Analyze JSONL event logs and propose reviewer-gated SKILL.md patches.
npx skills add https://github.com/duypham2801/ThS_LLM --skill meta-optimize-duypham2801
Please help me install this Agent Skill. Skill: meta-optimize Source: https://github.com/duypham2801/ThS_LLM/tree/main/.claude/skills/meta-optimize Command: npx skills add https://github.com/duypham2801/ThS_LLM --skill meta-optimize-duypham2801
ARIS skill prompts, defaults, and workflow gates can drift away from what users actually do, causing unnecessary manual overrides, retries, and inconsistent outcomes.
.aris/meta/events.jsonl to find recurring friction points in reviewer prompts, loop thresholds, retries, and workflow ordering.SKILL.md diffs, runs cross-model adversarial review, and never auto-applies changes.Run the meta-optimize skill after at least five logged workflow sessions so it can analyze .aris/meta/events.jsonl and recommend safe SKILL.md improvements.
High-intent search queries and answers about installing and using this skill.
To optimize skill prompts using execution logs, you analyze accumulated JSONL event logs to identify recurring friction points, then propose minimal, reviewer-gated diff patches to SKILL.md and workflow defaults based on log-backed evidence.
You need at least five logged workflow sessions before tuning workflow defaults. This ensures the log analysis has sufficient accumulated execution data to perform accurate frequency, failure, and convergence analyses for targeted improvements.
No, log-driven skill optimization never auto-applies SKILL.md patches. It runs cross-model adversarial review on proposed diff-style patches and applies changes only after receiving explicit user approval during the maintenance workflow.
The best way to analyze ARIS harness friction from JSONL logs is performing frequency, failure, and convergence analyses on events.jsonl. This identifies recurring issues in reviewer prompts, loop thresholds, retries, and workflow ordering for targeted optimization.
Yes, you can target a specific skill for prompt tuning instead of all skills. The optimization scope focuses on a specified skill if it has sufficient logged invocations within the accumulated event logs.
You should not use automated prompt tuning for workflow defaults when you have fewer than five logged workflow sessions, or when you need immediate auto-applied changes, as this process requires explicit reviewer approval and sufficient log data.
Shallow-WAnalyze ARIS usage logs and propose evidence-backed SKILL.md patches.
lightrain-aAnalyze ARIS usage logs to propose ranked optimizations for harness components.
goupup-aiAnalyze ARIS usage logs to propose evidence-backed SKILL.md patch improvements.
estevanhernandez-stack-edFilter and group friction log signals into a banner-style report.