Data

Tools

Indexes

Applications

Experiments

Awesome

An open API service indexing awesome lists of open source software.

https://github.com/snowfluke/my-tech-lead-flow

AI Agent Skills library to facilitate technical lead in a software project
https://github.com/snowfluke/my-tech-lead-flow

ai-agent claude-code coding-agents skill

Last synced: 13 days ago
JSON representation

AI Agent Skills library to facilitate technical lead in a software project

Awesome Lists containing this project

README 

          
# Claude Code Skills



Personal skill library for Claude Code. The core of this repository is a

**technical lead pipeline**: a sequence of skills that takes a product idea

from raw user stories all the way to a reviewed, documented, deployable

codebase. Each stage is a skill that reads the artifacts produced by the

previous stage and writes the next one, so the project documentation stays

consistent and every decision is traceable back to an acceptance criterion.



## The technical lead pipeline



The pipeline runs in order. Steps 1 and 3 are manual (the human supplies or

edits the source material); every other step is a skill invoked with `/`.

Step 0 is optional, for projects that start from an idea rather than a written

spec.



| Step | Stage | Skill | Produces |

| ---- | ----- | ----- | -------- |

| 0 | Discover (optional) | `product-discovery` | The step-1 requirements table, elicited by interview, when no user stories or AC exist yet |

| 1 | Capture requirements | (manual or step 0) | A Markdown table of user stories, acceptance criteria, and a sprint breakdown |

| 2 | Groom | `grooming` | Engineering's gaps before commitment. Two modes: blast every question into a file with recommendations (to take to a BA/PO), or interview the user one question at a time and refine the US/AC from their answers |

| 3 | Refine | (manual) | Edits to the US and AC based on grooming output |

| 4 | Format | `us-ac-formatter` | `docs/business/`: `user-story.md`, `sprint-breakdown.md`, and per-sprint Gherkin acceptance criteria |

| 5 | Specify | `technical-spec` | `docs/technical-specs/`: numbered `NN-topic.md` files plus `_index.md`, including ad-hoc trailing specs. Grills on tech stack and tooling first |

| 6 | Spec the API | `api-spec` | `docs/api-specs/`: numbered `NN-topic.md` files plus `_index.md`, deriving every operation, its inputs/outputs, errors, and access rules from the module definitions and data model, in the project's chosen protocol (REST/GraphQL/gRPC/SOAP) |

| 7 | Plan work | `task-breakdown` | `docs/TASK_BREAKDOWN.md`: sprint-by-sprint, role-assigned cards, with frontend/backend wiring as its own owned card. Grills on team shape and scope first |

| 8 | Set up project | `tech-lead-setups` | The Sprint 0 scaffold: folder structure, architectural patterns, commit hooks, tooling config, and endpoint/page stubs returning mock responses. Grills first, then executes |

| 9 | Set standards | `coding-standard` | `CODING_STANDARD.md` and `CODE_REVIEW_CHECKLIST.md`. Grills on every open rule first |

| 10 | Init GitHub project | `github-project-init` | Issues from every task card (assigned, labelled, milestoned, in the board Backlog), the Projects v2 Kanban board, `dev`/`test`/`main` branches with protection, issue and PR templates, CI quality and build workflows, dependabot, and a release template |

| 11 | Plan deployment | `deployment-plan` | `DEPLOYMENT_PLAN.md`: an operational runbook. Grills on infrastructure first |

| 12 | Orient newcomers | `project-docs` | `README.md`, `GLOSSARY.md`, `DEVELOPMENT_SCENARIO_GUIDE.md`, `ONBOARDING_GUIDE.md` |

| 13 | Document failures | `troubleshooting` | `TROUBLESHOOTING.md`: symptom-indexed guide, scaffolded from the architecture seams |

| 14 | Init agent manual | `init-claude` | `CLAUDE.md`: a dense agent operating manual distilled from the specs, standard, task breakdown, and deployment plan. Grills on the gaps the docs leave open first |

| 15 | Build | (manual + skills below) | The implementation |

| 16 | Review | `code-review` | A structured PR review against the standards, run in an isolated git worktree |



### How the stages connect



- `product-discovery` (step 0) is the on-ramp when no requirements exist yet: it

  interviews and emits the step-1 table, then hands off to `grooming`.

- `grooming`, `us-ac-formatter`, and `technical-spec` all treat `docs/business/`

  as the product source of truth.

- `technical-spec` is the architectural keystone: the tech stack, data model,

  and module boundaries it fixes are what `api-spec`, `task-breakdown`,

  `tech-lead-setups`, `coding-standard`, and `deployment-plan` build on.

- `api-spec` turns the module definitions and data model into concrete operation

  contracts; `task-breakdown` and `tech-lead-setups` cite those contracts when

  carving cards and stubbing endpoints.

- `tech-lead-setups` reads the technical specs to build the scaffold that

  `coding-standard` then describes and `code-review` enforces.

- `coding-standard` writes the two documents that `code-review` consumes as its

  source of truth.

- `github-project-init` turns `task-breakdown`'s cards into GitHub issues on a

  board, and its CI quality gate and PR template reference the coding standard.

- `deployment-plan` writes the runbook that `troubleshooting` references rather

  than duplicates.

- `init-claude` runs last, distilling the specs, standard, task breakdown, and

  deployment plan into the `CLAUDE.md` an agent reads before building.

- Every doc cross-links its siblings instead of restating their content, so

  facts live in exactly one place.



## Supporting skills



General-purpose skills that assist the pipeline at any stage:



- `grill-me`, `grill-with-docs`: stress-test a plan or design before building.

- `diagnose`: disciplined debugging loop. Feeds confirmed incidents back into

  `troubleshooting`.

- `tdd`: red-green-refactor build loop.

- `to-prd`, `to-issues`, `triage`: convert context into tracked work.

- `improve-codebase-architecture`, `zoom-out`, `prototype`, `simplify`.

- `git-commit`, `setup-pre-commit`, `git-guardrails-claude-code`.

- `stop-slop`: remove AI writing patterns from prose.

- `handoff`, `caveman`, `skill-creator`, `write-a-skill`.