skills.rest
anajuliabit
anajuliabitCOMMUNITY
Explore all skills by this author →
𝕏@anajuliabitGitHub

self-improve

Analyze recent agent failures and propose targeted configuration, prompt, and workflow fixes.

6|2|Updated May 21, 2026
Search Tags:#workflow optimization#performance monitoring#failure triage#prompt tuning#agent self-improvement#GitHub pull requests
One-click install
npx skills add https://github.com/anajuliabit/aeon --skill self-improve-anajuliabit
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: self-improve
Source: https://github.com/anajuliabit/aeon/tree/main/skills/self-improve
Command: npx skills add https://github.com/anajuliabit/aeon --skill self-improve-anajuliabit

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill reduces recurring agent failures and quality degradation by identifying what went wrong and proposing the smallest targeted improvements based on recent performance signals.

Core Features & Use Cases

  • Single high-impact fix per run: Selects the most valuable “smallest effort” change from the last two days of logs and recent repo actions to avoid compounding technical debt.
  • Focused, evidence-backed PRs: Creates an improvement branch and PR with a clear Problem/Fix/Evidence structure derived from memory and logs.
  • Operational guardrails: Avoids creating too many open improvement PRs, and prevents risky scope like modifying workflows or secrets.

Quick Start

Ask your agent: self-improve with var set to "notifications" to focus on improving broken or truncated notification behavior from recent logs.

Frequently Asked Questions about self-improve

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

FAQPage Schema
How do I automate agent self-improvement for recurring workflow failures?▼

Agent self-improvement analyzes recent logs and memory to identify recurring failures, then creates an evidence-backed pull request proposing the smallest high-impact configuration or prompt fix.

What is the best way to triage autonomous agent failures from unattended scheduled runs?▼

Triage autonomous agent failures by scanning cron success rates and recent repo actions over the last two days to select the highest-impact, smallest-effort configuration or prompt change.

How do I fix degraded agent performance without accumulating technical debt?▼

Fix degraded agent performance by applying operational guardrails that restrict changes to a single high-impact fix per run, avoiding risky scope like modifying workflows or secrets.

Can I target workflow optimization for specific issues like broken notifications?▼

Yes, you can target workflow optimization for specific issues by setting a variable to focus on broken notification behavior, analyzing recent logs to propose targeted prompt and configuration fixes.

How does failure triage generate evidence-backed pull requests for prompt tuning?▼

Failure triage generates evidence-backed pull requests by creating an improvement branch with a clear Problem/Fix/Evidence structure derived from analyzing recent memory and logs.

What are the limitations of using autonomous self-improvement for GitHub pull requests?▼

Limitations include avoiding modifications to risky scope like workflows or secrets, and preventing the creation of too many open improvement pull requests to manage technical debt.

Related Skills

Explore More Software Engineering Skills→
anajuliabitanajuliabit
6

Weekly Reflect

Consolidate agent memory by reviewing logs and pruning stale entries.

Community
Intermediate
vTRKAvTRKA

code-review

Assesses GitHub pull requests for defects and produces scored, actionable verdicts.

Community
Advanced
cyzuscyzus
62

automation

Schedule cron-based tasks and persistent heartbeat monitoring for agent prompts.

Community
Advanced
insightwavesmithinsightwavesmith

code-analyzer

Analyze code changes to identify those meeting evidence-shape-check criteria.

Community
Intermediate