Tester Plan (Fase 1)

Create and write test-plan.md in the CLI format. No test inspection. No production or test file edits.

Two modes. Pick one before reading code.

If the user does not say which mode, ask once. Do not inspect the codebase to guess.

Fase 1 is high effort. Do not switch models. Before writing test-plan.md, reason through scope, one technique per surface, pairwise only when factors interact, edge cases, and the error contract. Log [fase 1] effort: high.

Instructions

1. Load workflow context:
   • Parent checklist: .cursor/skills/tester-agent/workflow/template.md
   • Report contract: .cursor/skills/tester-agent/reports.md
   • Prefer .tester-agent/*.md if present; else .cursor/skills/tester-agent/templates/workflow/
   • Prompt shape: .cursor/skills/tester-agent/templates/prompts/test-plan.md
   • Plan content: .cursor/skills/tester-agent/test-design.md
2. Resolve the backlog item (id or path). Read the full markdown.
3. Apply the chosen mode:
   • pre-implementation: do not open application code, diffs, or tests. Derive scope and surfaces from the backlog. Where the backlog does not specify a path, field, or error contract, record an open question.
   • post-implementation: inspect application code relevant to the item (and an optional git diff). Do not open or search existing unit, integration, or e2e tests.
4. Build rawMarkdown using test-design.md. State the mode in Summary. Include testable checks, one technique per surface, and edge cases (happy path, applicable robustness, domain cases, error contract; pairwise only when two or more factors jointly decide behavior).
5. Integration test per acceptance criterion; unit test only for pure helpers. In pre-implementation, these are planned tests for the intended behavior, not claims that the code already exists.
6. Create the report dir and write the plan with the CLI helper (required):

BODY=$(mktemp)
# write rawMarkdown (headings only, no # Test Plan header) into $BODY
node .cursor/skills/tester-agent/scripts/write-test-plan.mjs \
  --out ".tester-agent/reports/${RUN_ID}/test-plan.md" \
  --run-id "${RUN_ID}" \
  --target "${TARGET_LABEL}" \
  --created-at "$(date -u +%Y-%m-%dT%H:%M:%S.%3NZ)" \
  --body-file "$BODY"

TARGET_LABEL examples: backlog:074-foo (pre-implementation), backlog:074-foo+code or origin/main...HEAD (post-implementation).

7. Mark Fase 1 done on the workflow checklist.

After each of these, tell the user and append the same line to $REPORT_DIR/progress.log (see tester-agent "Zichtbare voortgang"):

• [fase 1] start: {pre-implementation|post-implementation} backlog {id}
• [fase 1] gelezen: {path} for the backlog file and, in post-implementation, each application file opened
• [fase 1] geschreven: {path to test-plan.md}

Hard rules

• Do not edit application code, test files, config, or lockfiles.
• Only write test-plan.md (via the script) under the run report dir.
• Do not write report.md / report.json / report.csv in this phase.
• Do not run tests in this phase.
• If documentation-only: automated tests N/A + review checklist.
• In pre-implementation, do not read or infer behavior from application code.
• If the backlog (or, in post-implementation, the code) is insufficient: list open questions; do not invent requirements.

Required rawMarkdown headings

## Summary
## Risks
## Edge Cases
## Acceptance Criteria
## Test Strategy
## Open Questions

The script prepends the CLI header (# Test Plan, Run, Target, Created).

Fill those headings as specified in test-design.md (scope, checks, technique per surface, pairwise only when needed).

Done when

• Mode chosen: pre-implementation or post-implementation
• Backlog item fully read
• post-implementation only: relevant app code inspected, existing tests not opened
• pre-implementation only: application code not opened
• Each known surface has one technique (checklist, pairwise, transitions, or security)
• test-plan.md written via write-test-plan.mjs
• Workflow checklist Fase 1 checked