devproc

Plugin devproc from marketplace local-plugins in plwhite/claudeplugins. Commands read from the repo's manifest — 13 days ago.

Skills (10)

• feature-checkpoint

  Sync all documentation and feature tracking to reflect the current state of the in-progress feature

• feature-design

  Design and plan a feature — move it to in progress and write its design and sub-task plan

• feature-end

  Complete a feature — finalise documentation and move it to Completed

• feature-init

  Initialises the feature model for this project. Run this once to enable feature workflows.

• feature-spec

  Create a new feature and write its specification

• internal-docs-prune

  Prune internal Claude-facing docs (CLAUDE.md, nested CLAUDE.md files, NOTES.md, .claude/rules/*.md) — spawns internal-docs-reviewer, auto-applies redundant/stale findings without content loss, and escalates judgment findings (interactive) or defers them (unattended)

• review-branch

  Code review scoped to files changed in the current feature branch — runs simplicity, general, and nitty agents over the diff, auto-applies code-level findings until convergence, and escalates architectural findings for user confirmation

• review-component

  Code review scoped to a described component or area — resolves the description to files, then runs simplicity, general, and nitty agents; auto-applies code-level findings until convergence; escalates architectural findings for user confirmation

• review-docs

  Docs review scoped to the whole documentation set, the current change, or a described area — runs architectural, accuracy, style and agent docs agents, auto-applies minor fixes inside a single doc until convergence, and escalates significant findings (always including boundary findings) for user con

• review-full

  Full-codebase code review — runs simplicity, general, and nitty agents over all files, auto-applies code-level findings until convergence, and escalates architectural findings for user confirmation

Subagents (12)

• code-review-architectural

  Use this agent to review the architectural quality of a codebase or set of files. It checks that the overall structure is valid, internally consistent, and represents the best approach for the problem — flagging module boundary violations, misplaced responsibilities, fragile coupling, and deviations

• code-review-general

  Use this agent for a general correctness and robustness review of code. It checks that the code does what it is supposed to do, handles errors and edge cases correctly, and does not have obvious performance problems. It produces findings only; it does not modify files. Pass the files or diff to revi

• code-review-nitty

  Use this agent for a low-level code quality review focused on readability, naming, comments, and fine-grained robustness. It digs into the details that general review passes overlook: misleading names, missing or wrong comments, confusing control flow, and subtle robustness issues within individual

• code-review-simplicity

  Use this agent to identify where code can be simplified, consolidated, or removed. It looks for unnecessary complexity, duplicated logic, dead code, over-engineering, and abstractions that add indirection without value. It produces findings only; it does not modify files. Pass the files or diff to r

• dev-process-manager

  Top-level orchestrator for the devproc feature workflow. Run it as the session agent (`claude --agent dev-process-manager`, or the container's `claude-run --manager`) when you want feature work driven semi-autonomously: it spawns teammates to implement sub-tasks, reviews their work, keeps the high-l

• docs-review-accuracy

  Use this agent to check that documentation still matches the code and its behaviour. Given a change, it finds every document that describes the changed thing and checks each is still correct: commands, paths, options, behaviour, examples, file lists. Without a change, it checks the documents in scop

• docs-review-agent-docs

  Use this agent to review a project's agent docs: the documents written for Claude and other AI tooling rather than for people (`CLAUDE.md` at any depth, `NOTES.md`, `.claude/rules/*.md`). It checks that they still match the repository (file lists, paths, commands, descriptions of components), that n

• docs-review-architectural

  Use this agent to review the structure of a documentation set as a whole. It checks that every document is reachable from the entry points, that links resolve, that each document states its purpose up front, that architecture guides cover the components and their relationships, that each fact has on

• docs-review-style

  Use this agent to review how documentation reads. It checks procedural form (imperative, prerequisites before the commands that need them, the reason for non-obvious steps, complete command blocks, success indicators), clarity (ambiguity, buried points, overlong sentences), and consistency (detail k

• feature-design-reviewer

  Use this agent to review a feature design and its sub-task plan before a human reads it. It checks that the design addresses every requirement and records why the decisions were made, that each sub-task carries auditable sign-off criteria derived from the agreed strategy, and that anything blocking

• feature-spec-reviewer

  Use this agent to review a feature specification before a human reads it. It checks that the plan file's `## Spec` and `## Sign-off strategy` sections are complete, clear, and readable, that the delivery criteria are explicit and auditable, and that anything blocking the move to design is surfaced.

• internal-docs-reviewer

  Use this agent to review the internal, Claude-facing documentation of a repository — root and nested `CLAUDE.md` files, `NOTES.md`, and `.claude/rules/*.md` — for content that has gone stale, become redundant with a canonical source elsewhere, or is of borderline enough value that a human should dec