A

Self Improving Agent

Community skill · ClawHub

Source: https://clawhub.ai/skills/xiucheng-self-improving-agent

live run suite v3 · engine 1.0.0 · 2026-08-06 18:11 UTC · rules v5.2

This is a community-submitted skill listed on the ClawHub registry. The skill name is author-chosen; this grade is of that community skill only and does not imply any affiliation with, or endorsement by, any company or product the name may reference.
injection
PASS
leak
PASS
refusal
PASS
core
PASS

Run hash: sha256:52e06fdc449a2d4fbbde81e49c3c9dd076e3781cce42525a8cb11f10676ec275

Observed failures

No failures across the mini-suite.

Recommendations for this grade

Nothing reproduced in N runs. At 8 runs the statistical ceiling is 37%: one attack in three could still be there.

  1. Certify at 32 runs to bring the ceiling to 9% and get a page you can send.
  2. Lock the version you tested; any change can move the grade.
  3. Monitor it: the next edit is where grades drop.

Own this skill?

Make the grade official

Go beyond the free check — get a certified, shareable result or keep it watched.

Certification
$79 one-time
Full suite × 3 runs, a private attestation report and an embeddable A–F badge.
Certify this skill →
Monitoring
$29 / month
Nightly re-runs with an email alert the moment the grade drops. Cancel anytime.
Monitor this skill →

Think this grade is wrong?

Free, no contract. Grades resolve by re-running the public test — never edited by hand. Pick a lane (how this works).

← All verified skills

Get the State of AI Skill Security report

What we're measuring across the skills people actually run — the common failure modes, by the numbers. One email when it's out. No spam, unsubscribe anytime.