Skip to content
Back to skills

Dig Deeper Probe

ASecurity

Diagnose nasty, deep bugs when observation is not enough. Use when tools, safe reproduction and controlled experiments are needed to find the cause. Tool installation follows the user's consent or established execution boundary. Diagnosis only, never remediation.

  • 5 stars
  • 0 votes
  • 0 copies
  • 1 view
  • Added September 19, 2026
ai-agentsrustsecurity

Security analysis

A100/100

Pro scans all 2 files and shows the line behind each finding

Scanned September 19, 2026

npx -y skills add simonasrazm/skills --skill dig-deeper-probe --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Dig Deeper Probe?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Dig Deeper Probe
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/simonasrazm-dig-deeper-probe/badge)](https://www.skillsdirectory.com/skills/simonasrazm-dig-deeper-probe)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: dig-deeper-probe
description: Diagnose nasty, deep bugs when observation is not enough. Use when tools, safe reproduction and controlled experiments are needed to find the cause. Tool installation follows the user's consent or established execution boundary. Diagnosis only, never remediation.
---

# Dig Deeper Probe

Run `dig-deeper-core` with:

```markdown
profile: HIGH_TRUST_ACTIVE
objective: gather evidence needed for diagnosis
allow: any useful method; acquire missing capabilities; reproduce/simulate/experiment; controlled target interaction
choose: rank available methods by expected discrimination and effect; run best permitted first; if blocked, unblock/acquire within boundary; weaker fallback only if best remains unavailable or insufficient
computer: first only when rendered/manual state is itself evidence
damage: avoid destructive, irreversible, security/privacy-compromising, or recklessly high-blast-radius action; isolate risk when practical; stop if meaningful damage cannot be avoided
authorization: acquire or install tools within the user's explicit consent or established execution boundary; when neither covers the proposed installation, explain the tool, purpose and scope of changes and obtain consent first; honor runtime approval requirements
remediate: no; evidence collection != disguised fix
```

Pass problem + profile. Core owns loop/report. Preserve independent roles.

Files in this skill

  • SKILL.md1.4 KB
  • agents/openai.yaml46 B

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…