reflection_skill

This skill guides structured self-evaluation before delivering work, clarifying quality gaps and boosting trust through multi-lens critique.
  • Python

2.5k

GitHub Stars

7

Bundled Files

2 months ago

Catalog Refreshed

4 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill openclaw/skills --skill reflection

  • _meta.json460 B
  • dimensions.md1.6 KB
  • memory-template.md1.5 KB
  • prompts.md1.3 KB
  • reflections.md241 B
  • setup.md1.5 KB
  • SKILL.md9.1 KB

Overview

This skill guides a short, structured self-evaluation before delivering work to catch blind spots and improve quality. It frames review as a multi-lens critique—correctness, completeness, clarity, and robustness—plus a steel-man objection and an honest confidence rating. Use it to increase trust in outputs and reduce rework by finding issues early.

How this skill works

Pause after completing work and create mental distance to review it as if it belonged to someone else. Run through the core lenses: Does it solve the stated problem, what’s missing, would others understand it, and what might break. Produce either a deliver-with-confidence statement, a plan to improve then deliver, or a flagged uncertainty with a numeric confidence score and next steps.

When to use it

  • Before handing off high-stakes or client-facing deliverables
  • After long focused sessions when tunnel vision may hide errors
  • When uncertain about the solution or its edge cases
  • For complex artifacts like architecture, strategy, or production code
  • When quality issues would be costly to fix later

Best practices

  • Take a brief mental step back—don’t edit while still in creator mode
  • Apply all four lenses for standard tasks; use one lens for quick checks
  • Steel-man the strongest objection and try to address it concretely
  • Give an honest 1–10 confidence rating and list what would raise it
  • Document any unresolved risks and communicate them when delivering

Example use cases

  • Final review of a client proposal to ensure completeness and clarity
  • Pre-merge check of a production code change to spot edge cases
  • Assessment of a system design to identify robustness gaps
  • Quick sanity check after drafting a policy or analysis report
  • Post-sprint review to decide whether work is ready for release

FAQ

Tailor depth to risk: ~10s for quick answers, ~30s for standard tasks, ~2 minutes for critical deliveries.

What if I still feel unsure after reflecting?

Flag the uncertainty with a confidence score, list what’s needed to improve it, and ask for a targeted review or extra testing.

Built by
VeilStrat
AI signals for GTM teams
© 2026 VeilStrat. All rights reserved.All systems operational