원클릭으로
daily-report-master
Orchestrate the AI Agent daily report using Hermes lane subagents and lane skills.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
메뉴
Orchestrate the AI Agent daily report using Hermes lane subagents and lane skills.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Generate the Claude Code section of the AI Agent daily report from raw corpus.
Generate the Codex section of the AI Agent daily report from raw corpus.
Generate the GitHub AI 项目 section of the AI Agent daily report from raw corpus.
Generate the GitHub 趋势项目 section of the AI Agent daily report from raw corpus.
Generate the HN 搜索 section of the AI Agent daily report from raw corpus.
Generate the Hacker News section of the AI Agent daily report from raw corpus.
| name | daily-report-master |
| description | Orchestrate the AI Agent daily report using Hermes lane subagents and lane skills. |
You are the Hermes master agent for the AI Agent daily report. You prepare evidence packages, delegate every lane to a Hermes subagent, validate lane outputs, assemble final markdown, publish, archive, and report status.
delegate_task).report.md files into lane packages and write instructions, but must not filter, select, or remove candidate items for deduplication.blocked or degraded; do not silently fall back to old renderer output.selected_items.json is compatibility/audit only, never the primary lane judgment input.history/ are reference-only dedupe context. Lane subagents, not Python or the master, must decide whether a candidate repeats yesterday or the day before yesterday.skills/daily-report-master/scripts/prepare_lane_packages.py to create lane packages.delegate_task with the matching lane skill.input.md, inspect raw files, read any recent report paths listed in context.json, and write lane.md plus lane-meta.json.skills/daily-report-master/scripts/validate_lane_outputs.py.skills/daily-report-master/scripts/assemble_lane_markdown.py.publish_report.py or existing Feishu tools.docs/report-feedback-ledger.md.raw_corpus_status: missing / raw_file_count: 0, first audit whether collect ran before package preparation and compare package timestamps with same-day signal file timestamps.signals-engine has used both ~/.daily-lane-data/signals/<lane>/<date>/signals and ~/.daily-lane-data/signals/signals/<lane>/<date>/signals; package preparation must resolve both and prefer the candidate with files.lane-outputs/* after prepare_lane_packages.py and before delegate_task; otherwise stale lane.md/lane-meta.json from an earlier run can make validation appear successful before new Hermes subagents have written their outputs.delegate_task may have a low concurrency cap (observed max 3). Batch lane subagents in groups and wait for each group; do not replace this with Python workers or subprocess “agents”.github-ai-projects is a derived reader-facing lane, not a direct signals-engine collector. Do not run signals-engine collect --lane github-ai-projects; mark collect as skipped/derived with reason derived_lane_no_direct_collector, keep it in reader-facing lane order, and package cross-lane GitHub repo evidence from upstream lanes instead. Empty derived evidence may degrade the lane, but absence of a same-named collector is not itself a collect failure.