{
  "markdown": "# SDLC Strict Waterfall Kit\n\nDoc-first Software Development Lifecycle methodology for AI coding agents.\nWorks natively on six harnesses — **Claude Code** (CLI + VS Code + Cursor\nextension + JetBrains + `claude.ai/code` web), **Cursor Agent**, **GitHub\nCopilot CLI**, **OpenAI Codex CLI/App**, **Google Gemini CLI**, and **OpenCode**.\n\n**v2.0.0** · Released 2026-04-23 · MIT · Python 3.11+ only\n\n## What this kit is\n\nA scaffolded 6-phase waterfall SDLC tree (`docs/sdlc/01_requirement/` … `docs/sdlc/06_maintenance/`)\nplus a runtime that keeps AI coding agents honest about it.\n\nAt session start, a **5-rule contract** is auto-injected as `IMPORTANT` context.\nAt phase sign-off, a **two-pass reconciliation gate** blocks drift between\nsigned docs and shipped code. Throughout, an **on-demand skill**\n(`/sdlc-strict-waterfall`) provides the full protocol: Core rules 1-10,\nper-phase workflows, change/migration/hotfix/removal procedures, and a\ntraceability matrix from requirement to commit.\n\nEverything is **measurement-backed**: every NFR carries a numeric target; a\nmetrics subsystem (`tools/sdlc_metrics/`) reports token cost per phase; a\nresidue guard refuses to sign docs with placeholder content; a SHA-pinned hook\nscript refuses to inject a tampered contract.\n\n## Install\n\nPick your agent. One command. Done.\n\n| Agent | Install |\n|---|---|\n| **Claude Code** | `claude plugin marketplace add Electronics-Extreme/claude-sdlc && claude plugin install claude-sdlc@claude-sdlc` |\n| **Codex** | Clone repo → `/plugins` → Search \"SDLC\" → Install |\n| **Gemini CLI** | `gemini extensions install https://github.com/Electronics-Extreme/claude-sdlc` |\n| **Cursor** | `npx skills add Electronics-Extreme/claude-sdlc -a cursor` |\n| **Windsurf** | `npx skills add Electronics-Extreme/claude-sdlc -a windsurf` |\n| **Copilot** | `npx skills add Electronics-Extreme/claude-sdlc -a github-copilot` |\n| **Cline** | `npx skills add Electronics-Extreme/claude-sdlc -a cline` |\n| **Any other** | `npx skills add Electronics-Extreme/claude-sdlc` |\n\nInstall once. Use in every session. On first interaction, the agent loads the SDLC contract, detects a missing scaffold, and offers to bootstrap.\n\n## Quick start (after install)\n\n1. Open your project in the agent\n2. Ask: \"What are the 5 non-negotiable rules?\"\n3. Agent loads SDLC contract → detects no scaffold → offers bootstrap\n4. Accept → `docs/sdlc/01_requirement/` ... `06_maintenance/` created\n5. Start spec in `docs/sdlc/01_requirement/srs.md`\n\n## Manual bootstrap (alternative)\n\nNo agent install needed — use the CLI directly.\n\n```bash\n# Option A — npx (no global install)\ncd ~/Projects/MyApp\nnpx @electronics-extreme/claude-sdlc init --harness claude\n\n# Option B — global install (shorter command)\nnpm install -g @electronics-extreme/claude-sdlc\ncd ~/Projects/MyApp\nclaude-sdlc init --harness claude\n```\n\n## The methodology\n\nSix phases, strictly ordered. No phase skips. Every phase gates on four checks\nplus two-pass reconciliation before sign-off:\n\n```\n   +------------------+      +------------------+      +------------------+      +------------------+      +------------------+      +------------------+\n   | 01_requirement   |  →   | 02_design        |  →   | 03_implementation|  →   | 04_testing       |  →   | 05_deployment    |  →   | 06_maintenance   |\n   |                  |      |                  |      |                  |      |                  |      |                  |      |                  |\n   | SRS              |      | Architecture     |      | Source code      |      | Test report      |      | Release          |      | Change requests  |\n   | signed off       |      | + DB + API       |      | + unit tests     |      | + defects log    |      | + runbook        |      | + incident log   |\n   +------------------+      +------------------+      +------------------+      +------------------+      +------------------+      +------------------+\n```\n\n**Five non-negotiable rules** (always-loaded via SessionStart hook):\n\n1. REFUSE direct code edits — route via CR + doc amendment\n2. No code without a doc parent in signed 01/02/03\n3. No phase skipping — N+1 requires N sign-off\n4. Signed docs are frozen — amend via Post-vX.Y.Z sections\n5. Reconciliation gate before every phase sign-off (two ordered passes)\n\nEach rule ships with an **Excuse / Reality rationalization table** that\nanticipates the specific rationalizations an agent (or human) invents to skip\nthe rule.\n\n**Core rule 7 — TDD inside every slice.** RED-GREEN-REFACTOR per TO-###.\nTests derive from AC-### / TC-### in the signed docs — never invent behavior.\n\n**Core rule 9 — Four gates + two-pass reconciliation**:\n- typecheck + lint + full test suite + reconciliation Pass 1 (spec) before Pass 2 (quality) before sign-off\n\nSee `skill/sdlc-contract.md` for the 5 rules in full; `skill/SKILL.md` for\nCore rules 1-10, per-phase workflows, and protocols.\n\n## What's in the kit\n\n| Area | Contents |\n|---|---|\n| Phase scaffolds | `docs/sdlc/01_requirement/` … `docs/sdlc/06_maintenance/` — 39 artifact templates per phase |\n| Skill | `skill/sdlc-contract.md`, `skill/SKILL.md`, `skill/reconciliation.md`, `skill/traceability-matrix.md`, workflows × 4, protocols × 4, references × 2, required-reads manifest |\n| Hooks | `hooks/session_start.py` (6-harness env-detect + SHA integrity + banner), `run-hook.cmd` / `.sh` launchers |\n| Scripts | 12 Python 3.11+ stdlib scripts: bootstrap, build-archive, check_frontmatter, check_residue, check_task_types, update_contract_sha, sync_wrappers, reconcile, bump_version, release_check, add_frontmatter |\n| Metrics subsystem | `tools/sdlc_metrics/` — phase-aware token analyzer with 6 adapter stubs, 4 report formats (text/JSON/markdown/HTML), SQLite history, budget gate, secrets pre-filter |\n| Config | `config/pricing.yaml`, `budgets.yaml`, `phase-markers.yaml`, `task-types.yaml`, `harnesses.yaml`, `residue-exceptions.yaml` |\n| Schemas | `schemas/doc-frontmatter.schema.yaml`, `task-types.schema.yaml` |\n| Harness adapters | `.claude/`, `.claude-plugin/`, `.cursor/`, `.windsurf/`, `.clinerules/`, `.github/copilot-instructions.md`, `.codex/`, `.opencode/`, `gemini-extension.json`, `AGENTS.md`, `GEMINI.md` |\n| Policy docs | `PRIVACY.md`, `NOTICE.md`, `LICENSE`, `CHANGELOG.md`, `docs/VERSIONING.md` |\n| CI workflows | `.github/workflows/sdlc-tests.yml`, `release.yml`, `airgap.yml` |\n\nTotal: ~290 files, ~830 KB `dist.zip`.\n\n## Verifiable claims\n\nEvery claim in this README is backed by a measurement or CI gate:\n\n| Claim | Verified by |\n|---|---|\n| Zero telemetry, zero analytics, zero remote fetch except user-invoked | `.github/workflows/airgap.yml` runs every script with network denied; exit 0 required |\n| Works on Windows 11 natively | `.github/workflows/sdlc-tests.yml` matrix includes `windows-2022` |\n| Python 3.11 stdlib only | Grep-check on `import` lines + airgap job |\n| Every phase doc has valid frontmatter | `scripts/check_frontmatter.py` in CI |\n| No placeholder residue in signed docs | `scripts/check_residue.py` in CI |\n| Contract + hook integrity (SHA-pinned) | `scripts/update_contract_sha.py --check` in CI |\n| Harness wrappers stay in sync | `scripts/sync_wrappers.py --check` in CI |\n| SemVer correctness on release | `scripts/release_check.py --tag` in release workflow |\n| HTML metrics report is self-contained ≤ 500 KiB | NFR-METRICS-OUT-1 asserted in tests |\n| Metrics adapter ≤ 200 LOC for well-formed transcripts | NFR-METRICS-ADAPT-1 — Claude Code reference is 180 LOC |\n\n## Token budget\n\nDefault per-phase budgets (from `config/budgets.yaml`, editable per adopter):\n\n| Phase | Token hard-cap | Cost hard-cap | Cache floor |\n|---|---|---|---|\n| 01 requirement | 80K | $0.40 | 60% |\n| 02 design | 100K | $0.55 | 65% |\n| 03 implementation | 150K | $0.75 | 70% |\n| 04 testing | 90K | $0.45 | 70% |\n| 05 deployment | 60K | $0.30 | 55% |\n| 06 maintenance | 70K | $0.35 | 60% |\n\nRun `python3 tools/sdlc_metrics/analyze.py budget-check <session.jsonl>\n--phase 3` to enforce. Exit 1 on breach.\n\n## Versioning + stability\n\nStrict **SemVer 2.0**. The kit follows a bootstrap pattern — existing\nprojects stay on the kit version they installed; new features land in new\nbootstrapped projects. See `docs/VERSIONING.md`.\n\n## Contributing\n\nThis kit is distributed as-is. Changes go through the kit's own SDLC:\n\n1. File a CR in `docs/sdlc/06_maintenance/change_requests.md` (use the template)\n2. Approve per class (minor = TL; major = TL + PO; breaking = TL + PO + Security)\n3. Amend Phase 1/2/3 docs as declared in the CR\n4. Implement per Core rule 7 (TDD)\n5. Close reconciliation — Pass 1 spec compliance, then Pass 2 code quality\n6. Bump VERSION, promote CHANGELOG, tag\n7. `release.yml` publishes `dist.zip`\n\n## License\n\nMIT — see `LICENSE`.\n\n## Privacy\n\nZero telemetry. Zero analytics. Zero remote fetch except user-invoked\n`pricing-sync` and harness marketplace installs. See `PRIVACY.md`.\n",
  "bytes": 8891,
  "sha": "58d61ab4331482a30541efd9866297c28129b0fe1e27da85a5e40afb1d079539",
  "repo_slug": "electronics-extreme/claude-sdlc",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/plg_electronics_extreme_claude_sdlc_61344f76/readme"
}