Repair Dotnet Build

STOP - Read before ANY tool call

1. Read {{GUARDRAILS_PATH}}
2. Read _shared/sdd-artifacts/SESSION.md; load session-state for $Cwd
3. If the relevant gate is not approved: STOP - ask user (pt-BR) - do NOT Write/Shell
4. SDD/develop skills: after ONE step/task, STOP session - handoff only
5. This skill body is English; user-facing prompts may be (pt-BR)

Step -1 - Gate check (report in chat before continuing)

Gate check:
[ ] guardrails.mdc read
[ ] SESSION.md read; session-state loaded
[ ] PIPELINE.md read (SDD skills only)
[ ] User confirmed current action (sim)
-> If any unchecked: STOP

Skill: repair-dotnet-build

Trigger

Invoke when the user asks for: /repair-dotnet-build, fix build, /repair-dotnet-build, or when build/test failures block progress.

Arguments (optional):

Do not require an external CI build ID or any PAT.

Outcome

Structured diagnosis, proposed fixes with rationale, fixes applied only after user confirmation, local re-validation, then handoff to /commit if the user wants to commit.

Lazy-load

Never by default: do not preload all references/*.md or the full dotnet-guidelines tree. Load one references/<section>.md per Process step (SKILL-REFERENCE-RETRIEVAL.md).

Process

Read references/<section>.md for heuristics and templates — not full reference.md.

Step -1b - Caveman Mode (Full cap)

1. Read {{SDD_ROOT}}/preferences.json (create { "caveman_mode": false, "caveman_level": "full" } if missing).
2. If caveman_mode is false: continue without compression.
3. If true: load {{TOOLKIT_ROOT}}/skills/_shared/caveman/CAVEMAN.md; apply Full participation cap + prefs caveman_level (Lite skills never escalate); show once: [Caveman] Modo ativo (respostas compactas, level={effective}). Digite caveman off para desativar.
4. Honor caveman on|off|status|lite|full|ultra (and stop caveman / normal mode) during the session.
5. Auto-Clarity + never-compress gates/drafts/paths per CAVEMAN.md.

0. Workspace

Confirm target repository (.sln or test projects). Summarize failure source: local run or pasted CI log.

1. Collect failure evidence

Local (default): follow references/local-commands.md.

dotnet build
dotnet test --no-build

Capture errors: file, line, test name, expected vs actual.

Pasted log: extract compile errors, restore failures, and test failures (references/pasted-ci-log.md).

CI log (optional): if the user pastes a GitHub Actions (or other CI) log, parse failed steps the same way (references/ci-logs.md). Do not fetch remote CI via CLI or API - ask for a paste when remote logs are needed.

2. Structured diagnosis

Present using references/diagnosis-template.md:

## Build diagnosis

**Source:** local | pasted log | GitHub Actions
**Branch:** <current branch>
**Failures:** N

### Items
1. [<category>] <summary> - <file>:<line> or <test name>
...

Categories: compile, restore/NuGet, test assertion, configuration, pipeline config (YAML only if user supplied log).

3. Root-cause investigation

For each item, Read/Grep the codebase. Apply heuristics in references/common-causes.md (culture, timezone, Bogus seed, fixture order, glob in CI YAML).

Load csharp-patterns.md only when editing production or test code.

4. Propose fixes

List each change: file, problem, cause, proposed fix (references/fix-proposal.md). Wait for user confirmation before Edit/Write.

5. Apply and validate

After approval, apply minimal diffs. Re-run:

dotnet build
dotnet test --no-build

Or scoped test filter when the repo is large (see references/local-commands.md).

6. Handoff

When build and targeted tests pass, offer /commit (references/commit-handoff.md, references/related-skills.md).

Do not auto-commit. Do not push unless the user asks via commit skill or explicitly.

Must not

• External ALM/tracker REST, PATs, or org-specific credential stores
• Mandatory external CI API - local reproduction is enough
• Auto-commit or auto-push
• Corporate agent pool names or private feed assumptions without repo evidence

Handoff