@bastani/atomic 0.8.15 → 0.8.16-0

package/CHANGELOG.md

@@ -2,6 +2,20 @@
 
 ## [Unreleased]
 
+## [0.8.16-0] - 2026-05-26
+
+### Breaking Changes
+
+- Removed Ralph's `prompt` and `max_loops` compatibility inputs from the bundled workflow; use `objective` and `max_turns` instead ([#1061](https://github.com/flora131/atomic/issues/1061)).
+
+### Changed
+
+- Updated bundled Ralph docs and guide examples for the Goal Runner workflow with goal-ledger receipts, reviewer quorum, repeated-blocker gating, and `max_turns`/`objective` inputs ([#1061](https://github.com/flora131/atomic/issues/1061)).
+
+### Fixed
+
+- Clarified bundled Ralph's bounded `blocker_threshold` behavior ([#1061](https://github.com/flora131/atomic/issues/1061)).
+
 ## [0.8.15] - 2026-05-26
 
 ### Changed

package/dist/builtin/intercom/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@bastani/intercom",
-  "version": "0.8.15",
+  "version": "0.8.16-0",
   "private": true,
   "description": "Atomic extension providing a private coordination channel between parent and child agent sessions. Fork of: https://github.com/nicobailon/pi-intercom",
   "contributors": [

package/dist/builtin/mcp/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@bastani/mcp",
-  "version": "0.8.15",
+  "version": "0.8.16-0",
   "private": true,
   "description": "Atomic extension that adapts MCP (Model Context Protocol) servers into the coding agent. Fork of: https://github.com/nicobailon/pi-mcp-adapter",
   "contributors": [

package/dist/builtin/subagents/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@bastani/subagents",
-  "version": "0.8.15",
+  "version": "0.8.16-0",
   "private": true,
   "description": "Atomic extension for delegating tasks to subagents with chains, parallel execution, and TUI clarification. Fork of: https://github.com/nicobailon/pi-subagents",
   "contributors": [

package/dist/builtin/web-access/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@bastani/web-access",
-  "version": "0.8.15",
+  "version": "0.8.16-0",
   "private": true,
   "description": "Atomic extension for web search, URL fetching, GitHub repo cloning, PDF/video extraction. Fork of: https://github.com/nicobailon/pi-web-access",
   "contributors": [

package/dist/builtin/workflows/CHANGELOG.md

@@ -6,6 +6,20 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
 
 ## [Unreleased]
 
+## [0.8.16-0] - 2026-05-26
+
+### Breaking Changes
+
+- Removed Ralph's `prompt` and `max_loops` compatibility inputs; use the canonical `objective` and `max_turns` inputs instead ([#1061](https://github.com/flora131/atomic/issues/1061)).
+
+### Changed
+
+- Reworked Ralph into a Goal Runner workflow with a persisted goal ledger, bounded worker turns, parallel structured reviewers, reviewer-quorum completion, repeated-blocker gating, and final status reporting while excluding token-budget behavior ([#1061](https://github.com/flora131/atomic/issues/1061)).
+
+### Fixed
+
+- Hardened Ralph's goal-runner edge cases around base-branch sanitization coverage, worker/reviewer failure handling, cross-turn review context, bounded blocker-threshold handling, and unused ledger state ([#1061](https://github.com/flora131/atomic/issues/1061)).
+
 ## [0.8.15] - 2026-05-26
 
 ### Breaking Changes

package/dist/builtin/workflows/README.md

@@ -308,17 +308,19 @@ Final Markdown research documents are written to dated `research/` paths relativ
 
 ### `ralph`
 
-Plan → write a spec under `specs/` → orchestrate from that spec path while maintaining OS-temp implementation notes → simplify → infrastructure discovery → parallel review loop → PR preparation. The workflow result includes the generated spec as `plan_path`, the notes file as `implementation_notes_path`, and the final PR phase report as `pr_report`. The PR phase uses local git identity as a hint when multiple GitHub accounts are logged in, tries available credentials until one works, and posts the implementation notes as a PR comment when a PR is created. Named after the [Ralph Wiggum Method](https://ghuntley.com/ralph/).
+Goal Runner workflow: initialize a persisted goal ledger with a per-run goal id and lifecycle events, render goal-continuation context, run bounded worker LM turns, append receipts, run three independent reviewers, and let a TypeScript reducer decide `complete`, `continue`, `blocked`, or `needs_human`. Token budget behavior is intentionally excluded.
 
 ```text
-/workflow ralph prompt="Migrate the database layer to Drizzle ORM" base_branch=develop
+/workflow ralph objective="Migrate the database layer to Drizzle ORM" max_turns=5 review_quorum=2
 ```
 
-| Input         | Type     | Required | Default | Description                                      |
-| ------------- | -------- | -------- | ------- | ------------------------------------------------ |
-| `prompt`      | `text`   | ✓        | —       | High-level task or goal to accomplish.           |
-| `max_loops`   | `number` | —        | `10`    | Maximum plan → orchestrate → review loops.       |
-| `base_branch` | `string` | —        | `origin/main` | Branch reviewers compare the current delta with. |
+| Input               | Type     | Required | Default       | Description                                      |
+| ------------------- | -------- | -------- | ------------- | ------------------------------------------------ |
+| `objective`         | `text`   | ✓        | —             | Goal-runner objective.                           |
+| `max_turns`         | `number` | —        | `10`          | Maximum worker/review turns.                     |
+| `review_quorum`     | `number` | —        | `2`           | Reviewer `complete` votes required to complete.  |
+| `blocker_threshold` | `number` | —        | `3`           | Consecutive same-blocker turns required to block; requires at least two observations and is capped by `max_turns` when possible. |
+| `base_branch`       | `string` | —        | `origin/main` | Branch reviewers compare the current delta with. |
 
 ### `open-claude-design`