skills.rest
Vimurai
VimuraiCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

ai-insights

Generates a meta-cognition report from cross-project telemetry stored in a SQLite database.

Updated Feb 26, 2026
Search Tags:#monitoring#agent workflow#sqlite#telemetry#meta-analysis#cli automation#insights report
One-click install
npx skills add https://github.com/Vimurai/ai-os --skill ai-insights-vimurai
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: ai-insights
Source: https://github.com/Vimurai/ai-os/tree/main/.claude/skills/ai-insights
Command: npx skills add https://github.com/Vimurai/ai-os --skill ai-insights-vimurai

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Engineering teams running AI agent workflows accumulate telemetry across projects but lack a consolidated view of which tools fail, which are slow, and which CLI interactions could be automated. This Skill triggers the meta_analyst agent to turn raw telemetry into an actionable INSIGHTS.md report. ## Core Features & Use Cases - Telemetry Analysis: Reads cross-project telemetry from ~/.ai-os/telemetry.sqlite and produces optimization recommendations covering CLI automation candidates, tool deprecation candidates, and latency hardening candidates. - Staleness-Aware Triggering: Invoked on demand or when the ai-preflight staleness check emits [INSIGHTS_STALE], keeping meta-analysis out of the hot chat loop. - Guarded Execution: Verifies telemetry reachability, handles empty databases with [INSIGHTS_EMPTY], and stamps each run via task-synchronizer so report cycles correlate with sprint deltas. - Use Case: Before a sprint planning session, run this Skill to refresh INSIGHTS.md and surface which tools are failing or slow, so planning decisions reflect actual usage data. ## Quick Start Ask the AI to regenerate the AI-OS insights report from telemetry before starting the planning session.

Frequently Asked Questions about ai-insights

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I generate an insights report from AI agent telemetry?▼

Invoke this Skill to activate the meta_analyst agent, which runs five canonical aggregate queries against ~/.ai-os/telemetry.sqlite and writes recommendations to ~/.ai-os/INSIGHTS.md. The Skill verifies telemetry reachability first and stamps the run for sprint correlation.

When should I run the telemetry insights report?▼

Run it when ai-preflight emits [INSIGHTS_STALE] at session start, before a sprint planning session, or during monthly hygiene of ~/.ai-os/. Avoid running it mid-implementation or during an active debug cycle, since the analysis is noisy then.

What happens when the telemetry database is empty?▼

The Skill checks telemetry stats first; if the tool_executions count is zero, it writes [INSIGHTS_EMPTY] to INSIGHTS.md and exits without invoking the analyst agent. There is nothing to analyze until telemetry accumulates.

Can I query the telemetry SQLite database directly instead?▼

No. The meta_analyst agent enforces a read-only, aggregate-only contract that scrubs tool_name and session_id correlations. Direct ad-hoc queries bypass these privacy safeguards and are explicitly prohibited.

How do I reset or roll back the insights report?▼

Delete ~/.ai-os/INSIGHTS.md and the next invocation regenerates it from scratch; the telemetry database is untouched. Setting AI_TELEMETRY_DISABLE=1 pauses data collection, and the agent will emit [INSIGHTS_PAUSED].

Related Skills

Explore More Software Engineering Skills→
googlegoogle
68

dev-mcp

Bridge AI agents in a Capsem guest VM to external MCP servers.

Official
Advanced
bmad-code-orgbmad-code-org
52.5k

bmad-sprint-status

Forwards deprecated sprint status invocations to bmad-sprint-planning's status view.

Official
Basic
HugoRCDHugoRCD
1.8k

telemetry

Answers questions about evlog CLI production usage by querying the telemetry dashboard's MCP tools.

Community
Intermediate
wenermewenerme
7

openobserve-docs

Guide users to OpenObserve documentation for setup, ingestion, querying, and operations.

Community
Intermediate