Doc Bridge

Practical Doc Bridge paths for humans, agents, PR gates, and MCP — one repository, two audiences.

View raw Markdown · llms.txt · For agents

Documentation

Doc Bridge keeps one repository useful to people and coding agents. Pick the shortest path:

Start

GoalPage
Install + 60s proofGetting started
Guided installInstall and run
Machine-first entryFor agents · site route /for-agents
PR freshness gateGate and CI · Marketplace

Build workflows

GoalPage
Index + resolve ownershipIndex and query · Query
MCP for Cursor / ClaudeMCP for agents · MCP
Memory digest → draft docsMemory pipeline
Every CLI commandCLI map · CLI reference
Config sketchesExamples
Optional chat / RAGChat and RAG · Ollama demo
Study and measurementStudy overview and anonymized data · Study protocol v1
Controlled task suiteStudy task suite v1
Controlled study runnerStudy runner v1
Provider CLI adapterStudy provider CLI v1
Round 1 instrumentationInstrumentation plan v1 · adjudicated smoke ledger
Longitudinal measurementsStudy metrics v1
Study verificationStudy verification v1
A/B baseline resultA/B baseline result 2026-08-31 · analysis
Documentation quality auditDocumentation audit v1
Current documentation audit snapshotAudit round 2026-08-31
Controlled pilot evidencePilot round 2026-08-31

How it works

Repository docs


ak-docs index  ──►  .doc-bridge/index.json + llms.txt

      ├─► Human guides (Fumadocs / Docusaurus / md)
      ├─► Agent handoff (CLI / MCP)
      └─► PR gate (Marketplace Action)
  1. Index — map ownership from corpus + config
  2. Resolve — deterministic handoffs (startHere, editRoots, checks)
  3. Gate — fail stale context before agents run
  4. Promote — optional memory → draft docs loop

Product vs reference

Machine surfaces: llms.txt · llms-full.txt · raw Markdown

Ecosystem

Part of AgentsKit — next to AgentsKit, Registry, Chat, and Playbook.

The AgentsKit ecosystem

Build the agent. Then take it all the way.

On this page