agentleFS
Postpartum-genushyacinthus29Postpartum-genushyacinthus29/dotnet-skills/skills/dotnet-code-analysis/SKILL.md

dotnet-code-analysis

Skill13 starsChanged 8 months agoMITView on GitHub

Use the free built-in .NET SDK analyzers and analysis levels with gradual Roslyn warning promotion. Use when a .NET repo needs first-party code analysis, `EnableNETAnalyzers`, `AnalysisLevel`, or warning-as-error policy wired into build and CI.

The file

Show its rules and text

We split this skill into the rules it gives an agent

15 rules. About 0 are common, 6 are distinctive, and 9 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 9

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Never disable or remove TreatWarningsAsErrors or WarningsAsErrors if the project has set them.

    line 33
  2. Very specific:

    Never add <NoWarn> or #pragma warning disable for warnings the user chose to treat as errors, unless the user explicitly approves the …

    line 34
  3. Very specific:

    Never silently downgrade severity in .editorconfig (e.g.

    line 35
  4. Distinctive:

    If warnings-as-errors breaks the build — fix the code.

    line 36
  5. Distinctive:

    If warning volume is too large to fix in one pass — report count and categories to the user and ask which to tackle first.

    line 37
  6. Distinctive:

    Start with SDK analyzers before third-party packages.

    line 60
  7. Very specific:

    Add to WarningsAsErrors: CS8019;CS0219;CS0168.

    line 92
  8. Very specific:

    Ask: "Which categories next — Nullability, Performance, or Reliability?" Add selected IDs to WarningsAsErrors, fix, repeat.

    line 104
  9. Very specific:

    Run dotnet build, count warnings by ID.

    line 126
  10. Distinctive:

    Ask which batch to tackle.

    line 128
  11. Very specific:

    Add those IDs to WarningsAsErrors.

    line 130
  12. Distinctive:

    Never skip the ask step. The user decides the pace.

    line 133
  13. Very specific:

    Run dotnet build, return status: configured or status: improved.

    line 145
  14. Very specific:

    If repo defers analyzer policy to another build layer, return status: not_applicable.

    line 146
  15. Distinctive:

    Update plan after each iteration.

    line 167

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

.NET Code Analysis

Trigger On

  • the repo wants first-party .NET analyzers
  • CI should fail on analyzer warnings
  • the team needs AnalysisLevel or AnalysisMode guidance
  • the repo needs a gradual Roslyn warning promotion strategy

Do Not Use For

  • third-party analyzer selection by itself
  • formatting-only work

Inputs

  • the nearest AGENTS.md
  • project files or Directory.Build.props
  • current analyzer severity policy

Hard Rules for AI Agents

Non-negotiable. Violating these undermines the user's explicit intent.

  1. Never disable or remove TreatWarningsAsErrors or WarningsAsErrors if the project has set them. Do not comment them out, set to false, wrap in a condition, or add <TreatWarningsAsErrors>false</TreatWarningsAsErrors> to make the build pass.
  2. Never add <NoWarn> or #pragma warning disable for warnings the user chose to treat as errors, unless the user explicitly approves the suppression.
  3. Never silently downgrade severity in .editorconfig (e.g. error to warning or none) to make a build succeed.
  4. If warnings-as-errors breaks the build — fix the code. If the fix is too large, ask the user whether to defer that warning ID.
  5. If warning volume is too large to fix in one pass — report count and categories to the user and ask which to tackle first. Do not unilaterally disable the policy.

Workflow

flowchart TD
    A[Start] --> B{New or legacy project?}
    B -->|New| C[TreatWarningsAsErrors=true immediately]
    B -->|Legacy| D[dotnet build, count warnings by ID]
    D --> E{"< 30 warnings?"}
    E -->|Yes| F[Fix all, then enable TreatWarningsAsErrors]
    E -->|No| G[Report counts to user, ask which batch first]
    G --> H[Add selected IDs to WarningsAsErrors]
    H --> I[Fix that batch, verify build]
    I --> J{More batches?}
    J -->|Yes| G
    J -->|No| F
    C --> K[Set AnalysisLevel latest-recommended]
    F --> K
    K --> L[Promote security CA3xxx/CA5xxx to error in .editorconfig]
    L --> M[Validate: build + CI green]
  1. Start with SDK analyzers before third-party packages.
  2. Detect project maturity: new or existing/legacy.
  3. Enable EnableNETAnalyzers, AnalysisLevel, AnalysisMode in Directory.Build.props.
  4. Apply the right warning promotion strategy (see below).
  5. Per-rule severity goes in repo-root .editorconfig.
  6. dotnet build is the analyzer gate in CI.

Warning Promotion Strategy

New Projects

Set these in Directory.Build.props immediately:

  • TreatWarningsAsErrors = true
  • AnalysisLevel = latest-recommended
  • Security category = error in .editorconfig

Fix all warnings before merging.

Legacy Projects — Gradual Promotion

Blanket TreatWarningsAsErrors on a legacy codebase produces hundreds/thousands of errors. An agent cannot fix them all at once — context floods, fix quality drops. Promote in batches.

Phase 1: Trivial Hygiene (start here)

Mechanical fixes, lowest effort:

  • CS8019 — unnecessary using directive (remove it)
  • CS0219 — variable assigned but never used (remove it)
  • CS0168 — variable declared but never used (remove it)
  • CS1591 — missing XML comment for public member (add comment or disable for internal code)
  • CS0612 — obsolete member used, no message (replace with non-obsolete API)
  • CS0618 — obsolete member used, with message (follow migration guidance)

Add to WarningsAsErrors: CS8019;CS0219;CS0168. Fix all, then Phase 2.

Phase 2: Code Quality (ask user which categories)

  • CA2000 — dispose objects before losing scope (Reliability)
  • CA1062 — validate public method arguments (Design)
  • CA1822 — mark members as static (Performance)
  • CA1860 — avoid Enumerable.Any() for length check (Performance)
  • CA1861 — avoid constant arrays as arguments (Performance)
  • CA2007 — consider calling ConfigureAwait (Reliability)
  • CS8600–CS8610 — nullable reference type warnings (Nullability)

Ask: "Which categories next — Nullability, Performance, or Reliability?" Add selected IDs to WarningsAsErrors, fix, repeat.

Phase 3: Security (always promote early)

Set in .editorconfig regardless of project maturity:

[*.cs]
dotnet_analyzer_diagnostic.category-Security.severity = error

Covers CA3001 (SQL injection), CA3002 (XSS), CA3003 (path injection), CA3075 (insecure DTD), CA5350/CA5351 (weak crypto), CA5394 (insecure randomness).

Phase 4: Full Coverage

Once all batches pass, transition to:

<TreatWarningsAsErrors>true</TreatWarningsAsErrors>
<WarningsNotAsErrors>CA1707</WarningsNotAsErrors> <!-- explicit exceptions only -->

Interaction Protocol (legacy codebases)

  1. Run dotnet build, count warnings by ID.
  2. Report summary: "Found 47 CS8019, 23 CA1822, 12 CA2000, 8 CS8600."
  3. Ask which batch to tackle. Recommend starting with Phase 1.
  4. Fix selected batch, verify build.
  5. Add those IDs to WarningsAsErrors.
  6. Report back, ask about next batch.

Never skip the ask step. The user decides the pace.

Bootstrap When Missing

  1. Detect current state:
    • dotnet --info
    • rg -n "EnableNETAnalyzers|AnalysisLevel|AnalysisMode|TreatWarningsAsErrors|WarningsAsErrors" -g '*.csproj' -g 'Directory.Build.*' .
    • dotnet build SOLUTION_OR_PROJECT 2>&1 — count warnings by ID
  2. Classify: new (few/zero warnings) vs legacy (many warnings).
  3. Enable EnableNETAnalyzers, AnalysisLevel, AnalysisMode in MSBuild config.
  4. Apply promotion strategy matching project maturity.
  5. Per-rule severity in repo-root .editorconfig.
  6. Run dotnet build, return status: configured or status: improved.
  7. If repo defers analyzer policy to another build layer, return status: not_applicable.

Deliver

  • explicit, reviewable first-party analyzer policy
  • build-time analyzer execution for CI
  • warning promotion plan matching project maturity

Validate

  • analyzer behavior driven by repo config, not IDE defaults
  • CI reproduces same warnings/errors locally
  • no TreatWarningsAsErrors, WarningsAsErrors, or severity settings removed/weakened without user approval
  • promoted warnings produce build errors, not just IDE hints

Ralph Loop

  1. Plan: analyze state, define target, constraints, risks, execution plan, validation steps.
  2. Execute one step, produce concrete delta.
  3. Review result, capture findings.
  4. Apply fixes in small batches, rerun checks.
  5. Update plan after each iteration.
  6. Repeat until acceptable or only explicit exceptions remain.
  7. Missing dependency: bootstrap or return status: not_applicable.

Required Result Format

  • status: complete | clean | improved | configured | not_applicable | blocked
  • plan: concise plan and current step
  • actions_taken: concrete changes
  • validation_skills: final skills run or skipped with reasons
  • verification: commands, checks, or review evidence
  • remaining: unresolved items or none

Load References

  • references/rules.md
  • references/config.md
  • references/code-analysis.md

Example Requests

  • "Turn on built-in .NET analyzers."
  • "Make analyzer warnings fail the build."
  • "Set the right AnalysisLevel for this repo."
  • "Start treating unused usings and unused variables as errors."
  • "Help me gradually promote Roslyn warnings in my legacy project."
  • "Which warnings should I promote to errors next?"

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.