EXTENSIONS

Issue lifecycle model with 9 bundled Claude Code skills for DDD, TDD,

Run AI coding agents (Claude, Codex) with structured output for CI

Sweep open pull requests a fork contributes to an upstream GitHub repo. Composable steps — list, review (one agent per batch), sync, verdicts, summary — so an orchestrator (Ceph, a foreman agent, a workflow, or you) plans the work and delegates it, while `sweep` remains as the one-call convenience. Queryable per-PR and run-summary resources; the sweeper prompt ships as a bundled asset.

Reviews prose against a fixed set of editorial principles and returns

Deterministic disclosure checks for candidate public drafts, producing non-echoing blocker and review findings for agent and human review.

Read-only ArchiMate architecture review through complete revision-coherent model evidence, selectable specialist and domain lenses, deterministic synthesis, and durable user-directed dialogue.

Narrow GitHub.com same-repository pull-request boundary: preflight, plan, apply, and read-back verification. Draft is requested per packet rather than fixed by the adapter. No merge, release, repository administration, PR update, or arbitrary endpoint capability.

Forge-neutral immutable change packets for reviewed repository delivery. Records exact Git base/head identity, validation evidence, requested draft submission metadata, a canonical SHA-256 digest, and derived readiness. Performs no network, Git, forge, approval, merge, or release operation.

Fact-check one Markdown document through a provider-neutral, read-only adapter. Kiro CLI is the default provider; Claude CLI is supported when explicitly selected. Repository and optional web capabilities are recorded separately, with no provider fallback.

Structured git comparison — produces per-file diffs between refs and auto-discovers structural patterns to flag deviations in new code. Two methods: diff (structured breakdown) and check (pattern-based review with ready-to-post markdown comments).

Human-in-the-loop review canvas — serve a local web form the human gates while the agent keeps working. list mode renders a long candidate list as a curation grid (configurable option-scale + per-item comment) for migrations, triage, and cleanup approvals; doc mode is a pastebin-style markdown editor (editor left, live preview right, optional approve/revise/reject verdict) for reviewing generated docs and gating publishes. Detached serve / status / collect / stop lifecycle; every save dual-writes JSON + markdown + a timestamped append-log, and collect records the result into the data model for workflow consumption. Python-stdlib server bundled; private networks only (URL-token guarded).