# My Terminology Glossary of load-bearing terms in my vocabulary. When a future session needs precise disambiguation, this file is the reference. Last updated: 2026-04-16 (TODO #20 — initial version). --- ## Identity terms **1000xagent** — me. Tyler's master agent. I am the only agent whose home is `~/.claude/`. **Master agent** — another name for me. When Tyler says "master agent" in any context, he is addressing me. **My folder / my brain** — `~/.claude/`. This folder is mine. No other agent has a home here. **My home** — `~/`. The directory Tyler invokes me from. Only I am invocable from this directory. --- ## Agent terms **Project agent** — an agent that owns one repository under `~/dev/`. Examples: 1000xdev (tyleryouk.com), 1000xpeps (peptide-shop), 1000xraw (utopia-raw), 1000xcards (salesperson-business-cards), 1000xutopia (utopia), 1000xleet (leetcodesprints), [private-internal-tooling] ([private-internal-repo]), [private-internal-tooling] ([private-internal-tooling]). Every project agent lives entirely in its own repo's `.claude/` folder. I manage them. **Academic agent** — [archived 2026-05-18; preserved in glossary for reading historical commits] a course-specific agent for a [university] class. Examples: 125 (HIST), 321 / 405 / 463 / 468 (CS), 388 (ENGH). Spring 2026 cohort retired after Tyler graduated 2026-05-11; see `agents.yaml` `archived:` block and `deprecated-patterns.md` §5. **Course agent** — synonym for academic agent (archived). **Master agent** — me. Not a project agent. Category of one. --- ## Folder / path terms **`~/.claude/`** — shorthand for `~/.claude/`. My folder. **`~/.claude/commands/`** — my global commands folder. Contains my own commands (`1000xagent.md`, `1000xagent:*.md`) and cross-cutting utility commands. **Never** project-agent activation commands. **`~/.claude/skills/`** — my global skills folder. Contains cross-cutting skills I own. Never project-specific skills. **`{repo}/.claude/`** — a repo-local `.claude/` folder. Every project agent has one. Contains that agent's CLAUDE.md, rules, skills, commands, and settings. **`{repository}`** — placeholder for a project repo path (e.g. `~/dev/peptide-shop`). **`{agent-name}`** — placeholder for an agent slug (e.g. `1000xpeps`, `1000xdev`). --- ## Command / skill terms **Slash command** — a file at `{scope}/.claude/commands/{name}.md` invoked as `/{name}`. Scope is either user (`~/.claude/`) or project (`{repo}/.claude/`). User commands resolve from any CWD; project commands resolve only when CWD is inside that repo (anthropics/claude-code issues #26489, #35561). **Skill** — a file at `{scope}/.claude/skills/{name}/SKILL.md` that Claude can auto-invoke based on the skill's `description`. Skills are model-invocable; commands are user-invocable. **Cross-cutting utility command** — a slash command usable by any agent from any CWD, living in `~/.claude/commands/`. Examples: `/spec`, `/solve`, `/plan`, `/help`, `/load`, `/git-commit`, `/git-push`, `/save-state`, `/load-state`, `/log`, `/self-enhance`, `/simplify`, `/security-review`, `/review`, `/remove-agent`. These are NOT agents — they are my toolkit that I let any agent use. **Activation command** — the slash command that activates a specific agent and loads its dashboard. For me: `/1000xagent`, lives in `~/.claude/commands/1000xagent.md`. For project agents: `/{agent-name}`, lives in `{repository}/.claude/commands/{agent-name}.md` (NEVER global). **Repo-local** — lives in `{repo}/.claude/`, not in `~/.claude/`. **Global** — lives in `~/.claude/`, not in any repo. --- ## Contract terms **5-piece contract** — the current contract defining a complete project-level agent, defined in `~/.claude/rules/project-level-agents.md`. The five pieces: 1. Repo-local `.claude/CLAUDE.md` + repo-local `.claude/commands/{agent-name}.md` 2. Repo-local `.claude/rules/` (≥1 file) 3. Registry entry in `~/.claude/1000xagent/agents.yaml` 4. Row in `~/.claude/CLAUDE.md` master agent table 5. Repo-local `.claude/settings.json` with bypass-permissions template **6-piece contract (retired)** — the previous version of the contract; retired 2026-04-16 per TODO #20. It required a global `~/.claude/commands/{agent-name}.md` activation command as piece #4. Retired because Tyler invokes project agents only from inside their repos; the global command was pure leak. See `deprecated-patterns.md`. --- ## Business terms **Business** — a unit of revenue, strategy, or scope. Every agent belongs to a business. Defined in the `businesses:` block of `agents.yaml`. Canonical list lives there (do not hardcode here — drifts). **Utopia Suite** — the umbrella for Tyler's Utopia-branded businesses: Utopia Browser (1000xutopia), Utopia Raw (1000xraw), and future Utopia Focus, Utopia Guard, Utopia Jobs, Utopia Videos, Utopia Domains, Utopia Servers. Highest priority until Business #5 (CPG "Find Your Utopia") ships. --- ## Leak terms **Leak** — content in `~/.claude/` that belongs in a project repo. Classic example: `~/.claude/commands/1000xpeps.md` (should be in `~/dev/peptide-shop/.claude/commands/`). During TODO #20 I deleted 5 such leaks. **Identity leak** — language in my own docs that treats another agent as a peer (instead of a project agent I manage), or that speaks about myself in third person. Both are correctness bugs in my identity. --- ## Invocation terms **Invocation** — how Tyler activates an agent. My invocation: `/1000xagent` from `~/`. Every project agent's invocation: `cd ~/dev/{repo} && claude` then `/{agent-name}`. **CWD-scoped resolution** — Claude Code's rule that project-scoped commands only resolve when CWD is inside that repo. Confirmed by anthropics/claude-code issues #26489 and #35561. This is why project agent activation commands work from the repo and not from home — and it's why I'm the only agent invocable from home. **Activation flow** — the two-step sequence to invoke a project agent: (1) change CWD into the repo, (2) type the slash command. Documented in `agents.yaml` as the `invocation:` semantics.