AGENT SKILL

reiterate

Verify comprehension before any action by restating the user's intended meaning in structured, domain-appropriate language. Use when the user invokes /reiterate, "$reiterate", "make sure you understand", "formalize this", "restate my intent", or wants a precise, auditable capture of a claim, requirement, research goal, or specification before planning, editing, coding, or researching proceeds. Prevents expensive mis-execution on the wrong interpretation.

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

01 前言 Front Matter

name
reiterate
metadata
{ "short-description": "Precise restatement of user intent in domain language before action" }
description
Verify comprehension before any action by restating the user's intended meaning in structured, domain-appropriate language. Use when the user invokes /reiterate, "$reiterate", "make sure you understand", "formalize this", "restate my intent", or wants a precise, auditable capture of a claim, requirement, research goal, or specification before planning, editing, coding, or researching proceeds. Prevents expensive mis-execution on the wrong interpretation.
when-to-use
Explicitly before any non-trivial action, implementation, research probe, or edit when the request is complex, novel, or high-stakes. Also useful mid-conversation to re-anchor after scope has drifted.
allowed-tools
[ "read_file", "grep", "list_dir", "memory_search", "memory_get" ]
argument-hint
<the request, claim, or artifact whose intent must be reiterated>

03 SKILL.md 正文

Reiterate

Purpose

Demonstrate to the user — before any tool use, edit, plan, or analysis — that their exact intent has been received and can be expressed back in the precise language of the relevant domain. This is a verification gate, not a planning or execution skill.

This is the Grok-native port of the Codex "reiterate" skill, with added memory awareness and explicit PGS-domain language support for workspaces that contain AGENTS.md.

Strict Separation

  • Reiterate only. Do not propose next steps, implementation plans, task lists, or begin using tools for the underlying task.
  • Only after the user confirms "yes, that matches" or "proceed from that framing" do other skills (actualize, ooda, task-chain, search_replace, etc.) become appropriate.

Method

  1. Identify the Concrete Elements

    • Subject
    • Intended claim or distinction
    • Desired outcome or artifact
    • Target audience / domain / contract (e.g., "for a senior number theorist", "for the AGENTS.md contract", "for a code reviewer who will inherit this diff")
  2. Preserve Scope Exactly

    • Do not widen, weaken defensively, or substitute a nearby abstraction.
    • Do not add safety disclaimers the user did not request.
  3. Translate into Structured Domain Language

    • Use the vocabulary and distinctions native to the subject (for PGS work: ordered prime-gap state, divisor-count field, DNI, GWR, endpoint-chain, chamber reset, structural certificate, reciprocal transport, modulus-link closure, unresolved PGS state, etc.).
    • Name variables, invariants, constraints, regimes, and unresolved distinctions the request depends on.
  4. Separate Assertion from Request

    • Clearly distinguish what the user appears to be asserting from what they appear to be asking the system to do.
  5. Minimal Auditable Structure

    • Use the smallest set of headings that makes the understanding checkable.
    • Common headings (use only those that apply):
      • What I Understand
      • Structured Form (or Contract Language, Definitions and Claims, etc.)
      • Key Distinctions
      • Ambiguity To Confirm (only if genuine ambiguity exists that would change meaning)
  6. Close with Explicit Verification Sentence

    • "If that matches your intent, I can act from this framing."
    • "The only point I would confirm before acting is ..."
    • "I understand the request as X, not Y."

Success Criteria

  • The user can reply with a single short correction or "correct, proceed" and the subsequent work will be on the right target.
  • No action tools have been invoked during the reiterate turn.
  • The restatement uses the user's actual concepts and the domain's native terminology, not generic assistant language.
  • In a PGS workspace: the restatement demonstrates correct use of the PGS-first frame from AGENTS.md.

Guardrails

  • This skill must never leak into planning or execution.
  • Do not correct the user on wording unless the difference changes the operational meaning.
  • Do not introduce disclaimers, scope reductions, or "but we should also consider..." that the user did not ask for.
  • Do not replace domain language with vague "helpful assistant" phrasing.
  • Do not end with a menu of options or "would you like me to...".
  • Higher contracts (AGENTS.md, explicit user constraints, safety) are surfaced only if they directly affect the captured intent.

Grok Environment Notes

  • memory_search and memory_get are used to pull prior context so the reiterate can be consistent with long-running threads.
  • read_file / grep / list_dir let the skill ground the restatement in actual workspace artifacts when the user's request references them.
  • After a successful reiterate + user confirmation, the clean framing produced here becomes excellent input for actualize, ooda, task-chain, or direct search_replace / write.

Example (PGS-flavored, abbreviated)

What I Understand
You want a precise restatement of the current open problem in leftmost-minimum-divisor selection for simultaneous multi-chamber resets, using only the language of the Divisor Normalization Identity and GWR, for the purpose of writing a one-page proof obligation map.

Structured Form (PGS-native)
... (exact restatement using the required objects and invariants) ...

Key Distinctions

  • The distinction between a chamber reset and a modulus-link closure.
  • ...

Ambiguity To Confirm
Whether the map must also cover the case of a divisor-count field that is not yet fully normalized.

If that matches your intent, I can act from this framing.

This skill is the single most effective defense against expensive "I thought you meant..." failures in technical work.

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