Skwaq Development Assistant

An AI agent system (Skwaq - from Lushootseed word for Raven) for researching software vulnerabilities. This skill guides implementation following the project's specification documents, maintains status tracking, and enforces strict code quality standards.

What This Skill Does

Instructions

Before Starting Any Work

1. **Check current status**:

- Read `/Specifications/status.md` to understand current milestone

- Read `/Specifications/VulnerabilityAssessmentCopilot.md` for project context

- Read `/Specifications/ImplementationPlan.md` for implementation steps

2. **Update prompt history** (MANDATORY):

- Open `/Specifications/prompt-history.md`

- Add new entry: "## Prompt N (YYYY-MM-DD)"

- Record user's exact request

- After completing work, add 3-7 bullet points summarizing your actions

Code Implementation Guidelines

**Modularity & Design**:

**Code Quality Standards**:

**Project Structure**:

**Error Handling**:

Testing Requirements (CRITICAL)

**Before Moving to Next Milestone**:

1. Run full test suite: `pytest tests/ --cov=skwaq`

2. Ensure ALL tests pass (no failures anywhere in project)

3. If tests fail:

- STOP immediately

- Think carefully about the code being tested

- Consider necessary setup conditions

- Carefully construct/fix the test to validate functionality

- Fix either the code OR the test

- Re-run ALL tests

4. Only proceed when all tests pass

**Test Writing**:

Development Workflow

**For Each Implementation Step**:

1. **Read the plan**: Check `/Specifications/ImplementationPlan.md` for current step

2. **Implement**: Write code following guidelines above

3. **Run linters**: `pre-commit run --all-files`

4. **Run tests**: `pytest tests/ --cov=skwaq`

5. **Fix failures**: Do NOT proceed if any test fails

6. **Commit**: Use git commit after each completed step

7. **Update status**: Modify `/Specifications/status.md` with progress

8. **Update prompt history**: Add entry to `/Specifications/prompt-history.md`

**Commands**:

```bash

Run all tests with coverage

pytest tests/ --cov=skwaq

Run single test

pytest tests/path/to/test_file.py::TestClass::test_function

Run linters

pre-commit run --all-files

Format code

black .

Check types

mypy .

Run full CI locally

./scripts/ci/run-local-ci.sh

```

Important Technical Constraints

**Autogen Usage**:

**Azure & GitHub Access**:

**Database**:

When Blocked or Unclear

1. **Update status**: Record the blocker in `/Specifications/status.md`

2. **Update prompt history**: Document the issue

3. **Ask for clarification**: Do not guess or assume requirements

4. **Do not proceed**: Wait for user guidance before continuing

Mandatory Status Tracking

**Always maintain `/Specifications/status.md`**:

**Always maintain `/Specifications/prompt-history.md`**:

Code Reuse & Anti-Patterns

**Do**:

**Don't**:

Example Usage

**Starting work**:

1. User: "Implement the knowledge ingestion module"

2. You: Read status.md, check ImplementationPlan.md, update prompt-history.md

3. You: Implement following code guidelines

4. You: Run tests, fix any failures

5. You: Commit, update status.md and prompt-history.md

**Handling test failures**:

1. Test fails

2. STOP immediately (do not continue implementation)

3. Analyze the test and the code being tested

4. Determine root cause

5. Fix code OR fix test

6. Re-run ALL tests

7. Only proceed when all tests pass

Success Criteria