Rich Lander

Use before adding, growing, or restructuring content in AGENTS.md — keeps it under its 600-line cap, protects operator templates, and limits it to cross-cutting binding rules.

Use when shipping a new dotnet-inspect version; coordinate certification, NuGet and GitHub publication, and matching production and CoreCLR site deployments.

Inspect source mapped by Portable PDB and SourceLink data — map files and member locations, fetch source, or resolve checksum-matched content locally.

Surface a dependency's observable signals — provenance, compatibility, dependency risk, unsafe/PInvoke surface, artifact-text containment, and identifier confusion — to judge how much caution it warrants. Observations, not verdicts.

Map how code connects — implementors and subclasses, extension methods, dependency graphs, reverse callers, and ecosystem integrations. Many outputs are graph-shaped (add --mermaid).

Output formats, -D/-S discovery and selection, -Q query-capability discovery, value projection, @ categories, and output limits shared across commands.

Inspect packages from private and custom NuGet feeds safely — select and map sources, use credential providers or explicit configuration, diagnose authentication failures, and reason about source-bound caches and offline operation.

Whole-assembly call-graph leverage ranking and performance triage for libraries (experimental).

Discover and use version-matched agent skills from NuGet packages, and persist selected skills in a repository only when the user explicitly requests it.

Query raw ECMA-335 assembly metadata — discover table and heap sections, inspect decoded rows and handles, address table or heap coordinates, and project bounded machine-readable results.

Reconstruct a method or type as C# and IL — decompiled source, annotated source with hidden facts, raw IL, fidelity levels, and IL-offset lookup. Use --offline to prohibit package and PDB network acquisition.

Judge whether code is sound and safe to call — its exception surface (throws, catches, exception types) and the unsafe operations in a method body.

Decide whether a change or upgrade is safe — API-surface diffs, behavioral diffs (allocations, exceptions), feature switches, and version resolution.

Find and share evidence instead of guessing about .NET packages, libraries, APIs, dependencies, source, performance, and version changes.

Use for opt-in Deep Inspect run triage: test lane failures, census artifacts, pinned-subset drift, and routing findings into issues or tracker updates.

Use when a dotnet-inspect change needs expensive evidence outside normal PR CI, or when preparing release certification; coordinate Deep Inspect lanes (full slow tests, cross-platform certification, IL round-trip sweep, corpus sensors, package discovery, validity scans, and analysis census).

Use for daily chores in dotnet-inspect; inspect main CI and staging health, scheduled certification, performance, runtime-pin, security, and dependency runs, then classify and prioritize any action.

Use for authored-source correspondence corpus maintenance: verify the vendored CIVIL/EVIL snapshot for drift against upstream, re-harvest or re-pin it, run the offline benchmark, and decide when to advance the pin.