agi-tr-report-pipeline

star 0

Orchestrates AGI TR schedule/delay report production using subagents and skills. Use when inserting A~F content into the final report, weaving skill outputs into docs/reports/AGI_TR_일정_최종보고서.MD, or when the user asks for "보고서에 삽입", "최종본 반영", Evidence/VERIFY, or Windows/But-For/Forensic.

macho715 By macho715 schedule Updated 2/17/2026

name: agi-tr-report-pipeline description: Orchestrates AGI TR schedule/delay report production using subagents and skills. Use when inserting A~F content into the final report, weaving skill outputs into docs/reports/AGI_TR_일정_최종보고서.MD, or when the user asks for "보고서에 삽입", "최종본 반영", Evidence/VERIFY, or Windows/But-For/Forensic.

AGI TR Report Pipeline

Orchestrates production of AGI_TR_일정_최종보고서.MD with fixed insert positions and evidence-backed conclusions. Combines subagents and skills; no fabrication—use VERIFY when data is missing.

Skill Map (Single-Responsibility)

Skill Purpose Trigger terms
agi-report-ssot-float SSOT + Critical Path/Float/SV_days fixed paragraph SSOT, Critical Path, Float=0, Cycle 7.00d
agi-delay-decomp TR1TR3 delay as S1S5 Segment Δh table Delay Decomposition, Segment, Δh
agi-rca-pack TR2/TR3: 5-Why (2) + Fishbone 6M (1) 5-Why, Fishbone, 6M
agi-forensic-schedule Windows Analysis + Collapsed As-Built (But-For) Windows Analysis, But-For, Forensic
agi-weather-evidence METAR/NCM → Weather Gate + Evidence Scorecard Weather Gate, METAR, NCM, Evidence
agi-montecarlo-buffer TR4~TR7 P50/P80/P90 + Buffer from segment stats Monte Carlo, P80, Buffer
agi-evidence-web-collector Collect/pin web evidence (AACE, SCL, AHRQ, PMI, OGIMET, NCM) 근거 수집, Evidence Bundle, METAR 원문

Subagents (Cursor)

Subagent Role Trigger Use when
report-weaver Weave skill outputs into report at fixed insert positions "보고서에 삽입", "최종본 반영" Combining A~F into one MD
evidence-judge Verify Evidence Scorecard, conflicts, VERIFY "근거 점수", "VERIFY", "충돌" Independent evidence check
schedule-forensics Windows/But-For logic review, assumptions/limits "Windows Analysis", "But-For" Method and attribution

Pipeline (A~F → Report)

  1. A agi-report-ssot-float → append to "1. SSOT 및 계산 규칙"
  2. B agi-delay-decomp → table under "2. 1·2·3호기 실적"
  3. C agi-rca-pack → "3. 지연 원인 분석"
  4. D agi-forensic-schedule → Windows + But-For tables
  5. E agi-weather-evidence (optionally after agi-evidence-web-collector if METAR/NCM missing)
  6. F agi-montecarlo-buffer → "Contingency" section

Report-Weaver Rules

  • Keep existing headers/section numbers/tables; add only at defined insert positions.
  • Never assert without data; use VERIFY and request 1~3 inputs.
  • Every insert: (1) title of added block, (2) source/date, (3) assumptions/limits.

Evidence Rules

  • W=3: official (METAR, NCM bulletin)
  • W=2: AIS/logs
  • W=1: WhatsApp/verbal
  • On conflict: higher W + time/place fit; if insufficient → VERIFY.

Safety

  • No EVM/schedule terminology mix (SV_days vs EV−PV).
  • But-For = counterfactual → always label "가정:" and uncertainty.
  • 5-Why is not a single-cause tool; cite evidence weight and alternatives.

Full Specs

  • Subagent and skill file contents: docs/일정/SUBAGENTSKILL.MD
  • Web evidence collector and weather patch: docs/일정/SKILLPATCH.MD
Install via CLI
npx skills add https://github.com/macho715/tr_dash --skill agi-tr-report-pipeline
Repository Details
star Stars 0
call_split Forks 0
navigation Branch main
article Path SKILL.md
More from Creator