Help us improve
Share bugs, ideas, or general feedback.
Share bugs, ideas, or general feedback.
Share bugs, ideas, or general feedback.
Automatically maintains a structured knowledge base in docs/ and keeps CLAUDE.md in sync. Claude is briefed at session start via a routing table injected by the SessionStart hook.
npx claudepluginhub wuffig-coding-solutions/protocollantScaffold the docs/ knowledge base for this project. Analyzes the codebase and creates initial versions of all managed doc files. Run this once when setting up the protocollant plugin on an existing project.
Manually trigger a full audit and sync of all docs/ files. Use this after a major refactor, when docs have drifted, or to force a CLAUDE.md routing table update.
Automatically maintains a structured knowledge base in docs/ and keeps CLAUDE.md in sync. Claude knows about the docs from the first turn of every session — no on-demand skill trigger needed.
| Mechanism | What it does |
|---|---|
| SessionStart hook | Injects a routing table into Claude's context: which doc to read for which task type |
| PostToolUse hook | Detects architecture/API/bug/dep-relevant file edits and queues a doc update |
| Stop hook | If the queue is non-empty, tells Claude to call @doc-updater before finishing |
@doc-updater subagent | Writes structured entries to the relevant docs, then patches CLAUDE.md |
/docs-init command | One-time scaffold of all doc files from a codebase scan |
/docs-sync command | Manual full audit — use after major refactors |
All files live in docs/:
| File | Purpose |
|---|---|
architecture.md | Module map, layers, constraints |
known-issues.md | Open bugs with symptoms and workarounds |
resolved-bugs.md | Fixed bugs with root cause and fix |
future-work.md | TODOs, feature ideas, roadmap |
decisions.md | ADR-style technical decisions |
api-contracts.md | Endpoint and interface definitions |
data-models.md | DB schema, entities, migrations |
dependencies.md | Key packages and why they were chosen |
performance-notes.md | Bottlenecks, optimizations, benchmarks |
security-notes.md | Sensitive areas, mitigations, threat model |
testing-strategy.md | Test patterns, coverage goals |
deployment.md | Environments, CI/CD, rollback |
integrations.md | Third-party APIs, MCP servers, webhooks |
onboarding.md | Setup steps, gotchas for fresh agents |
changelog.md | Notable changes by date |
This plugin is distributed via agent-plugins. In any Claude Code session:
/plugin install protocollant@agent-plugins
This fetches the plugin from GitHub, registers the hooks, and makes @doc-updater, /docs-init, and /docs-sync available globally.
Open a session in the project you want to document, then run:
/docs-init
This scans the codebase and creates all 15 doc files under docs/. After that, documentation updates happen automatically — no further setup needed.
docs/ and the hooks live in the project. All docs are committed to git and readable by any Claude instance that clones the repo.Why not a skill? Skills are on-demand — Claude has to decide to invoke them. The SessionStart hook injects the routing table unconditionally, so Claude always knows which docs exist and when to read them, without any invocation step.
Why a subagent instead of inline? The doc-updater runs in an isolated context with clear instructions and no session history to distract it. It's faster and more consistent than asking the main Claude instance to "also update the docs" mid-task.
Why docs/ instead of .claude/docs/? These docs are project knowledge, not agent configuration. They should be committed, diffed, and reviewed like any other project documentation.
Modifies files
Hook triggers on file write and edit operations
Uses power tools
Uses Bash, Write, or Edit tools
Share bugs, ideas, or general feedback.
Based on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
llmdoc Claude Code plugin with a minimal workflow: init, update, and use
Automatically maintains CLAUDE.md files as codebases evolve using hooks, agents, and skills
Commands for generating documentation and managing changelogs
CLAUDE.md lifecycle toolkit: initialise, enhance, sync, and modularise CLAUDE.md files with a 150-line cap, automatic chaining, and Karpathy behavioural guardrails.
Automatic documentation-as-code engine. Maintains project documentation as a side effect of development — no manual invocation needed.
Use this agent when you need to analyze a service or codebase component and create comprehensive documentation in CLAUDE.md files. This agent should be invoked after implementing new services, major refactoring, or when documentation needs updating to reflect the current codebase structure. Examples: <example>Context: The user has just implemented a new authentication service and wants to document it properly. user: 'I just finished implementing the auth service, can you document how it works?' assistant: 'I'll use the codebase-documenter agent to analyze the authentication service and create detailed documentation in CLAUDE.md' <commentary>Since the user has completed a service implementation and needs documentation, use the Task tool to launch the codebase-documenter agent to create comprehensive CLAUDE.md documentation.</commentary></example> <example>Context: The user wants to ensure a newly added API module is properly documented for the team. user: 'We need documentation for the new payment processing API I just added' assistant: 'Let me use the codebase-documenter agent to analyze the payment processing API and create proper documentation' <commentary>The user needs documentation for a new API module, so use the codebase-documenter agent to create CLAUDE.md files with setup instructions and architectural notes.</commentary></example>
Persistent memory for Claude Code via mem0. Fully implicit — no manual commands. Stores preferences, architecture facts and known issues across projects using Qdrant (vector) + optional Neo4j (graph).
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge.
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge.
Sign in to claim