feat(claude-plugin): add remaining CLI adapters as slash commands (#1428)

Wire ask, context, symbol, why, and export into Claude Code, document them
on the website CLI reference, and fold Raghav's review fixes (include health,
drop export -f, plugin.json eleven tools) into one PR.

Author: Ayush7614

plugins/claude-code/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "repowise",
-  "description": "Codebase intelligence for Claude Code. Indexes your codebase into five layers (Graph, Git, Docs, Decisions, Code Health) and exposes them through ten task-shaped MCP tools — so Claude understands architecture, ownership, hotspots, why code is built the way it is, and where the defect risk lives.",
+  "description": "Codebase intelligence for Claude Code. Indexes your codebase into five layers (Graph, Git, Docs, Decisions, Code Health) and exposes them through eleven task-shaped MCP tools — so Claude understands architecture, ownership, hotspots, why code is built the way it is, and where the defect risk lives.",
   "version": "0.40.0",
   "author": {
     "name": "Repowise",

plugins/claude-code/CHANGELOG.md

@@ -2,6 +2,15 @@
 
 All notable changes to the Repowise Claude Code plugin are documented here.
 
+## Unreleased
+
+### Added
+- `/repowise:ask`, `/repowise:context`, `/repowise:symbol`, `/repowise:why`, and
+  `/repowise:export` — slash commands for the remaining CLI adapters so Claude
+  can synthesise answers, pull triage cards, read live-verified symbol bodies,
+  query decisions/archaeology, and export the wiki or a Structurizr model
+  without MCP-only flows.
+
 ## 0.40.0
 
 ### Added

plugins/claude-code/README.md

@@ -50,14 +50,19 @@ defect-validated score from deterministic markers).
 | `/repowise:status` | Health check — sync state, page counts, provider info |
 | `/repowise:update` | Incremental update — sync the index with recent code changes |
 | `/repowise:search` | Search the codebase wiki (fulltext, semantic, or symbol) |
+| `/repowise:ask` | Cited, synthesised answer to a codebase question (LLM call) |
+| `/repowise:context` | Triage card for files/modules/symbols (layer, hotspot, freshness) |
+| `/repowise:symbol` | Read one symbol body with live-verified line bounds |
 | `/repowise:reindex` | Rebuild the vector store (re-embed; no LLM calls) |
 | `/repowise:health` | Code-health KPIs, lowest-scoring files, refactoring targets, trends |
 | `/repowise:coverage` | Ingest or inspect coverage reports (lights up untested hotspots + per-test map) |
 | `/repowise:impacted-tests` | Tests whose coverage intersects a change (commit / range / staged) |
 | `/repowise:risk` | Defect-risk score for a change (commit or `base..head` range) |
 | `/repowise:security` | Full-history secret scan (`repowise security scan --history`) |
 | `/repowise:dead-code` | Unreachable files, unused exports, zombie packages by confidence |
+| `/repowise:export` | Export wiki pages or a Structurizr architecture model |
 | `/repowise:decision` | List, inspect, add, or confirm architectural decisions |
+| `/repowise:why` | Why the code is shaped this way (decisions + archaeology) |
 | `/repowise:doctor` | Diagnose (and optionally repair) the setup, keys, and index drift |
 
 ### Automatic skills

plugins/claude-code/commands/ask.md

@@ -0,0 +1,44 @@
+---
+description: Ask a codebase question and get a cited, synthesised answer with a confidence rating (costs an LLM call).
+allowed-tools: Bash, Read
+---
+
+# Repowise Ask
+
+Answer a question about this codebase with citations. This is the same
+synthesis the `get_answer` MCP tool performs: hybrid retrieval followed by an
+LLM answer over what it found. Unlike `/repowise:search`, this **costs an LLM
+call** — use it when the user wants an answer, not a hit list.
+
+## Steps
+
+1. If `.repowise/` doesn't exist: "This repo isn't indexed yet. Run `/repowise:init` first." Stop.
+2. Resolve the question from `$ARGUMENTS`. If empty, ask: "What do you want to know about this codebase?"
+3. Run the command and present the answer, confidence, and retrieval quality.
+   Prefer citing the paths the answer names. Do not invent citations.
+
+## Choosing the invocation
+
+- Question only → `repowise ask "<question>"`
+- Restrict retrieval to a subtree → `repowise ask "<question>" --scope packages/cli/`
+- Machine-readable → `repowise ask "<question>" --format json`
+- Raw MCP payload (including dropped blocks) → `repowise ask "<question>" --full`
+
+```
+repowise ask "how does the retry backoff work?"
+repowise ask "where is the session cookie set?" --format json
+repowise ask "how is width resolved?" --scope packages/cli/
+repowise ask "why is auth split across two modules?" --full
+```
+
+Shared targeting flags (`--path`, `--repo`, `--no-workspace`) work the same as
+the other tool-adapter commands.
+
+## How to present the result
+
+- `confidence: high` is content-grounded — cite it directly.
+- `medium` / `low` — say so, and follow any `best_guesses` / `fallback_targets`
+  into `/repowise:context` or `get_context` rather than inventing detail.
+- If a `note` warns that numbers may be synthesised, surface that caveat.
+- Prefer this over `/repowise:search` when the user asked a *question*. Prefer
+  search when they want a list of matching pages / symbols.

plugins/claude-code/commands/context.md

@@ -0,0 +1,50 @@
+---
+description: Triage card for files, modules, or symbols — layer, hotspot, fix history, freshness (relationships, not source bytes).
+allowed-tools: Bash, Read
+---
+
+# Repowise Context
+
+Pull a triage card for one or more files, modules, or symbols. This is the CLI
+adapter over `get_context`: title, summary, architectural layer, hotspot /
+bug-fix history, doc freshness, and optional relationship blocks. Relationships
+and risk signals — **not** source bytes by default.
+
+## Steps
+
+1. If `.repowise/` doesn't exist: "This repo isn't indexed yet. Run `/repowise:init` first." Stop.
+2. Resolve targets from `$ARGUMENTS`. If empty, ask which file / module /
+   `path::Symbol` to inspect.
+3. Run `repowise context` and present each card: layer, stale bit, hotspot,
+   summary. Call out fix history when present.
+
+## Choosing the invocation
+
+TARGETS are file paths, module paths, or `path/to/file.py::Symbol` ids. Batch
+them in one call.
+
+- Default triage → `repowise context <targets…>`
+- Opt-in blocks (repeatable) → `--include callers|callees|ownership|metrics|decisions|skeleton|…`
+- Richer card → `repowise context <targets…> --no-compact`
+- Machine-readable / raw payload → `--format json` / `--full`
+
+```
+repowise context src/api/routes.py src/api/auth.py
+repowise context src/api/routes.py::login --include callers --include metrics
+repowise context src/api/routes.py --include skeleton
+```
+
+`--include skeleton` adds the body-elided, line-verified file shape. For the
+exact function body, prefer `/repowise:symbol` (or `get_symbol`) with a
+`symbol_id` from the card.
+
+Shared targeting flags (`--path`, `--repo`, `--no-workspace`) work the same as
+the other tool-adapter commands.
+
+## Notes
+
+- Prefer this before opening many files with Read — one call batches cards.
+- A target the index cannot resolve returns an `error` card; report that
+  plainly instead of inventing structure.
+- For a synthesised Q&A answer, use `/repowise:ask`. For "why is it shaped
+  this way", use `/repowise:why`.

plugins/claude-code/commands/export.md

@@ -0,0 +1,60 @@
+---
+description: Export the wiki (or architecture model) to markdown, HTML, JSON, or Structurizr DSL.
+allowed-tools: Bash, Read
+---
+
+# Repowise Export
+
+Export indexed wiki pages to files, or emit a Structurizr DSL architecture
+model. Use this when the user wants a shareable dump, a static site, a JSON
+archive, or a C4/Structurizr model — not when they need an interactive Q&A
+answer (`/repowise:ask`).
+
+## Steps
+
+1. If `.repowise/` doesn't exist: "This repo isn't indexed yet. Run `/repowise:init` first." Stop.
+2. Resolve format and output from `$ARGUMENTS` (see below).
+3. Run `repowise export` and report where files were written (page count /
+   output path). Do not invent page content.
+
+## Choosing the invocation
+
+Default — markdown under `.repowise/export`:
+```
+repowise export
+```
+
+Handle `$ARGUMENTS`:
+- "html" / "site" → `repowise export --format html`
+- "json" → `repowise export --format json`
+- "json full" / "archive" → `repowise export --format json --full`
+  (`--full` keeps tombstones and adds decisions / dead code / hotspots)
+- "structurizr" / "c4" / "dsl" → `repowise export --format structurizr`
+- "structurizr standalone" → `repowise export --format structurizr --standalone`
+- "structurizr components" → add `--components`
+- "to <dir>" / "-o <dir>" → pass `--output <dir>`
+  (for structurizr, a path ending in `.dsl` names the file itself)
+- A repo path → `repowise export <path> …`
+
+```
+repowise export
+repowise export --format html -o ./site
+repowise export --format json --full
+repowise export --format structurizr --standalone --components
+repowise export --format structurizr -o architecture.dsl
+```
+
+## Structurizr notes
+
+- Default structurizr output is a **model fragment** to include from your own
+  `workspace.dsl`. Pass `--standalone` for a complete workspace with default
+  views.
+- `--force` overwrites an output file even if Repowise did not write it
+  (`--standalone` often targets `workspace.dsl`).
+- `--no-externals` leaves third-party dependencies out of the model.
+
+## Notes
+
+- Markdown / HTML / JSON write a directory of pages; structurizr writes DSL.
+- Never fabricate export contents. If the command reports no pages, say so and
+  suggest `/repowise:init`.

plugins/claude-code/commands/symbol.md

@@ -0,0 +1,48 @@
+---
+description: Read one symbol's body with live-verified line bounds (path::Name, live range, or distill omission ref).
+allowed-tools: Bash, Read
+---
+
+# Repowise Symbol
+
+Read one function, class, or constant with live-verified line bounds. This is
+the CLI adapter over `get_symbol`. Prefer it over a raw file Read when you
+already have a `symbol_id` from `/repowise:context` or `search_codebase`.
+
+## Steps
+
+1. If `.repowise/` doesn't exist: "This repo isn't indexed yet. Run `/repowise:init` first." Stop.
+2. Resolve the symbol id from `$ARGUMENTS`. If empty, ask for
+   `path/to/file.py::Name`, a live range (`path:start-end`), or a
+   `repowise#<hex>` omission ref from distill.
+3. Run `repowise symbol` and present the body. Report `verified: true/false`
+   when shown. If the id is ambiguous, show every matching body — do not
+   silently pick one.
+
+## Choosing the invocation
+
+- Named symbol → `repowise symbol "src/api/routes.py::login"`
+- Live range read → `repowise symbol "src/api/routes.py:140-180"`
+- Distill omission ref → `repowise symbol "repowise#a1b2c3d4e5f6"`
+- Extra surrounding lines → `--context-lines N` (0–50)
+- Filter restored omission lines → `--query <regex-or-substring>`
+- Machine-readable / raw payload → `--format json` / `--full`
+
+```
+repowise symbol "src/api/routes.py::login"
+repowise symbol "src/api/routes.py:140-180"
+repowise symbol "repowise#a1b2c3d4e5f6"
+repowise symbol "src/api/routes.py::login" --context-lines 3
+```
+
+A truncated body carries a `continuation` you can pass straight back to
+`repowise symbol`. Shared targeting flags (`--path`, `--repo`,
+`--no-workspace`) match the other tool-adapter commands.
+
+## Notes
+
+- Cheaper and more precise than Read + offset math when you have a
+  `symbol_id`.
+- For triage (layer / hotspot / callers) without bytes, use
+  `/repowise:context`.
+- Never invent source — if the command errors or returns no body, say so.

plugins/claude-code/commands/why.md

@@ -0,0 +1,48 @@
+---
+description: Why the code is shaped this way — decisions, rationale, and git archaeology (question, path, or decision-health dashboard).
+allowed-tools: Bash, Read
+---
+
+# Repowise Why
+
+Answer *why* the code looks the way it does: decision records, rationale, and
+git archaeology. This is the CLI adapter over `get_why`. Worth running before a
+refactor or a deliberate divergence from a pattern.
+
+`/repowise:decision` manages decision records (list / add / confirm). This
+command **queries** why something is shaped a certain way.
+
+## Steps
+
+1. If `.repowise/` doesn't exist: "This repo isn't indexed yet. Run `/repowise:init` first." Stop.
+2. Map `$ARGUMENTS` to a mode (below), run `repowise why`, and present
+   decisions, alignment, and archaeology. Never invent rationale — if the
+   command falls back to git archaeology, say that plainly.
+
+## Modes
+
+- Question → `repowise why "why is auth using JWT?"`
+- File path (governing decisions + origin story) → `repowise why src/api/auth.py`
+- Target-anchored search → `repowise why "why the retry cap?" --target src/api/client.py`
+  (`--target` is repeatable)
+- No args → `repowise why` (decision-health dashboard: stale / proposed /
+  ungoverned hotspots)
+- Machine-readable / raw payload → `--format json` / `--full`
+
+```
+repowise why "why is auth using JWT?"
+repowise why src/api/auth.py
+repowise why "why the retry cap?" --target src/api/client.py
+repowise why
+```
+
+Shared targeting flags (`--path`, `--repo`, `--no-workspace`) match the other
+tool-adapter commands.
+
+## Notes
+
+- Falls back to git archaeology when a path has no decisions, so it is never
+  empty — call that out so the user knows it is reconstructed, not recorded.
+- For *managing* ADR records (add / confirm / deprecate), use
+  `/repowise:decision`.
+- Before contradicting an existing decision, surface it to the user first.

plugins/claude-code/skills/architectural-decisions/SKILL.md

@@ -45,8 +45,10 @@ Call `get_why()` with no arguments to get the decision-health dashboard:
 - Proposed decisions awaiting confirmation
 - Ungoverned hotspots (high-churn files with no recorded decisions)
 
-The same signals surface in the CLI via `repowise decision health`, and you can
-review auto-proposed decisions with `repowise decision confirm`.
+The same signals surface in the CLI via `repowise decision health` /
+`/repowise:decision`, and you can query *why* mid-task with `repowise why` /
+`/repowise:why` (the `get_why` adapter). Review auto-proposed decisions with
+`repowise decision confirm`.
 
 ## When a file has decision markers
 

plugins/claude-code/skills/codebase-exploration/SKILL.md

@@ -55,6 +55,9 @@ Otherwise the response is current — act on it.
   Add `--no-editor-setup` if this repo is a scratch clone, a fixture, or a
   worktree: `init` otherwise repoints the user's single global `repowise` MCP
   entry at it.
+- MCP tools unavailable → prefer the matching CLI slash commands when the
+  plugin is installed (`/repowise:ask`, `/repowise:context`, `/repowise:symbol`,
+  `/repowise:search`) rather than grepping blind.
 - `get_answer`/`search_codebase` come back empty → the repo may have a
   template-rendered wiki. Fall back to `get_context` with explicit paths, and note
   that model-written pages (`repowise generate`, or `/repowise:init` with an LLM