 - Full English output, remove KNOWN_SCRIPTS. v1.0.2 - Fix access_type Chinese header. v1.0.1 - Fix sys.path/closure/naming/logpath bugs. v1.0.0 - Initial release.
v1.0.4
v1.0.4 - Add Output Language section to SKILL.md (English-only output, i18n considered for future). v1.0.3 - Full English output, remove KNOWN_SCRIPTS. v1.0.2 - Fix access_type Chinese header. v1.0.1 - Fix sys.path/closure/naming/logpath bugs. v1.0.0 - Initial release.
v1.0.3
v1.0.3 - Full English output, remove KNOWN_SCRIPTS (generic detection only). v1.0.2 - Fix access_type Chinese header. v1.0.1 - Fix sys.path/closure/naming/logpath bugs. v1.0.0 - Initial release.
v1.0.2
v1.0.2 - Fix access_type Chinese header → English. v1.0.1 - Fix sys.path/closure/naming/logpath bugs. v1.0.0 - Initial release.
v1.0.1
v1.0.1 - Fix sys.path import bug (Path(__file__).parent), fix scan_sessions closure bug (pass valid_skills as param), fix naming consistency (all skill-insight), fix cron log path (/var/log/ → ~/.local/log/), fix extract_skills_from_content empty set. v1.0.0 - Initial release.
v1.0.0
v1.0.0 - Initial release: skill registry, usage logging with success/failure tracking, multi-period reports (today/week/month/all), unused skill analysis with access_type classification (route/tool/script), auto-scan sessions, path-agnostic installation, user data isolation via data/ directory + .clawhubignore
Metadata
Frequently Asked Questions
What is Skill Insight?
Understand what your AI agent's skills are actually being used for — with usage reports, success/failure tracking, and unused-skill recommendations. Part of... It is an AI Agent Skill for Claude Code / OpenClaw, with 155 downloads so far.
How do I install Skill Insight?
Run "/install skill-insight" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Skill Insight free?
Yes, Skill Insight is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Skill Insight support?
Skill Insight is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Skill Insight?
It is built and maintained by YY (@yingy4); the current version is v1.0.8.
More Skills