AgentSkillsCN

eiffel-status

Eiffel Spec Kit 的第 3 阶段。将已审阅的契约分解为包含验收标准的实施任务。配合 /eiffel.tasks 命令使用。

SKILL.md
--- frontmatter
name: eiffel-status
description: Shows current Eiffel Spec Kit phase, evidence files, and workflow status. Use with /eiffel.status command to see where you are in the process.
allowed-tools: Read, Grep, Glob, Bash

/eiffel.status - Workflow Status

Purpose: Show current phase and evidence trail. Answer "where are we?"

Usage

code
/eiffel.status <project-path>

Example:

code
/eiffel.status d:\prod\simple_cache

If no path provided, ask: "Which project? Provide the full path."

Workflow

Step 1: Verify Project Exists

bash
test -d <project-path>/.eiffel-workflow && echo "Workflow initialized" || echo "No workflow - run /eiffel.intent first"

Step 2: Check Evidence Files

Evidence FilePhase
evidence/phase0-intent.txtPhase 0: Intent
evidence/phase1-compile.txtPhase 1: Contracts
evidence/phase2-chain.txtPhase 2: Review
evidence/phase3-tasks.txtPhase 3: Tasks
evidence/phase4-compile.txtPhase 4: Implement
evidence/phase5-tests.txtPhase 5: Verify
evidence/phase6-tests.txtPhase 6: Harden
evidence/phase7-ship.txtPhase 7: Ship

Step 3: Display Status

code
╔══════════════════════════════════════════════════════════════╗
║           EIFFEL SPEC KIT STATUS                             ║
║           Project: <project-path>                            ║
╠══════════════════════════════════════════════════════════════╣
║ Phase 0: Intent      [✓] intent-v2.md approved               ║
║ Phase 1: Contracts   [✓] Compiled                            ║
║ Phase 2: Review      [✓] Synopsis approved                   ║
║ Phase 3: Tasks       [✓] 5 tasks defined                     ║
║ Phase 4: Implement   [→] IN PROGRESS                         ║
║ Phase 5: Verify      [ ] Pending                             ║
║ Phase 6: Harden      [ ] Pending                             ║
║ Phase 7: Ship        [ ] Pending                             ║
╠══════════════════════════════════════════════════════════════╣
║ Current: Phase 4 - Implementation                            ║
║ Next: Complete implementation, then /eiffel.verify           ║
╚══════════════════════════════════════════════════════════════╝

Step 4: Drift Detection

Check for signs of drift:

  • Contracts modified since Phase 2?
  • Tasks.md out of sync?
  • Evidence files missing?
  • Skills modified during workflow?
code
DRIFT WARNINGS:
- [!] Contracts changed since review (re-run /eiffel.review)
- [!] phase4-compile.txt missing (compile not verified)
- [!] Skill files modified since workflow start (queue changes for after ship)

Step 4b: Skill Version Lock Check

Why: Modifying skills mid-workflow causes drift. Discovered improvements should be queued for after ship.

Check skill modification times vs workflow start:

bash
# Get workflow start time (phase0 evidence)
WORKFLOW_START=$(stat -c %Y <project-path>/.eiffel-workflow/evidence/phase0-intent.txt 2>/dev/null)

# Check if any eiffel-* skills were modified after workflow started
find ~/.claude/skills/eiffel-* -name "SKILL.md" -newer <project-path>/.eiffel-workflow/evidence/phase0-intent.txt

If skills were modified during workflow:

code
SKILL DRIFT WARNING:
- Skills modified during workflow execution
- Changes made: [list files]
- Recommendation: Complete current workflow, then formalize skill improvements

Best Practice: Create <project-path>/.eiffel-workflow/skill-improvements.md to queue discovered improvements:

markdown
# Skill Improvements Discovered During This Workflow

## To Apply After Ship
1. [Improvement description]
2. [Improvement description]

## Applied During Workflow (drift)
1. [What was changed and why]

Step 5: Suggest Next Action

Based on current state, suggest the next command:

code
Suggested next command:
  /eiffel.implement <project-path>   (continue implementation)

Or if implementation done:
  /eiffel.verify <project-path>      (run tests)

Context Management (RLM Pattern)

This skill focuses ONLY on: <project-path>

DO NOT:

  • Read files outside this project directory
  • Load entire ecosystem into context to check status
  • Keep large file contents in working memory

DO:

  • Only read .eiffel-workflow/ directory and evidence files
  • Use Task tool with Explore agent if you need ecosystem-wide status

Example - Ecosystem Query:

code
Need to check if dependencies have workflow issues? Don't read all projects.
Instead: Task(subagent_type=Explore) → "Does simple_mml have a .eiffel-workflow directory?"

The sub-agent searches and returns ONLY what you need. Your context stays focused on this project.

Output

Always show:

  1. Phase checklist with status
  2. Current phase
  3. Drift warnings (if any)
  4. Suggested next command