ClawHub

Think Block Leakage

v1.0.0

Internal reasoning from <think> blocks leaks into the final user-facing reply instead of being stripped.

⭐ 0· 44·0 current·0 all-time
by@mvogt99

Install

OpenClaw Prompt Flow

Install with OpenClaw

Best for remote or guided setup. Copy the exact prompt, then paste it into OpenClaw for mvogt99/think-block-leakage.

Previewing Install & Setup.
Prompt PreviewInstall & Setup
Install the skill "Think Block Leakage" (mvogt99/think-block-leakage) from ClawHub.
Skill page: https://clawhub.ai/mvogt99/think-block-leakage
Keep the work scoped to this skill only.
After install, inspect the skill metadata and help me finish setup.
Use only the metadata you can verify from ClawHub; do not invent missing requirements.
Ask before making any broader environment changes.

Command Line

CLI Commands

Use the direct CLI path if you want to install manually and keep every step visible.

OpenClaw CLI

Bare skill slug

openclaw skills install think-block-leakage

ClawHub CLI

Package manager switcher

npx clawhub@latest install think-block-leakage
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Benign
high confidence
✓
Purpose & Capability
Name/description claim (detect and strip leaked internal reasoning) matches the SKILL.md content. The skill requires no binaries, env vars, or configs that would be unexpected for this purpose.
✓
Instruction Scope
The runtime instructions focus on inspecting raw LLM output for unmatched reasoning tags, redacting or regenerating replies, trimming planning preambles, and tightening prompts. They do not instruct the agent to read unrelated files, transmit data externally, or access unrelated credentials.
✓
Install Mechanism
No install spec or code files are present (instruction-only), so nothing will be written to disk or downloaded. This is the lowest-risk install posture.
✓
Credentials
No environment variables, credentials, or config paths are requested. The skill's operations do not require elevated access or external tokens.
✓
Persistence & Privilege
always is false and the skill is user-invocable; it does not request persistent or cross-skill configuration changes. Autonomous invocation is allowed by platform default but is not combined with broad privileges here.
Assessment
This is an instruction-only, low-risk skill whose behavior is coherent with its purpose. Before installing, confirm that: (1) your agent/platform exposes the raw LLM output or a post-generation hook so the skill can inspect and redact leaked reasoning; (2) automatic regeneration or trimming is acceptable for your use case (it can change or remove content the user might expect); and (3) your provider may already offer a separate reasoning channel — in that case the skill may be redundant. If you rely on audit logs or transcripts, be aware the skill will redact or regenerate text, which may affect debugging or record-keeping. No credentials or installs are required.

Like a lobster shell, security has layers — review code before you run it.

Runtime requirements

OSmacOS · Linux · Windows
latestvk97c9rkc9kbmvmmctnds0v637585k68r
44downloads
0stars
1versions
Updated 2d ago
v1.0.0
MIT-0
macOS, Linux, Windows
@mvogt99
latest
Download

think-block-leakage

The model's internal reasoning escapes into the reply the user sees. This usually means an unclosed <think> / <thinking> tag, or a reply that begins with planning prose instead of the answer itself.

Symptoms

  • Reply contains literal <think>, <thinking>, or similar reasoning tags.
  • Reply opens with "Let me think...", "Okay, the user wants...", "First I'll need to...", or other planning preamble.
  • Reply is cut off mid-sentence and an opening reasoning tag has no matching close.
  • A </think> appears with only a few dozen characters of content after it.

What to do

  • Inspect the raw LLM output for unmatched reasoning tags before returning it. Strip or redact any content inside reasoning tags.
  • If the provider supports a separate reasoning channel, emit reasoning there and keep it out of the reply body entirely.
  • If leakage is detected, regenerate the reply. Do not ship reasoning-as-answer.
  • If the reply starts with planning language, trim the preamble. The user should see the answer first.
  • For persistent leakage, tighten the system prompt to forbid meta-commentary in the reply body.

Comments

Loading comments...