@shivaedev/pr-watch 0.1.1 → 0.1.2

package/CHANGELOG.md

@@ -1,5 +1,11 @@
 # Changelog
 
+## 0.1.2 - 2026-10-24
+
+### Changed
+
+- Require Effect and its companion packages at exactly 4.0.5. Generated applications use the same versions.
+
 ## 0.1.1 - 2026-10-18
 
 ### Fixed

package/README.md

@@ -23,7 +23,7 @@ Each successful observation compares a PR with its saved snapshot. The first obs
 Install with Node 24 or later and the repository's Effect peer version:
 
 ```sh
-npm install @shivaedev/pr-watch effect@4.0.4 @effect/platform-node@4.0.4 @effect/platform-node-shared@4.0.4
+npm install @shivaedev/pr-watch effect@4.0.5 @effect/platform-node@4.0.5 @effect/platform-node-shared@4.0.5
 export GH_TOKEN=YOUR_READ_ONLY_TOKEN
 pr-watch --repo acme/project 42 57 --state ./watch-state.json
 pr-watch --repo acme/project --fleet --interval 60 --state ./fleet-state.json
@@ -68,7 +68,3 @@ The state file records discovered open PR numbers and the last successfully comp
 Transitions that reverse between observations, including while the watcher is down, are not recovered. For example, leaving and rejoining the merge queue before the next observation produces no `dequeued` event. Stdout delivery is at-least-once across a crash: a crash after output but before the state replacement can repeat a line. There is no event history or acknowledged output protocol.
 
 The exported `StateFile.layer` requires Effect `FileSystem`, and `GitHubHttp.layer` requires Effect `HttpClient`. The CLI supplies `NodeFileSystem.layer` and a `FetchHttpClient` layer that refuses redirects. HTTP and file failures remain typed `WatchError` values.
-
-### Evidence
-
-Synthetic HTTP tests cover conditional 304 responses, all eight events, already-stuck PRs at startup, terminal pruning across restarts, queue entry with a cleared auto-merge request, confirmed queue drops and rebuilds, manual auto-merge disablement, native IO injection and typed failures, checks independent of PR ETags, fleet discovery and pagination, quota slowdown, failed requests, and clean restart comparison. A CLI subprocess test verifies that pending mergeability diagnostics stay on stderr while stdout contains JSON event lines. The packed consumer runs the installed executable against a synthetic server. These tests establish the REST and GraphQL handling and installed package behavior; they do not establish live GitHub or Monitor adoption.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@shivaedev/pr-watch",
-  "version": "0.1.1",
+  "version": "0.1.2",
   "description": "One REST watcher for pull request transitions across a repository",
   "homepage": "https://github.com/ShivaeDev/platform/tree/main/packages/pr-watch#readme",
   "bugs": {
@@ -39,30 +39,28 @@
     "!src/**/*.test.ts",
     "!src/**/*.spec.ts",
     "!src/**/test-support",
-    "docs",
     "CHANGELOG.md",
     "README.md"
   ],
   "dependencies": {
-    "@shivaedev/effect-service": "0.4.0"
+    "@shivaedev/effect-service": "0.4.1"
   },
   "devDependencies": {
-    "@effect/platform-node": "4.0.4",
-    "@effect/platform-node-shared": "4.0.4",
-    "@effect/vitest": "4.0.4",
-    "@shivaedev/effect-test": "0.6.0",
-    "@shivaedev/quality": "0.25.0",
-    "@shivaedev/test-story": "0.7.1",
+    "@effect/platform-node": "4.0.5",
+    "@effect/platform-node-shared": "4.0.5",
+    "@effect/vitest": "4.0.5",
+    "@shivaedev/effect-test": "0.7.2",
+    "@shivaedev/quality": "0.34.0",
+    "@shivaedev/test-story": "0.18.2",
     "@types/node": "24.19.2",
-    "@typescript/native": "npm:typescript@7.0.2",
-    "effect": "4.0.4",
-    "typescript": "npm:@typescript/typescript6@6.0.2",
-    "vitest": "5.0.3"
+    "effect": "4.0.5",
+    "typescript": "7.0.2",
+    "vitest": "5.0.5"
   },
   "peerDependencies": {
-    "@effect/platform-node": "4.0.4",
-    "@effect/platform-node-shared": "4.0.4",
-    "effect": "4.0.4"
+    "@effect/platform-node": "4.0.5",
+    "@effect/platform-node-shared": "4.0.5",
+    "effect": "4.0.5"
   },
   "engines": {
     "node": ">=24"

package/docs/north-star.md

@@ -1,7 +0,0 @@
-# One process for pull request observations
-
-An orchestrator needs to see merges, conflicts, failed checks and merge queue drops across a repository. One REST watcher shares discovery and queue requests, uses conditional responses, and keeps successful snapshots across restarts.
-
-Evidence wins over speed. A failed or rate-limited request cannot stand in for a negative observation. Check runs and commit statuses belong to the observed head; queue discovery comes from the queue refs, and a suspected drop is confirmed against the PR’s own queue entry and auto-merge state. Native Effect services own composition and interruption.
-
-Durability means comparison between successful snapshots. It does not recover transitions that reversed between observations. Stdout is at-least-once across a crash. Event history, acknowledged delivery, PR changes and automated reactions belong outside this package.

package/docs/roadmap.md

@@ -1,18 +0,0 @@
-# Roadmap
-
-## Built
-
-- A REST executable for explicit PRs or repository-wide fleet discovery, using effect-service services.
-- ETag response reuse, quota pacing, queue-ref observations and eight transition events, distinguishing queue entry from disabled auto-merge.
-- Startup events for already-stuck auto-merges, terminal PR pruning and GraphQL confirmation of suspected queue drops.
-- Native Effect filesystem and HTTP services with typed failures.
-- Atomic snapshot persistence and comparison across clean restarts.
-- Synthetic HTTP tests and packed executable consumer scenarios.
-
-## Next
-
-Work Fleet checks and delivery integrations can consume the event lines. Application adoption needs its own evidence.
-
-## Maintainer decisions
-
-No additional delivery protocol or event history is part of this package.