Agent Skills
› NeverSight/learn-skills.dev
› gpd-suggest-next
gpd-suggest-next
GitHub分析项目当前状态，通过调用 GPD CLI 扫描阶段、计划及阻碍项，生成按优先级排序的下一步行动建议列表。包含上下文信息如进度和阻塞问题，帮助用户快速确定最高效的后续操作。
Trigger Scenarios
用户询问接下来该做什么
需要快速了解项目优先事项
查看项目当前状态和待办事项
Install
npx skills add NeverSight/learn-skills.dev --skill gpd-suggest-next -g -y
SKILL.md
Frontmatter
{
"name": "gpd-suggest-next",
"description": "Suggest the most impactful next action based on current project state",
"context_mode": "project-required",
"allowed-tools": [
"read_file",
"shell",
"grep",
"glob"
]
}
<codex_runtime_notes> Codex shell compatibility:
- When shell steps call the GPD CLI, use /Users/charlie/.gpd/venv/bin/python -m gpd.runtime_cli --runtime codex --config-dir ./.codex --install-scope local instead of the ambient
gpdon PATH. - If you intentionally need the repo environment, keep the runtime pin:
GPD_ACTIVE_RUNTIME=codex uv run gpd .... </codex_runtime_notes>
This is the fastest way to answer "what should I do next?" without reading through progress reports.
Step 1: Run the suggest CLI
SUGGESTIONS=$(/Users/charlie/.gpd/venv/bin/python -m gpd.runtime_cli --runtime codex --config-dir ./.codex --install-scope local --raw suggest)
if [ $? -ne 0 ]; then
echo "ERROR: suggest-next failed: $SUGGESTIONS"
echo ""
echo "Try $gpd-progress for manual project status."
exit 1
fi
Step 2: Parse and present
Parse the JSON output. It contains:
suggestions: Array of{priority, action, command, reason}sorted by priority (1=highest)context: Object withcurrent_phase,status,progress_percent,paused_at,active_blockerssuggestion_count: Total number of suggestions
Step 3: Display
## What's Next
[For each suggestion, ordered by priority:]
**{priority}. {command}**
{reason}
---
Context: Phase {current_phase} | {progress_percent}% complete | {status}
If there's only one suggestion, present it as the clear next step:
## >> Next Up
**{command}**
{reason}
<sub>`/clear` first -> fresh context window</sub>
If there are blockers, highlight them before suggestions:
## !! {active_blockers} Blocker(s)
Resolve before continuing:
- {blocker description}
---
<success_criteria>
- suggest-next command executed successfully
- Suggestions presented in priority order
- Context shown (phase, progress, status)
- Blockers highlighted if present
- User has a clear next action </success_criteria>
Version History
- e0220ca Current 2026-07-05 22:21
