Skip to content
Back to search

sdlc-skills

MarketplaceWorks with Claude CodeGitHub

Kevin McHale's agentic SDLC plugins for Claude Code

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/kmchale11--sdlc-claude-skills/badge.svg)](/listings/kmchale11--sdlc-claude-skills)

HTML

<a href="/listings/kmchale11--sdlc-claude-skills"><img src="/api/listings/kmchale11--sdlc-claude-skills/badge.svg" alt="findsafeskills safety rating" /></a>

Install

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

In Claude Code:

/plugin marketplace add kmchale11/SDLC-Claude-Skills
/plugin install sdlc@sdlc-skills

Or from your shell:

claude plugin marketplace add kmchale11/SDLC-Claude-Skills
claude plugin install sdlc@sdlc-skills

You also need Node 20+ and the OpenSpec CLI (`npm install -g @fission-ai/openspec`). Then, in your project, run `/sdlc:init`.

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

.claude-plugin/marketplace.json · Manifest · 15 lines
  1. {
  2. "name": "sdlc-skills",
  3. "description": "Kevin McHale's agentic SDLC plugins for Claude Code",
  4. "owner": {
  5. "name": "Kevin McHale",
  6. "url": "https://github.com/kmchale11"
  7. },
  8. "plugins": [
  9. {
  10. "name": "sdlc",
  11. "source": "./plugins/sdlc",
  12. "description": "Spec-driven, test-first delivery with OpenSpec, enforced by hooks"
  13. }
  14. ]
  15. }
README.md · README · 134 lines
  1. # SDLC Claude Skills
  2. A Claude Code plugin that runs spec-driven, test-first delivery as a set of
  3. skills, with hooks that enforce each stage's rules.
  4. You decide **what** to build and **whether it ships**. Claude writes the spec,
  5. the failing tests, the code and the pull request, and an independent reviewer
  6. checks the result against the spec.
  7. ```
  8. /sdlc:propose → you review → /sdlc:tests → /sdlc:build → /sdlc:verify → /sdlc:ship → you merge → /sdlc:archive
  9. spec only the spec must fail tests frozen fresh reviewer opens PR specs updated
  10. ```
  11. Built on [OpenSpec](https://github.com/Fission-AI/OpenSpec). Pairs with
  12. [Agentic-SDLC](https://github.com/kmchale11/Agentic-SDLC), which runs the same
  13. pipeline in GitHub Actions.
  14. ## Install
  15. In Claude Code:
  16. ```
  17. /plugin marketplace add kmchale11/SDLC-Claude-Skills
  18. /plugin install sdlc@sdlc-skills
  19. ```
  20. Or from your shell:
  21. ```bash
  22. claude plugin marketplace add kmchale11/SDLC-Claude-Skills
  23. claude plugin install sdlc@sdlc-skills
  24. ```
  25. You also need Node 20+ and the OpenSpec CLI (`npm install -g @fission-ai/openspec`).
  26. Then, in your project, run `/sdlc:init`.
  27. ## The skills
  28. | Skill | Stage | What happens | Who can start it |
  29. |---|---|---|---|
  30. | `/sdlc:init` | — | Sets up OpenSpec, the `spec-driven-tdd` schema and `.sdlc/config.json` | You |
  31. | `/sdlc:propose <idea>` | propose | Writes proposal, specs, design, a test plan mapping each scenario to a test ID, and tasks. No code | You or Claude |
  32. | `/sdlc:tests <change>` | tests | Writes a failing test per scenario, then runs the **red check** | You |
  33. | `/sdlc:build <change>` | build | Implements until green, then runs the **green check** | You |
  34. | `/sdlc:verify <change>` | verify | The `spec-verifier` subagent, which never saw the build, writes a PASS/FAIL review | You |
  35. | `/sdlc:ship <change>` | ship | Pushes a branch and opens the PR | You |
  36. | `/sdlc:archive <change>` | — | After merge, folds the change into `openspec/specs/` | You |
  37. | `/sdlc:status` | — | Where things stand and the next command | You or Claude |
  38. Every stage after propose is user-only, so Claude can't move work forward
  39. without you.
  40. ## What's enforced
  41. Three layers, from soft to hard:
  42. **1. Skills** tell Claude what each stage is for.
  43. **2. A PreToolUse hook** blocks file edits and shell commands that break the
  44. current stage's rules, and tells Claude why:
  45. | Stage | Claude can change | Blocked |
  46. |---|---|---|
  47. | propose | `openspec/` | everything else |
  48. | tests | test folders, the change's `tasks.md` | source code, spec artifacts |
  49. | build | source, new test files, `tasks.md` | the tests written in the tests stage, `openspec/specs/`, approved artifacts |
  50. | verify | `.sdlc/reviews/` | everything else, and git changes |
  51. | ship | git, `openspec/` | source and tests |
  52. | any | — | `gh pr merge`, force-pushes, pushing to `main` |
  53. **3. Checks the next stage depends on:**
  54. - **Red check:**
  55. - every scenario has a test,
  56. - the suite fails, and the failure output names the new tests or their files (so a broken runner doesn't count as red),
  57. - no source file changed.
  58. `build` won't start until it passes.
  59. - **Green check:**
  60. - frozen tests are byte-for-byte unchanged,
  61. - approved artifacts are unchanged,
  62. - all tasks are ticked,
  63. - suite, typecheck and `openspec validate --strict` pass.
  64. `ship` won't start until it passes.
  65. ### Limits
  66. The hook catches direct edits and the common shell writers (`>`, `sed -i`,
  67. `rm`, `git checkout`, …), but a determined agent could still get around a
  68. pattern match. The checks compare file hashes, so they catch changes however
  69. they were made. For a hard guarantee on shared branches, run the same gates
  70. in CI. That is what [Agentic-SDLC](https://github.com/kmchale11/Agentic-SDLC)
  71. is for.
  72. ## Configuration
  73. `.sdlc/config.json`, written by `/sdlc:init` and committed:
  74. ```json
  75. {
  76. "srcDirs": ["src"],
  77. "testDirs": ["test"],
  78. "testCommand": "npm test",
  79. "typecheckCommand": "npm run typecheck",
  80. "testFilePattern": "\\.(test|spec)\\.[cm]?[jt]sx?$|_test\\.(py|go)$|^test_.*\\.py$"
  81. }
  82. ```
  83. `.sdlc/state.json` (the current stage) and `.sdlc/reviews/` are local and git-ignored.
  84. ## Repository layout
  85. ```
  86. .claude-plugin/marketplace.json marketplace "sdlc-skills"
  87. plugins/sdlc/
  88. .claude-plugin/plugin.json
  89. skills/<stage>/SKILL.md one skill per stage
  90. agents/spec-verifier.md independent reviewer subagent
  91. hooks/hooks.json PreToolUse guard
  92. scripts/sdlc.mjs stage state, red and green checks
  93. scripts/guard.mjs the hook
  94. scripts/check-test-plan.mjs scenario → test traceability
  95. schema/spec-driven-tdd/ OpenSpec schema with the test-plan artifact
  96. test/ guard unit tests and an end-to-end flow
  97. ```
  98. ## Development
  99. ```bash
  100. npm test # guard rules and a full propose→ship flow in a temp project
  101. npm run validate # claude plugin validate, for the plugin and the marketplace
  102. claude --plugin-dir ./plugins/sdlc # try local edits without installing
  103. ```

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.