github-sync git:20260821.19573ecA

Immutable. This exact content is served forever at /api/v1/blob/f131f6a81074e91e.

---
name: github-sync
description: Read or synchronize bounded GitHub issues, threads, and pull requests through the local authenticated gh CLI or any compatible hosted connector; writes require approval and readback.
runx:
  category: ops
---

# GitHub Sync

Define one bounded synchronization between local Runx state and a GitHub
repository. The skill makes direction, resource set, filters, content identity,
scope, cursor, and approval posture explicit before the selected GitHub
transport is allowed to read or mutate GitHub.

The default `github-sync` runner performs the requested pull or push. `plan`
is the explicit no-effect runner; `pull` and `push` remain explicit execution
lanes for compatible callers. Pulls need no human approval because they are
bounded reads. A push carries one to eight typed mutations, uses one approval
bound to the exact batch digest, and returns an individual status, sync
reference, and mutation digest for every item. It closes only on independent
GitHub readback for every applied item. A partial or unknown item is deferred
for reconciliation and is never silently retried. No GitHub token enters this
package.

## Composes

<!-- Generated from the native execution closure; run pnpm core-skills:composes:generate. -->

- `data-store#append_event`
- `data-store#read_projection`

## When to use it

Use `github-sync` when an operator needs a reproducible pull or push contract for
issues, pull requests, threads, or a mixed `batch`—especially when cursor state must survive
across runs. Use `plan` only when the requested outcome is a plan. Use `pull`
for an explicitly selected read lane. Use `push` for one already-decided issue,
pull-request, or thread mutation. Batches are bounded and best-effort, not
atomic: if one item fails, the result identifies each item and the operator
must reconcile before retrying. Use `issue-triage` to decide what an issue means
and `issue-to-pr` to govern an implementation lane.

Do not use a plan as evidence that remote state was read or changed. Only the
native `runx.provider.operation.v1` result from `pull` or `push` is provider
evidence. Do not silently turn a denied push into a pull.

## How it works

1. Validate the exact `owner/name` repository, direction, resource kind, bounded
   filters, maximum result count, and requested scope.
2. A `pull` plan requires read scope and records the exact resources a provider
   may fetch.
3. A `push` plan requires write scope and one to eight typed mutations. It
   rejects unknown fields and oversized content, then returns
   `ready_for_approval`; planning itself does not open or satisfy that gate.
4. Optional `plan_and_append_cursor` composes the canonical `data-store` skill
   to append the bounded plan and read the projection back. That proves local
   cursor persistence, not GitHub synchronization; this package does not own a
   second cursor database or storage adapter.
5. The default runner selects the same bounded read or write path from the
   validated direction. `pull` executes the plan with `repo.read`. `push`
  hashes the exact mutation set
   with the native digest tool, requests approval, executes it with
   `repo.write`, verifies the returned digest, and reads the resource back.
   Both runners use the native provider lane.

## Inputs and result

- `repo` is exactly `owner/name`.
- `direction` is `pull` or `push`.
- `resources` contains the issue, PR, thread, or mixed `batch` selector and a
  limit no greater than 100. A push has one to eight `mutations` entries with `ref`, `op`, and an
  exact typed `payload`. Issue and PR changes use `op: update`; new thread
  comments use `op: comment`. Bodies are capped at 65,536 characters. Pulls
  return compact metadata and body digests by default; set `include_body: true`
  only when the full bounded body is needed.
- `scope` is the requested read or write scope.

The `runx.github_sync.v1` plan records exact direction, provider operation,
scope, filters, blockers, approval posture, cursor state when used, and the
explicit absence of remote effects. Execution additionally emits the native
provider-operation packet; it does not rewrite the planning packet to imply a
remote effect.

## Stop conditions

- Refuse malformed repositories, unknown resource kinds, unbounded selectors,
  limits above the contract, more than eight mutations, unknown mutation fields, or
  content beyond the declared bounds.
- Refuse push when write scope is missing; do not degrade silently.
- Do not treat local cursor persistence as remote provider readback.
- Resolve the repository from explicit input or the current checkout before
  selecting transport. Prefer the already-authenticated local `gh` path; use a
  compatible hosted grant when local GitHub authority is unavailable or the
  operator explicitly binds hosted transport.
- Refuse missing, ambiguous, wrong-provider, or under-scoped authority. Never
  treat an available hosted grant as evidence of the intended repository.
- Do not claim comments, labels, issues, or PRs were read or changed until the
  native provider operation proves the expected access, operation, and readback.

## Example

A caller wants to pull the next 50 open issues after cursor `2`. Planning can
persist that cursor locally; `pull` then performs `issues.read` through the
configured grant and returns the provider result. To close issue 241, a push
carries `ref: issues/241`, `op: update`, and
`payload: {state: closed, state_reason: completed}`. Runx hashes and approves
that exact mutation, applies only that issue update, and reads issue 241 back.