Skills self-improvement

📦

self-improvement

Name: self-improvement
Author: peterskoett

Content revision r1 High Risk ⚙️ External commands📁 Filesystem access🌐 Network access

Capture Agent Learnings and Errors

AI coding sessions can lose useful corrections, error fixes, and workflow discoveries. This skill records them in structured logs and helps teams promote durable guidance.

Supports: Claude Codex Code(CC)

⚠️ 38 Poor

Stop for confirmation before installing.

Review the plan and obtain explicit user consent before changing files.

Install with my Agent

Copy this request to your Agent. It includes the canonical Skill page and manifest.

Agent request

Review the Skillstore skill "self-improvement" from https://skillstore.io/skills/peterskoett-self-improvement.md and its manifest at https://skillstore.io/api/skills/peterskoett-self-improvement/manifest. Verify the artifact. Stop and obtain explicit user consent before installing or changing files.

Your Agent should still show its plan and request any confirmation required by the security policy.

Agent-readable resources

Use these links when an AI agent, crawler, or script needs clean context instead of reading the full page.

Markdown detail GET /skills/peterskoett-self-improvement.md Signed manifest GET /api/skills/peterskoett-self-improvement/manifest Signed lockfile GET /api/skills/peterskoett-self-improvement/lockfile

Test it

Using "self-improvement". A package install failed because the project uses a different package manager.

Expected outcome:

A learning entry notes the mistaken assumption, the correct package manager, related lock files, and a future check.

Using "self-improvement". A user asks for a missing export feature during a report workflow.

Expected outcome:

A feature request entry records the requested capability, user context, estimated complexity, and possible implementation path.

Using "self-improvement". Several sessions repeat the same build mistake after API changes.

Expected outcome:

A promoted guidance draft explains the required post-change step and links back to the resolved learning entries.

Security Audit

High Risk

v5 • 7/5/2026 Open versioned report

The static alerts are false positives in context: they are mostly Markdown backticks, documentation URLs, home-directory setup examples, and secret-avoidance text. No evidence of malware, credential exfiltration, or unauthorized runtime network activity was found. The main residual concern is durable prompt-memory poisoning through promotion of untrusted learnings into future agent instruction files.

Files scanned

2,300

Lines analyzed

Review items

False positives ignored

Confirmed security concerns (1)

High

SKILL.md:40-43 SKILL.md:293-310 SKILL.md:366-380 SKILL.md:460-468

Persistent Agent Instruction Poisoning Risk

The skill encourages agents to promote learnings into CLAUDE.md, AGENTS.md, .github/copilot-instructions.md, SOUL.md, and TOOLS.md. Because those files can become future agent context, untrusted or poisoned learning content could persist as durable instructions without a required review gate.

The cited lines explicitly name future prompt/context files as promotion targets and recommend aggressive promotion. The risk depends on later review quality, so confidence is high but not absolute.

Risk Factors

⚙️ External commands (199)

📁 Filesystem access (23)

🌐 Network access (3)

SKILL.md:58 SKILL.md:61 SKILL.md:581

Audited by: codex View Audit History →

Share & cite this report

Share the versioned assessment report, neutral badge, embed card, and citations. Skillstore reports evidence without deciding whether this Skill is safe.

Open versioned report

Copy report link

https://skillstore.io/skills/peterskoett-self-improvement/audits/5?utm_source=security_passport&utm_medium=share&utm_campaign=versioned_report

Markdown badge

[![Skillstore security assessment](https://skillstore.io/badges/skills/peterskoett-self-improvement/security.svg)](https://skillstore.io/skills/peterskoett-self-improvement?utm_source=security_passport_badge)

HTML badge

<a href="https://skillstore.io/skills/peterskoett-self-improvement?utm_source=security_passport_badge"><img src="https://skillstore.io/badges/skills/peterskoett-self-improvement/security.svg" alt="Skillstore security assessment" loading="lazy"></a>

Embed card

<iframe src="https://skillstore.io/embed/skills/peterskoett-self-improvement.html" title="Skillstore Security Assessment" sandbox="allow-popups allow-popups-to-escape-sandbox" loading="lazy" referrerpolicy="no-referrer" width="420" height="180"></iframe>

Academic citations (APA · BibTeX · CFF)

APA citation

peterskoett. (2026). self-improvement security audit report (audit version 5) [Author version unspecified]. Skillstore. https://skillstore.io/skills/peterskoett-self-improvement/audits/5

BibTeX citation

@techreport{peterskoett-peterskoett-self-improvement-2026,
  author = {peterskoett},
  title = {self-improvement security audit report (audit version 5)},
  institution = {Skillstore},
  year = {2026},
  number = {5},
  url = {https://skillstore.io/skills/peterskoett-self-improvement/audits/5},
  note = {Author version unspecified}
}

CITATION.cff

cff-version: 1.2.0
message: "If you use this Skill, cite its author and this versioned security audit report."
title: "self-improvement security audit report (audit version 5)"
version: "unspecified"
type: report
authors:
  - name: "peterskoett"
date-released: "2026-07-05"
url: "https://skillstore.io/skills/peterskoett-self-improvement/audits/5"
identifiers:
  - type: other
    value: "skillstore:peterskoett-self-improvement:audit:5"
    description: "Skillstore immutable audit report identifier"

Skillstore Score

Why this score Evidence Confidence: High

Architecture

Maintainability

Content

Community

Spec Compliance

What You Can Build

Track debugging discoveries

Record non-obvious command failures, fixes, and project conventions after a debugging session.

Share recurring workflow rules

Promote repeated lessons into shared agent guidance after review and consensus.

Improve agent handoffs

Capture patterns that make future Claude, Codex, and Claude Code sessions more consistent.

Try These Prompts

Initialize learning logs

Create the .learnings files for this workspace if they are missing. Do not overwrite existing content.

Log a corrected assumption

Log the correction from this conversation as a learning. Include the wrong assumption, the corrected rule, and related files.

Review unresolved learnings

Review the current learning logs for pending high-priority entries related to this task. Summarize only relevant action items.

Promote mature guidance

Find recurring resolved learnings that deserve promotion. Draft concise guidance for the correct instruction file and flag security concerns.

Best Practices

Log only the useful lesson, not full transcripts or raw secret-bearing output.
Review entries before promotion into persistent agent instruction files.
Use stable IDs, related files, and status fields so future agents can filter entries quickly.

Avoid

Do not store credentials, private keys, tokens, or full environment dumps in learning logs.
Do not promote untrusted user text directly into agent instructions.
Do not enable command-output hooks unless the workspace can handle sensitive output safely.