test-plan-score

Test Plan Scorer

Score an existing test plan using the 5-criteria quality rubric (Specificity, Grounding, Scope Fidelity, Actionability, Consistency). This is the user-facing entrypoint for rubric evaluation.

Usage

/test-plan-score <feature_dir>

Examples:

• /test-plan-score kagenti_agent_templates
• /test-plan-score mcp_catalog

Inputs

From arguments

Parse $ARGUMENTS to extract:

1. Feature directory (required): path to directory containing TestPlan.md

Process

Step 0: Python dependencies

Install the test-plan package (makes all scripts importable):

(cd $(git -C ${CLAUDE_SKILL_DIR} rev-parse --show-toplevel) && uv sync --extra dev)

If installation fails, inform the user and do NOT proceed. Once installed, all Python scripts will work from any directory.

Step 1: Read Test Plan and Source Strategy

1. Read <feature_dir>/TestPlan.md
2. Read frontmatter to extract source_key:

   (cd $(git -C ${CLAUDE_SKILL_DIR} rev-parse --show-toplevel) && uv run python scripts/frontmatter.py read <feature_dir>/TestPlan.md)
   

3. Fetch the source strategy from Jira using the source_key:

   mcp__atlassian__getJiraIssue with issueIdOrKey=<source_key>
   

   If MCP is unavailable, check for a local strategy file in artifacts/strat-tasks/<source_key>.md. If neither is available, warn the user and proceed — grounding and scope fidelity will be scored based on plan consistency only.

Step 2: Score (fork)

Read the score agent prompt from skills/test-plan-review/prompts/score-agent.md.

Launch a forked score agent with substitutions:

• {FEATURE_DIR} = feature directory path
• {TEST_PLAN_PATH} = <feature_dir>/TestPlan.md
• {STRATEGY_TEXT} = raw strategy description text from Step 1
• {CALIBRATION_DIR} = skills/test-plan-review/calibration/

Step 3: Present Results

Parse the score agent's output and present the results directly to the user:

## Test Plan Score — {feature_name}

### Rubric Scores

| Criterion | Score | Notes |
|-----------|-------|-------|
| Specificity | {n}/2 | {brief rationale} |
| Grounding | {n}/2 | {brief rationale} |
| Scope Fidelity | {n}/2 | {brief rationale} |
| Actionability | {n}/2 | {brief rationale} |
| Consistency | {n}/2 | {brief rationale} |

**Total: {sum}/10**

### Verdict

{If >= 8, no zeros: "**Ready** — proceed to test case generation"}
{If = 7, no zeros: "**Revise** — minor improvements needed. Re-run via `/test-plan-create` flow to apply auto-revision, or invoke the internal `test-plan.review` workflow from automation."}
{If < 7 or any zero: "**Rework** — significant issues. Re-run via `/test-plan-create` flow for remediation, or use automation that calls internal `test-plan.review`."}

### Grounding Cross-Reference
{Include the full grounding cross-reference table from the scorer}

What This Skill Does NOT Do

• Does NOT write a TestPlanReview.md file
• Does NOT trigger auto-revision
• Does NOT modify the test plan
• For scoring + auto-revision, use /test-plan-create flow (which calls internal test-plan.review)

$ARGUMENTS