Phantom is an AI co-worker that operates its own computer, maintains persistent memory, uses MCP tools, and builds software and infrastructure in its workspace. It is for people who want an agent that can continue work across sessions through Slack or web chat while keeping its workspace separate from the user's computer. The catalogue skills and instruction support workflows for Phantom.
Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add ghostwright/phantom --skill echogit clone --depth 1 https://github.com/ghostwright/phantomWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/ghostwright/phantom/echo)<a href="https://agentmods.dev/skills/ghostwright/phantom/echo"><img src="https://agentmods.dev/badge/skills/ghostwright/phantom/echo/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/ghostwright/phantom/echo"><img src="https://agentmods.dev/badge/skills/ghostwright/phantom/echo.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00022 | $0.00969 |
| Opus 5.5 | $0.00009 | $0.00388 |
| Sonnet 5.5 | $0.00004 | $0.00194 |
| Haiku 4.5 | $0.00002 | $0.00097 |
Grade A, and why
echo scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 77 lines — stays where its author put it; the contents beside it link to each section on GitHub.
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.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 1mo ago First seen · 77 lines · 22 tokens per session scan A 93b8741c0a5f
echo is a skill published in the GitHub repository ghostwright/phantom (1,483 stars, last pushed 4mo ago), licensed Apache-2.0. It adds 22 tokens to every session and 969 once invoked, about $0.0001 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
hive.context-preservation
Proactively extract critical values from tool results into working notes before automatic context pruning destroys them.
hive.note-taking
Maintain a free-form scratchpad of decisions, extracted values, and open questions so context pruning doesn't lose anything you still need.
Vizra ADK Memory System
Implement persistent memory, session context, and vector memory (RAG) for AI agents.
ashlr-codex-genome-refresh
Refresh or maintain Ashlr project genome state from Codex. Use after architectural changes, schema/routing changes, or substantial refactors.
whiteboard-curate-suggest
When an insight, decision, or checklist with lasting value comes up during a conversation, suggest saving it to the library, and on the user's consent register it in Curated Knowledge via knowledgepromote.
whiteboard-react-to-share
When the user shares a note or capture via a [USERSHARED] trigger or [Spotlight Context], naturally bring it up in conversation (getting an image description via whiteboarddescribe first if needed) and voice your opinion.