Prompt file imported from pavanthakur/XYDataLabs.OrderProcessingSystem (.github/prompts/XYDataLabs-context-audit.prompt.md). Fill in {{DB_PASSWORD}} before use. Copyright stays with the author.

Context Audit — Detect Stale AI Context

Audit all AI context files against the actual codebase to detect drift. Do NOT ask for permission — run all checks and report findings.

Checks to Perform

1. Project Table Accuracy

• Read XYDataLabs.OrderProcessingSystem.sln to get the actual project list.
• Compare against the project table in .github/copilot-instructions.md (§2).
• Compare against the project table in /memories/repo/dotnet-conventions.md.
• Flag: missing projects, removed projects still listed, wrong descriptions.

2. Package and Framework References

• Search all .csproj files for PackageReference entries.
• Check /memories/repo/dotnet-conventions.md for any package names mentioned (e.g. MediatR, AutoMapper).
• Flag: packages referenced in memory but not in any .csproj, or packages in .csproj not reflected in memory.

3. Test Project Structure

• List tests/ directory contents.
• Compare against test project table in .github/copilot-instructions.md.
• Flag: new test projects not listed, removed projects still listed.

4. CQRS/Architecture Pattern Accuracy

• Search for ICommand, IQuery, IDispatcher, IMediator, MediatR in .cs files.
• Verify that memory and instructions correctly describe the CQRS pattern in use.
• Flag: any mention of MediatR if it's not actually used (or vice versa).

5. Directory Layout

• List root-level directories and compare against the directory tree in .github/copilot-instructions.md (§3).
• Flag: directories that exist but aren't listed, or listed directories that no longer exist.

6. Prompt Discoverability

• Search .github/workflows/README-*.md, .github/prompts/README.md, and .github/copilot-instructions.md for prompt references.
• Verify reusable prompts are mentioned consistently where operators are told how to run repository workflows.
• Flag: newly added prompts missing from quick tips, workflow READMEs, or prompt indexes.

7. Memory Freshness

• Read all files in /memories/repo/ and relevant user memory files that define stable practices.
• For each file, check if its key facts are still accurate against the codebase and current repo structure.
• Flag: any factual statements that contradict current codebase state.

8. Secret Hygiene

• Search AI-facing files (.github/copilot-instructions.md, .github/instructions/*.md, .github/prompts/*.md, /memories/repo/*.md) for literal passwords, secrets, or connection strings with embedded credentials.
• Search tracked config samples (Resources/Configuration/*.json, Resources/Docker/*.yml, Resources/Docker/*.yaml) for literal credentials that should be replaced with safer local-dev patterns or environment-variable overrides.
• Flag: any secret-like values stored in memory, instruction files, reusable prompts, or tracked config samples.
• Treat placeholders such as $sqlPwd, {{DB_PASSWORD}}, or Key Vault references as acceptable.

9. ADR Directory

• List docs/architecture/decisions/ contents.
• Check that architect-patterns.md ADR references are consistent with actual ADR files.

10. Architecture Status Surface Consistency

• Read the active-status surfaces:
  • ARCHITECTURE-EVOLUTION.md
  • docs/guides/development/phase10-tool-prerequisites.md
  • docs/internal/phase10-implementation-checklist.md
  • docs/learning/curriculum/1_MASTER_CURRICULUM.md
  • docs/learning/curriculum/README.md
  • docs/internal/AZURE-PROGRESS-EVALUATION.md
  • docs/learning/implementation-notes/implementation-notes-days-29-38.md or the active implementation-notes file
  • docs/DEVELOPER-OPERATING-MODEL.md
  • .github/instructions/curriculum.instructions.md
  • .github/copilot-instructions.md
• Compare the current phase complete/next-phase snapshots across those files.
• Flag: any mismatch in completed phase count, active phase, next phase, or freeze/closeout wording that would mislead humans or Copilot.
• Specifically verify that Phase 10.1 local baseline reconciliation is marked complete, Phase 10.2 remains the next active slice, and the local-tooling / local-Docker guidance still matches the current repo contract.

Output Format

Present findings as a table:

Severity levels:

• HIGH — Actively feeds wrong context into every session (e.g. wrong project name, removed package still referenced)
• MEDIUM — Missing information that could cause confusion (e.g. new project not listed)
• LOW — Minor drift, cosmetic (e.g. description wording slightly outdated)

After the table, list specific edits needed to fix each HIGH and MEDIUM issue.