agent-orchestration-improve-agent

Systematic improvement of existing agents through performance analysis, prompt engineering, and continuous iteration.

Install with Tessl CLI

npx tessl i github:sickn33/antigravity-awesome-skills --skill agent-orchestration-improve-agent

What are skills?

1.24x

Quality

43%

Does it follow best practices?

Impact

81%

1.24x

Average score across 3 eval scenarios

Optimize this skill with Tessl

npx tessl skill review --optimize ./skills/agent-orchestration-improve-agent/SKILL.md

SKILL.md

Review

Evals

Evaluation results

70%

20%

Diagnose a Struggling AI Coding Assistant

Performance baseline analysis

Criteria

Without context

With context

context-manager command

100%

Task completion rate metric

83%

100%

Corrections per task metric

66%

100%

Tool call efficiency metric

User satisfaction score

100%

Latency and token metrics

All six failure categories

53%

46%

All five feedback patterns

60%

100%

Hallucination incidents tracked

100%

Baseline report section

37%

50%

Failure mode examples from logs

100%

No deployment recommendations

83%

Without context: $0.5408 · 2m 47s · 14 turns · 63 in / 9,280 out tokens

With context: $0.6222 · 2m 46s · 21 turns · 328 in / 9,345 out tokens

88%

10%

Redesign the System Prompt for a Legal Document Assistant

Prompt engineering improvements

Criteria

Without context

With context

Step-by-step phrasing

100%

Verification checkpoint phrasing

100%

Both positive and negative examples

25%

100%

Few-shot example structure

50%

100%

Core purpose statement

100%

Constraints section

100%

Success criteria defined

60%

80%

Tool proficiency defined

60%

Five constitutional principles

73%

80%

Critique-and-revise loop

70%

80%

Structured output templates

100%

Examples ordered simply to complex

80%

prompt-engineer technique named

100%

Without context: $0.3691 · 2m 14s · 12 turns · 12 in / 6,633 out tokens

With context: $0.5747 · 2m 48s · 23 turns · 19 in / 8,305 out tokens

86%

18%

Plan the Production Launch of an Improved Customer Support Agent

Deployment and version control planning

Criteria

Without context

With context

Version format used

100%

MINOR vs MAJOR distinction

80%

100%

Git-based storage

100%

Changelog requirement

100%

Six test suite categories

60%

100%

parallel-test-runner referenced

100-task minimum per variant

60%

100%

Statistical significance threshold

100%

40%

Cohen's d effect size

100%

Five-stage rollout

80%

90%

Rollback triggers present

37%

100%

Success criteria thresholds

12%

50%

30-day review plan

42%

100%

Regression testing before deploy

100%

Rollback process steps

87%

100%

Without context: $0.3873 · 2m 29s · 11 turns · 11 in / 7,709 out tokens

With context: $0.5933 · 3m 27s · 15 turns · 16 in / 10,795 out tokens

Evaluated: 8 days ago
Agent: Claude Code
Model: Claude Sonnet 4.6

Table of Contents

Diagnose a Struggling AI Coding Assistant Redesign the System Prompt for a Legal Document Assistant Plan the Production Launch of an Improved Customer Support Agent

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.