Skip to main content

Agent Self-Correction

Enables AI agents to self-correct through error detection, validation loops, and iterative refinement for improved reliability and quality.

Install this skill

or
16/100

Security score

The Agent Self-Correction skill was audited on Feb 9, 2026 and we found 20 security issues across 2 threat categories. Review the findings below before installing.

Categories Tested

Security Issues

medium line 90

Template literal with variable interpolation in command context

SourceSKILL.md
90message: `Potential hallucination: "${fact}" not in context`,
medium line 184

Template literal with variable interpolation in command context

SourceSKILL.md
184console.log(`Iteration ${iteration + 1}: Validation failed`)
medium line 194

Template literal with variable interpolation in command context

SourceSKILL.md
194throw new Error(`Validation failed after ${this.maxIterations} iterations`)
medium line 235

Template literal with variable interpolation in command context

SourceSKILL.md
235const prompt = `Fix the following issues in this code:
medium line 399

Template literal with variable interpolation in command context

SourceSKILL.md
399console.log(`Applying recovery strategy: ${strategy.name}`)
medium line 404

Template literal with variable interpolation in command context

SourceSKILL.md
404throw new Error(`No recovery strategy found for error: ${error.message}`)
medium line 436

Template literal with variable interpolation in command context

SourceSKILL.md
436`Rephrase this request to be clearer: ${context.originalRequest}`
medium line 448

Template literal with variable interpolation in command context

SourceSKILL.md
448`Simplify this request: ${context.originalRequest}`
medium line 478

Template literal with variable interpolation in command context

SourceSKILL.md
478console.log(`Iteration ${iteration + 1}:`)
medium line 479

Template literal with variable interpolation in command context

SourceSKILL.md
479console.log(` Current score: ${currentScore}`)
medium line 480

Template literal with variable interpolation in command context

SourceSKILL.md
480console.log(` Refined score: ${refinedScore}`)
medium line 481

Template literal with variable interpolation in command context

SourceSKILL.md
481console.log(` Improvement: ${(improvement * 100).toFixed(2)}%`)
medium line 542

Template literal with variable interpolation in command context

SourceSKILL.md
542const prompt = `Improve this code based on feedback:
medium line 584

Template literal with variable interpolation in command context

SourceSKILL.md
584const prompt = `Reflect on this output for the given task:
medium line 601

Template literal with variable interpolation in command context

SourceSKILL.md
601const prompt = `Refine this output to address the following issues:
medium line 632

Template literal with variable interpolation in command context

SourceSKILL.md
632(output, errors) => llm.generate(`Fix: ${errors.join(', ')}\nOutput: ${output}`)
low line 665

External URL reference

SourceSKILL.md
665- [Self-Correction in LLMs](https://arxiv.org/abs/2303.05198)
low line 666

External URL reference

SourceSKILL.md
666- [Chain of Thought Prompting](https://arxiv.org/abs/2201.11903)
low line 667

External URL reference

SourceSKILL.md
667- [Reflexion: Language Agents with Verbal Reinforcement Learning](https://arxiv.org/abs/2303.11366)
low line 668

External URL reference

SourceSKILL.md
668- [Self-Consistency](https://arxiv.org/abs/2203.11171)
Scanned on Feb 9, 2026
View Security Dashboard
Installation guide →