AGENT SKILL

prompts-check-for-signal

Apply the "Check for Signal" prompt-library workflow: given a hypothesis or claim, design and execute (or scaffold the execution of) a decisive test that definitively proves or falsifies the claim — never artificially. Create a new isolated folder under experiments/ (or equivalent), document everything in FINDINGS.md that **leads with the conclusion**, followed by technical supporting evidence. Do not modify artifacts outside the new folder. Use when the user says "check for signal on this claim", "design a decisive test for", "falsify this hypothesis", "prove or disprove", "use check-for-signal", or runs /prompts-check-for-signal.

License 许可
MIT
最近更新
2026-07-19
Stars
★ 1
Forks
1

01 前言 Front Matter

name
prompts-check-for-signal
metadata
{ "version": "1.0.0", "short-description": "Design and document a decisive falsifying/proving test for a claim (experiments/ isolation)", "canonical-codex-skill": "/Users/velocityworks/.codex/skills/prompts-check-for-signal/SKILL.md" }
description
Apply the "Check for Signal" prompt-library workflow: given a hypothesis or claim, design and execute (or scaffold the execution of) a decisive test that definitively proves or falsifies the claim — never artificially. Create a new isolated folder under experiments/ (or equivalent), document everything in FINDINGS.md that **leads with the conclusion**, followed by technical supporting evidence. Do not modify artifacts outside the new folder. Use when the user says "check for signal on this claim", "design a decisive test for", "falsify this hypothesis", "prove or disprove", "use check-for-signal", or runs /prompts-check-for-signal.
when-to-use
Designing or running a clean, high-leverage experiment or test whose outcome will materially update belief about a specific claim. Strong triggers: 'decisive test', 'falsify', 'prove or disprove', 'check for real signal', 'experiments/ folder'.
allowed-tools
[ "read_file", "grep", "list_dir", "write", "search_replace", "run_terminal_cmd", "web_search" ]
argument-hint
<hypothesis, claim, or research question to design a decisive test for>

03 SKILL.md 正文

Check for Signal — Grok Port

Design and document (and where possible execute) a clean, high-power test that can actually move the needle on a hypothesis instead of producing ambiguous or noisy results.

Purpose

The original prompt insists on decisive tests that "definitively prove or falsify (never artificially)" and requires strict workspace isolation plus conclusion-first documentation. This port makes the skill an active experiment designer and documenter inside the Grok tool environment.

Invocation

/prompts-check-for-signal <hypothesis or claim description>

Any additional context in the conversation (prior results, existing code, data files) is part of the input material.

Core Contract (Preserved)

Goal: Design and execute a test that will definitively prove or falsify (never artificially) the following hypothesis/claim.

  • Create a new folder under experiments/ (or project-appropriate equivalent) for all artifacts and documentation.
  • Do not modify any artifacts outside of the new folder.
  • Meticulously document findings in FINDINGS.md that leads with the conclusion, followed by technical supporting evidence.

Grok Workflow

  1. Clarify the Claim

    • Restate the exact hypothesis in the clearest, most falsifiable form possible.
    • Surface hidden assumptions, scope, and what "definitively" would look like for this claim.
  2. Design the Decisive Test

    • Choose the smallest, highest-power experiment or analysis that can produce a clear positive or negative result.
    • Prefer deterministic or low-variance designs over statistical ones when both are viable (aligns with project AGENTS.md preference for deterministic reasoning).
    • Define the success/failure criteria in advance (what numbers, what structural property, what invariant violation, etc. would count as proof or falsification).
    • Identify required inputs, controls, and how to make the result auditable.
  3. Scaffold the Experiment Workspace

    • Target: experiments/<kebab-case-claim-slug>/ or experiments/signal-<slug>/
    • Create inside it (using write/search_replace):
      • README.md (hypothesis, test design, how to run)
      • FINDINGS.md (template or initial entry — see format below)
      • Any scripts, config, or small programs needed for the test (keep minimal)
      • Makefile or run instructions if compilation/execution is involved
  4. Execute or Guide Execution (when possible)

    • Use available tools (run_terminal_cmd, code via write+exec patterns, web tools for data, etc.) to carry out the test.
    • If full execution is not possible in the current turn, produce a complete, runnable recipe that a follow-up agent or human can execute with zero ambiguity.
  5. Document Conclusion-First

    • The FINDINGS.md must open with a clear prose sentence or short paragraph stating the conclusion.
    • Then provide the technical evidence, methodology, raw outputs, statistical or structural results, and any caveats.
    • Example opening:

      "The test falsifies the claim that chamber resets are uniformly distributed: in the 10,000-sample run, reset events clustered 3.2× more frequently immediately after leftmost-minimum transitions than expected under uniformity (p < 0.001 via exact permutation test)."

  6. Verification & Audit Trail

    • Record exact commands, seeds (if any), tool calls, file hashes, and timestamps where relevant.
    • Make it possible for a third party to reproduce the exact conclusion from the artifacts.

FINDINGS.md Required Shape

# FINDINGS — <short claim label>

**Conclusion**: [One or two sentences that a reader can quote as the result. Lead with this.]

**Hypothesis under test**: [restatement]

**Test design**: [1-3 paragraphs describing the decisive procedure and why it has power]

**Execution details**:
- Date/time
- Exact commands / scripts run
- Inputs / datasets / parameters
- Environment (if relevant)

**Results**:
[Tables, key numbers, structural observations, logs, images descriptions, etc.]

**Interpretation**:
[How the data supports or refutes the claim. Be quantitative where possible.]

**Limitations & caveats**:
[Honest statement of what the test does not cover]

**Next actions** (optional):
- ...

Success Criteria

  • A new isolated experiments/... folder exists containing the complete record.
  • FINDINGS.md literally leads with the conclusion in plain prose.
  • The test design is genuinely capable of producing a decisive (non-ambiguous) result.
  • No files were touched outside the experiments folder.
  • A later reader can understand exactly what was tested and what the outcome means without re-running the entire conversation.

Guardrails

  • "Never artificially" means do not design a test whose outcome is foreordained by construction or by cherry-picked data. If the only honest answer is "this claim is currently untestable with available resources," say so clearly instead of forcing a weak test.
  • Prefer tests that can be executed deterministically or with full auditability.
  • In PGS contexts: tests should target PGS objects, invariants, or rule applications directly when the hypothesis touches them.
  • Workspace isolation is absolute — this skill does not "helpfully" edit files elsewhere in the project.

This port converts the original "Check for Signal" prompt into a disciplined experiment harness creator and conclusion-first reporter that works with Grok's full toolset.

04 同仓库更多 Skills

zfifteen
zfifteen/skills-catalog

Zotero

Use Zotero Desktop from Codex to enable/probe the local API, search a local Zotero library, list items/collections/tags, export BibTeX, insert citation keys into LaTeX or Markdown drafts, read indexed full text when requested, and import BibTeX/RIS records into Zotero through the connector server. Use when the user mentions Zotero, citations, references.bib, BibTeX export, local Zotero API, localhost:23119, or adding citations from a Zotero library.

★ 1
zfifteen
zfifteen/skills-catalog

zoom-apps-sdk

Use when using Apps SDK.

★ 1
zfifteen
zfifteen/skills-catalog

youtube

Obtain accurate, timestamped YouTube video transcripts (preferring manual captions over auto-generated) as the primary deliverable, then derive summaries, key claims, or notes from that transcript. Use when the user provides a YouTube URL (or video ID) and wants the transcript, a faithful summary, key claims with attribution, timestamped notes, or "what does this video actually say". The skill is transcript-first: never substitute metadata, title, description, or search snippets for the spoken content. Triggers: "youtube transcript", "get the transcript of this video", "summarize this youtube video", "what did they say in https://youtube.com/...", "key claims from this talk".

★ 1
zfifteen
zfifteen/skills-catalog

yeet

Publish local changes to GitHub by confirming scope, committing intentionally, pushing the branch, and opening a draft PR through the GitHub app from this plugin, with `gh` used only as a fallback where connector coverage is insufficient.

★ 1
zfifteen
zfifteen/skills-catalog

x-post-creator

Create X-ready post threads and co-located story packages from the active conversation or referenced project material. Use when the user says turn this into an X-post, /x-post, make this into an X thread, or wants plain-language social posts with supporting images saved to docs/x-posts/[story-name]/.

★ 1
zfifteen
zfifteen/skills-catalog

x-api

Search, read, and write X posts through the local X API MCP server (the same backend used by the Codex x-api skill). Use for searching X/x.com, reading threads and replies, creating lists, staging/publishing posts for @alltheputs, finding high-visibility reply targets for research outreach, ingesting public context, or any request that mentions /x-api or "search X".

★ 1