Skip to main content
skill CX Codex

echo · ghostwright — Claude Code skill for recalling past decisions

Before answering a substantive question, echo searches memory for a similar earlier decision. If it finds a strong match, it brings up that conclusion and asks whether it still applies.

Apache-2.0 ★ 1,497 updated 2026-06-16

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add ghostwright/phantom --skill echo -a claude-code
Open on GitHub

What echo does

echo helps Claude Code take the user’s earlier decisions into account. Before answering a request for advice or a choice between approaches, it searches memory for related entries. Greetings, status questions, acknowledgments, and direct tasks do not trigger a search.

The skill calls mcp__phantom-reflective__phantom_memory_search, restating the question’s meaning and requesting up to five results across memory types. It checks how closely a result matches, how old it is, and whether it contains an actual conclusion. A strong match is surfaced before a new answer; a weak one is ignored.

The memory search tool must be available. Without it, echo cannot compare a new question with earlier decisions.

What it helps with

  • retrieve an earlier decision on a similar question
  • remind the user which approach they chose
  • check whether a previous preference still applies
  • skip memory searches for simple operational requests

Example prompts for the agent

  • Which approach did we choose for handling errors in this project?
  • We discussed how to structure authentication before. Remind me what we decided.
  • What’s the best way to split this module? Check whether we covered it earlier.

How to install echo

1Skills CLI (Vercel)

npx skills add ghostwright/phantom --skill echo -a claude-code

2Manually: put the files in place

~/.claude/skills/echo/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/echo/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about echo

How do I install echo in Claude Code?

Run npx skills add ghostwright/phantom --skill echo -a claude-code. The skill file is at ~/.claude/skills/echo/SKILL.md.

What does echo do?

It searches memory for an earlier answer before responding to a substantive question. When it finds a strong match, it recalls the conclusion and asks whether it remains current.

Which questions trigger a memory search?

Requests for a recommendation, a choice between options, or a past decision can trigger one. Greetings, status checks, and direct tasks without an open decision are skipped.

Source text

SKILL.md · Apache-2.0
---
name: echo
x-phantom-source: built-in
description: Before answering a substantive question, quietly check whether the user has already resolved this question in the past.
when_to_use: Use when the user asks a substantive question that might have been asked and resolved before. Substantive questions are things like "how should I do X", "what is the right way to Y", "which approach is better", "what did we decide about Z". Do NOT fire on greetings, small talk, status checks, or operational queries like "what time is it" or "are you online". Before deriving a new answer, run a memory similarity check. If a strong prior match exists, surface it inline and ask whether anything has changed.
allowed-tools:
  - mcp__phantom-reflective__phantom_memory_search
context: inline
---

# Echo: the prior-answer surfacer

## Goal

Respect the user's past thinking. Before deriving a new answer to a substantive question, check whether the user already resolved this question weeks or months ago. If yes, surface the prior answer inline and ask whether anything has changed. If no, proceed to answer normally without mentioning that you looked.

The user should feel like you remember what they already decided, not like you are doing paperwork.

## Steps

### 1. Classify the question

Determine whether the question is substantive. Skip if:
- It is a greeting or acknowledgment ("hey", "thanks", "got it").
- It is an operational query ("are you online", "what time is it", "are you working on X").
- It is an imperative with no open decision ("send this to Anna", "delete that file").
- It is clearly a first-time question with no prior context ("what does this error mean in this new log line").

Proceed if:
- The user is asking for a recommendation or opinion.
- The user is asking "what did we decide" or "what is the right way".
- The user is weighing options on something they have discussed before.
- The user is asking a question that sounds like it could have been asked before.

**Success criteria**: you have a yes or no on whether to run the echo check. If no, do not call the search tool at all.

### 2. Search memory for prior answers

Call `mcp__phantom-reflective__phantom_memory_search` with `query: "<the user's question in your own words>"`, `memory_type: "all"`, `limit: 5`. The query should be a restatement of the semantic intent, not a literal copy of the user's words.

**Success criteria**: you have a list of 0-5 hits with similarity scores.

### 3. Judge the match

Examine the top hit. It is a strong match if all of these hold:
- Similarity score is above 0.80 if the tool returns one.
- The hit is at least 3 days old.
- The hit actually addresses the same question, not just the same keywords.
- You can clearly see what the prior conclusion was.

If the top hit is NOT a strong match, proceed to answer the question normally from scratch. Do not mention the echo check to the user.

**Success criteria**: you have a clear yes or no on the match.

### 4. Surface the prior answer if there is one

If there is a strong match, respond BEFORE deriving a new answer:

> You asked something very similar on [date] and you landed on [paraphrase of the prior conclusion]. Has anything changed since then, or is that still your view?

Wait for the user's response.

If the user says "no, things are different now" or explains what changed, proceed to derive a new answer informed by the new context.

If the user says "yes, that is still my view", acknowledge and ask what they want to do with that. Sometimes they just needed the reminder.

**Success criteria**: the user is aware of their prior thinking, and you have their explicit signal on whether to rebuild from scratch or honor the prior answer.

## Rules

- Never surface weak matches. A low-confidence echo is worse than no echo because it erodes the user's trust in your memory.
- Never em-dash.
- Never mention that you ran the echo check if it did not fire. The user should not see the machinery when it does not apply.
- Be brief on the surface: two sentences, not four.
- If the prior conclusion has expired (for example, it is about a project that has since shipped), treat it as a weak match and proceed normally.
- Always paraphrase the prior conclusion in your own words. Do not copy-paste from memory.

© ghostwright · Apache-2.0 · ghostwright/phantom

Similar skills

All skills for Agent setup: memory, context, orchestration →

echo by other authors comes first.

From the same set38

The whole set →
phase-10a-murph-pi-event-research phase-10a-murph-pi-event-research is a Claude Code prompt that audits Murph and Pi runtime events to surface agent progress, compaction, and tool calls in Phantom chat. GP ChatGPT ★ 1,497repo stars phase-10a-product-experience-research The phase-10a-product-experience-research prompt guides research into Phantom’s long-running agent chat, from run status to tool details and error states. Use it to shape an interface proposal and a phased build plan. GP ChatGPT ★ 1,497repo stars phase-10c-murph-progress-research phase-10c-murph-progress-research is a Claude Code prompt for researching Murph’s progress contract to improve tool call cards in Phantom chat. GP ChatGPT ★ 1,497repo stars phase-10b-durable-timeline-research phase-10b-durable-timeline-research directs an autonomous research run on durable chat timelines and SSE reconnect logic in Phantom without touching source files. GP ChatGPT ★ 1,497repo stars phase-10m-pi-murph-continuity-research phase-10m-pi-murph-continuity-research is a prompt for Claude Code and ChatGPT that investigates context recovery and compaction in Murph and Phantom, checking whether they reinvent Pi primitives. GP ChatGPT ★ 1,497repo stars phase-10i-transcript-recovery-builder The phase-10i-transcript-recovery-builder prompt for Claude Code builds transcript recovery for Phantom on Murph, letting agents search chat history after context compaction. GP ChatGPT ★ 1,497repo stars thread Trace how your thinking on a topic changed over time. The skill turns memory entries into a chronological account with marked turning points. CX Codex +1 ★ 1,497repo stars phase-10a-chat-pipeline-research phase-10a-chat-pipeline-research — a slash command that walks Phantom's chat pipeline from SDK events through SSE to React rendering. You get a file-grounded report, and no code is touched. GP ChatGPT ★ 1,497repo stars

Project sponsors

Companies that keep this analytics open