How do I install the Expert skill?

Run npx skillmds@latest add capitalone/expert in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Expert safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Expert?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Expert free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Expert?

capitalone (@capitalone) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Expert by capitalone — Agent Skill

Name: Expert
Author: capitalone

Expert

This project's long-term memory — how to run, validate, and extend it; its architecture, patterns, hard invariants, worked examples, and current decisions/direction. Consult when planning a feature, validating a spec, implementing a slice, or writing intent for this project. Routing table below points to one small reference file per topic.

capitalone 6c1f3f9 2 files · 3.8 KB Updated Sep 17, 2026

File contents

The Expert — this project's long-term memory

How to use this memory

Scan the routing table, open only the reference files whose USE WHEN matches the task at hand. Each file is one topic; files cross-link with [[wikilinks]] (resolve [[name]] to references/name.md). Do not page through everything.

Facts vs. decisions

Most shards describe the code as it IS — plan and build consistently with them. A decision- shard describes where the project is HEADING; the code may not reflect it yet.

When your task touches an area a decision covers, make a call: does this work advance the decision (build the new way) or stay consistent with current code (build the old way)? Read the decision's "Until fulfilled" note first — it answers this for its case. If it's silent: new isolated code follows the decision; changes to existing code stay consistent with what surrounds them, unless the task is the migration. Never assume a decision is already implemented — verify the current state in code before planning against it.

Routing table

Reference	USE WHEN
(empty — the first /learn, Reflect, or your own edit adds the first shard)

Writing to this memory

One topic per file, named <prefix>-<topic>.md, opening with a USE WHEN: line.
Anything that helps the next agent belongs here — facts about the code as it is (cite file paths, shas) AND decisions, direction, and aspirations about where it's going. Make clear in the prose which is which.
Facts → how-to-/concept-/pattern-/invariant-/example-, cited. Direction (a choice the code hasn't caught up to) → decision-<slug>.md: the direction and why, a [[concept-…]] pointer to the current state, and an Until fulfilled: note (what advances the decision vs. what stays consistent with today's code). No status field — the file existing means it's adopted.
Reconcile, don't accumulate — when reality or intent changes, edit or delete the shard. A merge that fulfills a decision promotes it to a concept-/pattern- fact and deletes the decision; a decision you've walked back gets deleted, not appended to.
Add the routing-table line in the same commit as the shard.

---name: expertdescription: This project's long-term memory — how to run, validate, and extend it; its architecture, patterns, hard invariants, worked examples, and current decisions/direction. Consult when planning a feature, validating a spec, implementing a slice, or writing intent for this project. Routing table below points to one small reference file per topic.---# The Expert — this project's long-term memory## How to use this memoryScan the routing table, open **only** the reference files whose `USE WHEN` matchesthe task at hand. Each file is one topic; files cross-link with `[[wikilinks]]`(resolve `[[name]]` to `references/name.md`). Do not page through everything.## Facts vs. decisionsMost shards describe the code as it IS — plan and build consistently with them. A`decision-` shard describes where the project is HEADING; the code may not reflectit yet.When your task touches an area a decision covers, make a call: does this work**advance** the decision (build the new way) or stay **consistent with currentcode** (build the old way)? Read the decision's "Until fulfilled" note first — itanswers this for its case. If it's silent: new isolated code follows the decision;changes to existing code stay consistent with what surrounds them, unless the task*is* the migration. Never assume a decision is already implemented — verify thecurrent state in code before planning against it.## Routing table| Reference | USE WHEN ||---|---|| _(empty — the first /learn, Reflect, or your own edit adds the first shard)_ | |## Writing to this memory- **One topic per file**, named `<prefix>-<topic>.md`, opening with a `USE WHEN:` line.- **Anything that helps the next agent belongs here** — facts about the code as it is (cite file paths, shas) AND decisions, direction, and aspirations about where it's going. Make clear in the prose which is which.- **Facts** → `how-to-`/`concept-`/`pattern-`/`invariant-`/`example-`, cited. **Direction** (a choice the code hasn't caught up to) → `decision-<slug>.md`: the direction and why, a `[[concept-…]]` pointer to the current state, and an **Until fulfilled:** note (what advances the decision vs. what stays consistent with today's code). No status field — the file existing means it's adopted.- **Reconcile, don't accumulate** — when reality *or intent* changes, edit or delete the shard. A merge that fulfills a decision promotes it to a `concept-`/`pattern-` fact and deletes the decision; a decision you've walked back gets deleted, not appended to.- Add the routing-table line in the same commit as the shard.

Frequently asked questions