Menü
Use when starting a Miller task, choosing a Miller tool, or choosing a Miller search mode.
Mit Codex oder Claude installieren Kopieren Sie diesen Prompt, fügen Sie ihn in Codex, Claude oder einen anderen Assistant ein und lassen Sie die Skill-Seite prüfen und installieren.
Use before changing existing indexed files with Miller edit, especially symbol rewrites, text replacements, renames, or refactors.
Use when an agent needs extractor-recognized code-shape facts such as routes, htmx attributes, Alpine directives, SQL DDL/DML, or data-document structure.
Use when tracing Miller bridge paths in supported providers: dotnet-web URL literals and fetch/axios client requests to ASP.NET signals, client requests to Next.js route handlers or Nuxt server routes, Next.js/Nuxt route references to file routes, Vue/React references to route definitions, or backend-http client requests to Express/FastAPI/Flask/Django/Spring/Go/gin/echo/Rails/NestJS/Laravel/Phoenix/axum/actix/Symfony/Ktor route facts.
Use when orienting on an unfamiliar code area with Miller, explaining a module, finding entry points, or gathering context before a change.
Use when assessing blast radius with Miller, choosing tests for a change, answering who uses a symbol, or planning a refactor.
Use when resuming Miller-backed work from a handoff packet or validating a packet before continuing in another harness, model, or session.
Basierend auf der SOC-Berufsklassifikation
| name | miller-orientation |
| description | Use when starting a Miller task, choosing a Miller tool, or choosing a Miller search mode. |
| user-invocable | true |
| arguments | <what you want to find or do> |
| allowed-tools | mcp__miller__search, mcp__miller__inspect, mcp__miller__context, mcp__miller__trace, mcp__miller__impact, mcp__miller__workspace, mcp__miller__patterns, mcp__miller__content, mcp__miller__edit |
Pick the right Miller tool and mode on the first call instead of guessing or falling back to Grep/Read. One call from the table below answers most tasks.
If you are unsure whether the index is current, start here:
workspace(operation="status")
workspace(operation="health")
Refresh only if stale or health reports a missing/corrupt sidecar:
workspace(operation="refresh")
| If the intent is | First call |
|---|---|
| Find a symbol/identifier | search(query="...") (mode=auto) |
| Find docs/config prose | search(query="...", mode="content") |
| Find source-body text | search(query="...", mode="source") |
| Find text in comments/strings | search(query="...", regions="comment,doc_comment,string_literal") |
| Audit TODO/FIXME/HACK/XXX | search(query="TODO,FIXME,HACK,XXX", mode="markers") |
| Read a file you can name | inspect(target="<file>") |
| Understand a symbol first | inspect(target="<symbol>", depth="overview") |
| Need complete body or complete relations | inspect(target="<symbol>", depth="full") |
| Orient on an unfamiliar area/task | context(query="<task>") |
| Who calls / what does this reach | trace(target="<symbol>") |
| How does A reach B | trace(target="A", mode="path", to="B") |
| Where is this name referenced | trace(target="<symbol>", mode="refs") |
| Scope a refactor / choose tests | impact(target="...") or impact(git=true) |
| Make a localized existing-file text edit | edit(operation="replace_text", target="<file>", old_text="<known-old>", new_text="<new>", match_mode="auto", query="<nearby text>") |
| Prepare a handoff to another harness/model | Use the handoff-out skill |
| Resume from a handoff packet | Use the handoff-in skill |
| List known code shapes | patterns(operation="list"), then follow its Next: actions or search a shown pattern_id |
| Read a large log/report | content(operation="import", path=...) -> content(operation="search", query="...") -> bounded content(operation="read", source_id="...", line=...) |
| Work in another registered repo | workspace(operation="list") then pass workspace_id="<selector>" |
search hides tests by default; set exclude_tests=false when tests are expected.inspect depth is summary; use depth=overview for the first symbol read, then depth=full only when you need the complete body or complete relation lists.workspace_id=all is for content(operation="search") text audits only, not symbol/code read tools.trace empty results include Next: / JSON next_actions; follow them before treating a missing path or ref as proof.content(operation="read") should use the source_id from content(operation="search") or content(operation="list"); pass the hit's workspace_id for cross-workspace reads.patterns list and no-match results include Next: / JSON next_actions; run patterns(operation="list") before raw route/HTML/JSON/YAML/Markdown greps.trace mode=bridge is provider-scoped to dotnet-web, nextjs, nextjs-api, nuxt, nuxt-api, vue, react, and backend-http; on another stack use mode=refs/path, and use inspect depth=full for callers and callees.name + signature only; docs/literals/broad source text need the row above.edit for localized text changes when query, anchor, or line can avoid a full-file read; use normal patching for broad multi-hunk edits.Keep it compact:
inspect, a follow-up trace, impact before a change, or edit preview for a localized change)Do not reach for Grep/Read/find when a row above fits. Do not read a whole file before inspect.