agentleFS
boyingliu01boyingliu01/xp-gate/skills/sprint-flow/AGENTS.md

xp-gate / sprint-flow

AGENTS.md4 starsChanged 6 months agoMITView on GitHub

Generated: 2026-11-04 Commit: eb2097b0 Branch: sprint/2026-11-03-02 Version: 0.37.0.0 v2.0 Compact Redesign (Issue #290): Merged from 11 phases to 6. PREP = old ISOLATE + AUTO-ESTIMATE. DESIGN = old THINK + PLAN. BUILD = old BUILD. VERIFY = old REVIEW + FEEDBACK. SHIP = old SHIP + LAND. CLOSE = old USER ACCEPTANCE + CLEANUP.

The file

We split this AGENTS.md into the rules it gives an agent

7 rules. About 0 are common, 6 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈6

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  • could be automatic: a setting could enforce it
  1. Distinctive:

    delphi-review HARD-GATE in Phase 2/6 DESIGN: all three expert roles must execute with three distinct requested model IDs, then aggregate …

    line 67
  2. Very specific:

    Do NOT skip delphi-review in Phase 2/6 DESIGN — HARD-GATE blocks implementation.

    line 74
  3. Distinctive:

    Do NOT use parallel build mode unless explicitly requested.

    line 75
  4. Distinctive:

    Do NOT enter Phase 2/6 DESIGN (PLAN) without completing DESIGN (brainstorming).

    line 76
  5. Distinctive:

    Do NOT implement before design is APPROVED — Phase 2/6 DESIGN must reach Delphi consensus first.

    line 77
  6. Distinctive:

    Do NOT merge an Emergent Requirement into the original Sprint silently — capture it via the template in Phase 6/6 CLOSE.

    line 78
  7. Distinctive:

    Do NOT terminate Delphi review before ≥90% consensus or 5 rounds, whichever first.

    line 79

≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
AGENTS.md

SKILLS/SPRINT-FLOW KNOWLEDGE BASE

Generated: 2026-11-04 Commit: eb2097b0 Branch: sprint/2026-11-03-02 Version: 0.37.0.0

OVERVIEW

6-phase development pipeline (v2.0 compact redesign, Issue #290): PREP → DESIGN → BUILD → VERIFY → SHIP → CLOSE. Phase 3/6 BUILD default build mode is ralph-loop (REQ-level iteration, 40-67% token savings vs parallel). HARD-GATE between DESIGN (2/6) and BUILD (3/6): design must pass Delphi review (≥90% consensus) before any coding.

v2.0 Compact Redesign (Issue #290): Merged from 11 phases to 6. PREP = old ISOLATE + AUTO-ESTIMATE. DESIGN = old THINK + PLAN. BUILD = old BUILD. VERIFY = old REVIEW + FEEDBACK. SHIP = old SHIP + LAND. CLOSE = old USER ACCEPTANCE + CLEANUP.

STRUCTURE

skills/sprint-flow/
├── SKILL.md              # 6-phase pipeline definition (canonical, v2.0)
├── AGENTS.md             # This file (mirrored to 7 other locations — DO NOT edit mirrors)
├── evals/                # Evaluation test cases
├── evolution-history.json
├── evolution-log.md
├── references/
│   ├── phase-1-prep.md                    # Phase 1/6: PREP (worktree isolation + sizing)
│   ├── phase-2-design.md                  # Phase 2/6: DESIGN (brainstorming + delphi-review HARD-GATE)
│   ├── phase-3-build.md                   # Phase 3/6: BUILD (ralph-loop default + TDD + test-align)
│   ├── phase-4-verify.md                  # Phase 4/6: VERIFY (code-walkthrough + QA + feedback)
│   ├── phase-5-ship.md                    # Phase 5/6: SHIP (PR + merge + deploy + canary)
│   ├── phase-6-close.md                   # Phase 6/6: CLOSE (UAT + cleanup)
│   ├── force-levels.md                    # Phase forcing rules
│   ├── orchestration-rules.md             # Agent dispatch, context inheritance, transition rules
│   └── components/                        # Reusable phase building blocks
└── templates/
    ├── auto-estimate-output-template.md
    ├── auto-estimate-learning-log.md
    ├── pain-document-template.md
    ├── sprint-progress-template.md
    ├── sprint-summary-template.md
    └── emergent-issues-template.md

WHERE TO LOOK

TaskLocationNotes
Pipeline definitionSKILL.md6 phases with HARD-GATE between DESIGN (2/6) and BUILD (3/6)
PREP phasereferences/phase-1-prep.mdworktree isolation + sizing pass
DESIGN phase + HARD-GATEreferences/phase-2-design.mdbrainstorming + autoplan + delphi-review → specification.yaml
BUILD phasereferences/phase-3-build.mdralph-loop (default) vs parallel
VERIFY phasereferences/phase-4-verify.mdcode-walkthrough + QA + retro + learn
SHIP phasereferences/phase-5-ship.mdPR + merge + deploy + canary
CLOSE phasereferences/phase-6-close.mdUAT + cleanup
Force-level rulesreferences/force-levels.mdDefines when each phase becomes mandatory
Orchestration rulesreferences/orchestration-rules.mdAgent dispatch, context inheritance, transition gates
Templatestemplates/Auto-estimate, sprint progress/summary, pain doc, emergent issues

THE 6 PHASES (v2.0)

PhaseNameMaps From (old 11)Key ActionHard Gate
1/6PREPISOLATE + AUTO-ESTIMATEWorktree isolation + sizing—
2/6DESIGNTHINK + PLANbrainstorming → autoplan → Delphi consensusHARD-GATE: design must reach ≥90% Delphi consensus
3/6BUILDBUILDralph-loop (default) + TDD + test-spec-alignment—
4/6VERIFYREVIEW + FEEDBACKcode-walkthrough + QA + retro + learnHARD-GATE: feedback-log.md must exist
5/6SHIPSHIP + LANDPR → squash-merge → deploy → canary—
6/6CLOSEUSER ACCEPT + CLEANUPManual UAT → emergent issues → cleanupHARD-GATE: UAT mandatory manual

CONVENTIONS

  • ralph-loop is Phase 3/6 BUILD default. Each REQ runs in a clean context, saving 40-67% tokens.
  • delphi-review HARD-GATE in Phase 2/6 DESIGN: all three expert roles must execute with three distinct requested model IDs, then aggregate ≥90% consensus within 5 rounds. Unapproved or unverifiable → BLOCK coding.
  • learn is called twice: once per REQ in Phase 3/6 (ralph-loop internal) and once in Phase 4/6 VERIFY (Sprint-level retro).
  • Phase isolation: each phase has explicit entry/exit criteria documented in its references/phase-*.md file.
  • Emergent Requirements discovered in Phase 6/6 CLOSE (USER ACCEPTANCE) are explicitly captured via templates/emergent-issues-template.md — never silently merged.
  • Auto-detection: Phase 2/6 DESIGN uses src/npm-package/lib/ui-detector.ts to pick the right tech-stack templates.

ANTI-PATTERNS (THIS PROJECT)

  • Do NOT skip delphi-review in Phase 2/6 DESIGN — HARD-GATE blocks implementation.
  • Do NOT use parallel build mode unless explicitly requested. Ralph-loop is the default for a reason.
  • Do NOT enter Phase 2/6 DESIGN (PLAN) without completing DESIGN (brainstorming).
  • Do NOT implement before design is APPROVED — Phase 2/6 DESIGN must reach Delphi consensus first.
  • Do NOT merge an Emergent Requirement into the original Sprint silently — capture it via the template in Phase 6/6 CLOSE.
  • Do NOT terminate Delphi review before ≥90% consensus or 5 rounds, whichever first.

UNIQUE STYLES

  • 6 phases (v2.0 Issue #290) — compact design with merged phases (1-6 numbering)
  • HARD-GATE between DESIGN and BUILD enforced both in SKILL.md and Claude Code plugin's PreToolUse hook.
  • Per-REQ clean context in ralph-loop = the core efficiency mechanism.
  • Tech-stack auto-detection via --type and --lang flags or ui-detector.ts.

COMMANDS

/sprint-flow "开发用户登录"                                     # Full 6-phase pipeline
/sprint-flow "开发用户登录" --type web-nextjs --lang typescript # Pin tech stack
/sprint-flow "开发用户登录" --phase build-only                  # Skip design (advanced)
/sprint-flow "开发用户登录" --mode parallel                     # Legacy all-at-once (NOT default)
/sprint-flow "开发用户登录" --stop-at design                    # Stop after DESIGN phase
/delphi-review "开发用户登录" --type web-nextjs --lang typescript

NOTES

  • Integrates: brainstorming, autoplan, delphi-review, TDD, test-specification-alignment, qa, design-review, benchmark, systematic-debugging, retro, learn, finishing-a-development-branch, land-and-deploy.
  • ralph-loop's internal learnings are persisted via progress.log (permanent vs contextual classification).
  • Phase 4/6 VERIFY calls gstack/learn for Sprint-level retrospective.
  • Phase 6/6 CLOSE cleanup behavior is governed by docs/plans/2026-06-06-sprint-branch-cleanup-design.md.
  • This AGENTS.md is the canonical version. 7 byte-identical mirrors exist. Mirrors are updated by scripts/copy-skills.sh. Do NOT edit them by hand.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.