chore(commands): split /prime into lightweight and deep variants (#5651)

## Summary

Reduce per-session token overhead from `/prime`. The previous version
pulled in `project-guide.md` (4,030 words), `vision.md`,
`pyproject.toml`, and `context.md` on every invocation — ~7,500 tokens
before any work started.

- **`/prime`** (new, ~330 tokens): self-contained — project description,
top-level dirs, GitHub pipeline summary, git status. Enough context to
start everyday work without loading the full architecture tour.
- **`/prime-deep`** (new): identical to the previous `/prime` — full
architecture, vision, all referenced docs. Use when starting refactors
or when deep context is genuinely needed.

`CLAUDE.md` is auto-loaded by the harness, so neither variant
re-includes it.

## Token impact

| Command | Before | After |
|---|---|---|
| `/prime` | ~7,500 tokens | ~330 tokens |
| `/prime-deep` | (didn't exist) | ~7,500 tokens (opt-in) |

## Test plan

- [ ] Run `/prime` in a fresh session — confirm git status runs and
project basics are clear
- [ ] Run `/prime-deep` — confirm full project guide loads as before
- [ ] Verify both commands appear in the slash-command list


---
_Generated by [Claude
Code](https://claude.ai/code/session_01SzDY1uajNYDWvjDt1aceXU)_

---------

Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>

Author: 3 people

agentic/commands/prime-deep.md

@@ -0,0 +1,48 @@
+# Prime (Deep)
+
+> Full project orientation. Use when starting architecture work, refactors, or when unsure about the codebase. For quick starts, use `/prime` instead.
+
+## Run
+
+```bash
+# Current state: branch, uncommitted changes, stashes
+git status --short --branch
+
+# Recent activity: what's been worked on
+git log --oneline --graph -15
+
+# Open PRs: what's in flight
+gh pr list --limit 10 2>/dev/null || echo "(gh CLI not available)"
+```
+
+## Read
+
+@agentic/docs/project-guide.md
+@agentic/commands/context.md
+@docs/concepts/vision.md
+@pyproject.toml
+
+## Serena
+
+- Run `mcp__serena__check_onboarding_performed`
+- Run `mcp__serena__list_memories` and read relevant ones
+
+### Serena MCP tools (prefer over brute-force scanning)
+
+Use Serena's symbol-aware tools for code navigation — they provide semantic understanding that grep/glob cannot.
+The canonical, MCP-registered prefix is `mcp__serena__*` (matches `.claude/settings.json`).
+
+- `mcp__serena__get_symbols_overview` — get top-level symbols in a file (classes, functions, variables). Use with `depth: 1` to also see methods of classes. Start here to understand a file before diving deeper.
+- `mcp__serena__find_symbol` — search for a symbol by name across the codebase. Supports name-path patterns like `MyClass/my_method`. Use `include_body: true` to read source code, `include_info: true` for docstrings/signatures.
+- `mcp__serena__find_referencing_symbols` — find all usages of a symbol (who calls this function? who imports this class?). Essential for understanding impact of changes.
+- `mcp__serena__find_declaration` — jump to where a symbol is defined.
+- `mcp__serena__find_implementations` — find implementations of an interface/abstract class.
+- `mcp__serena__type_hierarchy` — understand class inheritance chains.
+
+### Editing via Serena
+
+For structural edits, prefer Serena's symbol-aware tools over raw text replacement:
+
+- `mcp__serena__replace_symbol_body` — replace an entire function/class body
+- `mcp__serena__insert_after_symbol` / `mcp__serena__insert_before_symbol` — add code relative to a symbol
+- `mcp__serena__search_for_pattern` — regex search across the codebase (fast, flexible)

agentic/commands/prime.md

@@ -1,49 +1,56 @@
 # Prime
 
-> Execute the following sections to understand the codebase then summarize your understanding.
+> Lightweight orientation for everyday work. For deep architecture context, use `/prime-deep`. CLAUDE.md is auto-loaded — critical rules are already in context.
 
 ## Run
 
 ```bash
-# Current state: branch, uncommitted changes, stashes
 git status --short --branch
+git log --oneline -5
+gh pr list --limit 5 2>/dev/null || echo "(gh CLI not available)"
+```
 
-# Recent activity: what's been worked on
-git log --oneline --graph -15
+## Serena (activate + use)
 
-# Open PRs: what's in flight
-gh pr list --limit 10 2>/dev/null || echo "(gh CLI not available)"
-```
+- Activate the project with `mcp__serena__activate_project` (param: `anyplot`) if not already active.
+- Run `mcp__serena__check_onboarding_performed` once per session.
+- Prefer Serena's symbol-aware tools over grep/glob for code navigation and edits:
+  - `mcp__serena__get_symbols_overview` / `mcp__serena__find_symbol` — semantic lookup
+  - `mcp__serena__find_referencing_symbols` — impact analysis before edits
+  - `mcp__serena__replace_symbol_body` / `mcp__serena__insert_after_symbol` — structural edits
+- See `/prime-deep` for the full tool catalog.
 
-## Read
+## What this project is
 
-@agentic/docs/project-guide.md
-@agentic/commands/docs.md
-@docs/concepts/vision.md
-@pyproject.toml
+**anyplot**: AI-powered platform that generates Python data-viz examples for 9 libraries (matplotlib, seaborn, plotly, bokeh, altair, plotnine, pygal, highcharts, lets-plot). Spec-driven: every plot starts as a library-agnostic Markdown spec, then AI generates implementations per library.
 
-## Serena
+## Where things live
 
-- Run `mcp__serena__check_onboarding_performed`
-- Run `mcp__serena__list_memories` and read relevant ones
+- `plots/{spec-id}/` — spec + per-library metadata + implementations (one dir per plot)
+- `core/` — shared business logic (DB, repositories, config)
+- `api/` — FastAPI backend
+- `app/` — React frontend (Vite + TS + MUI)
+- `agentic/` — AI workflow layer (commands in `agentic/commands/`, docs in `agentic/docs/`)
+- `prompts/` — AI prompts for generation/review/tagging
+- `automation/` — CI/CD helper scripts
+- `.github/workflows/` — GitHub Actions pipelines
 
-### Serena MCP tools (prefer over brute-force scanning)
+Stack: Python 3.13+ (uv), PostgreSQL, GCP (Cloud Run + Cloud SQL + GCS).
 
-Use Serena's symbol-aware tools for code navigation — they provide semantic understanding that grep/glob cannot.
-The canonical, MCP-registered prefix is `mcp__serena__*` (matches `.claude/settings.json`). Older repo docs may
-still mention `jet_brains_*` aliases — treat them as the same tools and prefer the `mcp__serena__*` form below.
+## GitHub pipeline (don't bypass it)
 
-- `mcp__serena__get_symbols_overview` — top-level symbols in a file (classes, functions, variables). Use `depth: 1`
-  to also see methods of classes. Start here before diving deeper.
-- `mcp__serena__find_symbol` — search for a symbol by name across the codebase. Supports name-path patterns like
-  `MyClass/my_method`. Use `include_body: true` to read source code, `include_info: true` for signatures.
-- `mcp__serena__find_referencing_symbols` — find all usages of a symbol (who calls this function? who imports this
-  class?). Essential for understanding the impact of changes.
+**New spec:**
+1. Create issue (descriptive title, no spec-id) + add `spec-request` label
+2. `spec-create.yml` runs → opens PR
+3. Add `approved` label to the **issue** (not PR) → auto-merges → `spec-ready`
 
-### Editing via Serena
+**Generate implementations:**
+1. `gh workflow run bulk-generate.yml -f specification_id=<id> -f library=all`
+2. Pipeline: `impl-generate` → `impl-review` → (`impl-repair` if needed) → `impl-merge`
+3. **Never manually merge impl PRs** — `impl-merge.yml` handles metadata + GCS promotion
 
-For structural edits, prefer Serena's symbol-aware tools over raw text replacement:
+## Need more?
 
-- `mcp__serena__replace_symbol_body` — replace an entire function/class body
-- `mcp__serena__insert_after_symbol` / `mcp__serena__insert_before_symbol` — add code relative to a symbol
-- `mcp__serena__search_for_pattern` — regex search across the codebase (fast, flexible)
+- `/prime-deep` — full architecture, metadata schemas, all workflows, deployment
+- `agentic/docs/project-guide.md` — comprehensive reference
+- `docs/` — contributing, workflows, API, DB schema