From claude-toolkit
Challenge recent assertions — find evidence or explain reasoning for claims made in the conversation.
How this skill is triggered — by the user, by Claude, or both
Slash command
/claude-toolkit:proveThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Pause and review your recent statements in this conversation. The goal is to identify assertions you made and back them up with evidence or reasoning.
Pause and review your recent statements in this conversation. The goal is to identify assertions you made and back them up with evidence or reasoning.
Scan your recent messages (focus on the last few exchanges) for factual claims, especially:
List each assertion clearly, quoting the relevant text.
For each assertion, attempt to verify it in this order of preference:
Source code / local files: If the claim is about the codebase, find the specific code or config that proves it. Show the file path, line number, and relevant snippet.
Official documentation: Use WebSearch and WebFetch to find authoritative documentation (official docs, RFCs, specs, release notes) that confirms the claim. Link to the source.
Reasoning from first principles: If external evidence is unavailable or the claim is about general software concepts, explain the logical reasoning step by step. Be explicit about what you know vs. what you're inferring.
For each assertion, present:
| # | Assertion | Verdict | Evidence |
|---|---|---|---|
| 1 | "X is a known limitation of Y" | Confirmed | [link or file:line] |
| 2 | "The recommended approach is X" | Partially confirmed | Docs say X but with caveats... |
| 3 | "This causes X" | Reasoning only | No external source found. Reasoning: ... |
| 4 | "X was deprecated in v2" | Could not confirm | No evidence found — retracting this claim |
For any assertion you cannot prove:
Provides behavioral guidelines to reduce common LLM coding mistakes, focusing on simplicity, surgical changes, assumption surfacing, and verifiable success criteria.
Searches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.
Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.
npx claudepluginhub kfitzpatrick/claude-toolkit --plugin claude-toolkit