npm - @ckelsoe/prompt-architect - Versions diffs - 3.2.2 → 3.3.1 - Mend

@ckelsoe/prompt-architect 3.2.2 → 3.3.1

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Files changed (34) hide show

package/.claude-plugin/marketplace.json CHANGED Viewed

@@ -5,14 +5,14 @@
     "email": "charles@kelsoe.com"
   },
   "metadata": {
-    "description": "Prompt engineering skill using 27 research-backed frameworks across 7 intent categories"
+    "description": "Prompt engineering skill using 29 frameworks across 7 intent categories"
   },
   "plugins": [
     {
       "name": "prompt-architect",
       "source": "./",
-      "description": "Analyzes and improves prompts using 27 research-backed frameworks. Intent-based selection routes to the right framework — from ultra-minimal (APE) to agentic (ReAct).",
-      "version": "3.2.1",
+      "description": "Analyzes and improves prompts using 29 frameworks. Intent-based selection routes to the right framework — from ultra-minimal (APE) to agentic (ReAct).",
+      "version": "3.3.1",
       "keywords": ["prompt-engineering", "prompt-frameworks", "agent-skills"],
       "category": "productivity"
     }

package/.claude-plugin/plugin.json CHANGED Viewed

@@ -1,7 +1,7 @@
 {
   "name": "prompt-architect",
-  "description": "27-framework prompt engineering skill for analyzing and improving prompts across 7 intent categories",
-  "version": "3.2.2",
+  "description": "29-framework prompt engineering skill for analyzing and improving prompts across 7 intent categories",
+  "version": "3.3.1",
   "author": {
     "name": "Charles Kelsoe",
     "email": "charles@kelsoe.com"

package/CHANGELOG.md CHANGED Viewed

@@ -5,6 +5,117 @@ All notable changes to the Prompt Architect Claude Code skill will be documented
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
+## [3.3.1] - 2026-07-19
+### Fixed
+- **Publishing was broken before this release and would have failed for any version.** `.github/workflows/publish.yml` pinned Node 20 while running `npm install -g npm@latest`; once npm 12 shipped (requiring `^22.22.2 || ^24.15.0 || >=26`), that step failed with `EBADENGINE` before the publish ran. The workflow now uses Node 24. This is why 3.3.0 has no published artifact — its tag exists, but the release never completed.
+*All 3.3.0 changes below are included in this release.*
+---
+## [3.3.0] - 2026-07-19
+### Added
+- **Chain of Density is now the real method.** `chain-of-density.md` was rewritten around entity densification at *fixed* length, scoped to summarization, and cited to Adams et al., "From Sparse to Dense" (arXiv 2309.04269, NewSum @ EMNLP 2023). The previous content described generic progressive shortening — the opposite of the published technique, which holds length constant.
+- **New framework: Iterative Compression.** The general multi-pass shortening technique that used to be filed under Chain of Density is preserved as its own framework, with its own reference doc and template, and is no longer attributed to CoD. Framework count is now 29.
+- **Quality scoring is implemented.** `SKILL.md` and `adapters/system-prompt.md` now instruct the model to score prompts 1-10 on the five dimensions the README advertises — Clarity, Specificity, Context, Completeness, Structure — with rubric anchors defining the 1-3 / 4-6 / 7-8 / 9-10 bands. Previously this was advertised in three places and implemented in none, and the skill's dimension list disagreed with the README's.
+- **Framework combination guidance.** A "Combining Frameworks" section with concrete pairings (CO-STAR + Self-Refine, RISEN + ReAct, BROKE + Devil's Advocate) that finally routes to `hybrid_template.txt`, previously the only template with no path leading to it.
+- **Self-Consistency citation and a corrected template** in `chain-of-thought.md`, now citing Wang et al. (arXiv 2203.11171, ICLR 2023) and describing a real majority vote over independently sampled reasoning paths rather than asking the model to judge which of three different approaches it liked best.
+- **Origin lines for 10 previously uncited frameworks**: CO-STAR, Chain-of-Thought, RISEN, APE, RTF, CTF, RACE, BAB, plus corrected provenance for RISE and TIDD-EC.
+- **Acronym-collision disambiguation** for APE (vs. "Automatic Prompt Engineer," Zhou et al., arXiv 2211.01910) and RACE (vs. Dave Chaffey's 2010 Reach-Act-Convert-Engage marketing model).
+### Changed
+- **Provenance is now stated honestly per framework.** Frameworks with peer-reviewed backing cite it; community conventions say so plainly rather than leaving a silent gap. CO-STAR is credited to GovTech Singapore's Data Science & AI team — as Sheila Teo's own article does — rather than to Teo, who popularized it.
+- `rise.md`: "RISE-IE" and "RISE-IX" are now labeled as this skill's internal shorthand, not established terminology. RISE-IX is identified as our own composition; its three previously cited sources do not support it. `ctf.md` no longer propagates "RISE-IE" as if it were standard vocabulary.
+- `tidd-ec.md`: dropped the "originally documented alongside CO-STAR" claim, which implied a shared provenance that does not exist, and reassessed its three "authoritative sources" as one origin post, one same-organization republication, and one passing mention.
+- `SKILL.md` step numbering fixed — it had two sections numbered `### 4.` and now runs 1-7, matching the adapter.
+- `SKILL.md` template section reduced from a 30-line filename listing to one line; filenames are mechanically derivable.
+- Installer `--force` now does something: it is required to replace a **symlinked** install path, which previously was deleted without warning.
+### Fixed
+- **Installer no longer installs to Claude Code only.** Mode 5 (the `postinstall` path most users hit) and Mode 3 (`--yes`) now install to every detected agent, as documented. Mode 5 previously hard-skipped every agent but Claude and ignored `detect()` entirely.
+- **Windsurf adapter no longer duplicates itself.** A fresh `.windsurfrules` was written without marker comments, so the next run could not find or remove it and appended a second copy; the marker-stripping regex also lacked the `g` flag. Verified stable across repeated installs.
+- **`--project` is no longer ignored** for the Gemini and universal Agent Skills targets — only the Claude entry honored it.
+- **`engines.node` raised to `>=20.19.0`.** It declared `>=14.0.0`, but `@clack/prompts@1.1.0` is ESM-only, so `require()` of it throws `ERR_REQUIRE_ESM` below 20.19 — the headline `npx` experience failed across most of the declared range with a bare "Installation error". The CI matrix floor was raised to match.
+- **Step-Back statistic corrected**: MuSiQue is +7%, not +25% (a 3.5x overstatement); MMLU is +7% Physics / +11% Chemistry, not a "7-27% range."
+- **Socratic Prompting citation corrected**: single author Edward Y. Chang, IEEE CCWC 2023 — not "Chang et al." and not EMNLP/NAACL.
+- **Pre-Mortem attribution corrected**: the ~30% finding belongs to Mitchell, Russo & Pennington (1989), popularized by Gary Klein (HBR 2007); it measures identification of "reasons for future outcomes" against a might-happen framing, not "failure causes" against forward risk analysis. The unsupported "Brookings Institution" affiliation was removed.
+- **Skeleton-of-Thought quality claim qualified**: parity in roughly 60% of cases, with degradation on writing, math, and coding — previously "maintained or improved quality."
+- **Constitutional AI claim scoped**: the critique-before-revision effect is strongest for smaller models; the authors found no noticeable difference at 52B and kept critiques for transparency.
+- README no longer instructs users to create an `.npmrc` and a GitHub token for a registry the package is not published to. It is on public npm and needs no authentication.
+- README: duplicate `## Quick Start` heading renamed to `## Verifying Your Installation` (the table of contents anchor resolved to the wrong section), and the TOC completed.
+### Removed
+- `framework_analyzer.py` and `prompt_evaluator.py` (~50 KB). Nothing invoked them, they had no CLI, and `framework_analyzer.py` had already drifted — its header read "all 27" above a 28-entry dict. Their scoring dimensions were carried into the SKILL.md rubric before removal.
+- `.github/workflows/version-bump.yml`. It ran `npm version`, which updates only `package.json`, then immediately pushed the tag — landing a broken release tag on the remote before anyone could intervene.
+### Validation
+- `validate-skill.js` now derives framework and template checks from the filesystem instead of a hardcoded 7-entry subset, cross-checks `claudeCode.frameworks` against what is on disk, and **asserts every advertised framework count agrees with reality**. This is the check whose absence let "27" ship while 28 frameworks existed.
+- `validate-skill.js` now verifies **all five version sites** — `package.json`, `claudeCode.version`, `plugin.json`, `marketplace.json`, and `SKILL.md` frontmatter. It previously checked only `plugin.json`, and did not check SKILL.md's version at all, despite documentation claiming otherwise.
+- `test.js` gained a **drift guard** asserting `SKILL.md` and `adapters/system-prompt.md` keep identical step headings, sequential numbering, and the same framework set. These two files are hand-maintained copies and had already diverged.
+---
+## [3.2.2] - 2026-03-30
+### Added
+- Prominent Quick Start at the top of the README with the `npx @ckelsoe/prompt-architect` command
+- Codex CLI installation instructions via `$skill-installer install`
+- `/install-skill` documented as the primary Claude Code installation method, with the plugin marketplace commands retained for update support
+### Changed
+- README installation section consolidated: separate per-tool sections for Gemini CLI and for Cursor/Copilot/Windsurf/Codex were replaced by a single "Other Agents" section pointing at `~/.agents/skills/`
+- npm demoted to an alternative installation method
+### Fixed
+- `.claude-plugin/plugin.json` and `.claude-plugin/marketplace.json` versions brought back in sync with `package.json`; both had been left behind at 3.2.1
+---
+## [3.2.1] - 2026-03-24
+### Fixed
+- **Direct download link for ChatGPT**: the `.skill` link now points at `releases/latest/download/prompt-architect.skill` rather than the releases page, making it a one-click download
+- **Nothing after the revised prompt**: SKILL.md now states explicitly that no text may follow the closing backticks — the revised prompt must be the last element of the response. The example interaction was updated to comply.
+### Changed
+- `prompt-architect.skill` removed from git tracking and added to a new `.gitignore` alongside `node_modules/`. It is a build artifact produced by `npm run build:skill` and attached to releases by CI.
+---
+## [3.2.0] - 2026-03-24
+### Changed
+- **SKILL.md "Present Improvements" restructured for copy-pasteable output.** Responses now follow a fixed order: (A) analysis — framework selected, changes made, components applied; (B) a usage-instructions block explaining how to use the prompt in a new chat or the same chat; (C) the revised prompt last, in a fenced code block.
+- **The revised prompt is now clean flat text**: no framework section headers (`BEFORE:`, `BRIDGE:`, `CONTEXT:`), no gratuitous indentation, and no internal markdown unless the prompt genuinely requires it — so it can be copied verbatim with zero editing.
+- Example interaction rewritten to model the new output structure
+- Version synced across SKILL.md, `package.json`, `plugin.json`, and `marketplace.json`
+---
+## [3.1.1] - 2026-03-24
+### Fixed
+- **CI pipeline**: committed `package-lock.json` so `npm ci` works, and removed the unreliable `npm ci || npm install` fallback
+- Test matrix moved to Node 18/20/22; Node 14/16 dropped, as `@clack/prompts` requires a newer runtime
+---
+## [3.1.0] - 2026-03-24
+> The 3.x restructure described in the `[3.0.0]` entry below shipped across the 3.0.x line and landed in git under the `v3.1.0` tag. The items here are the parts of that work not already listed there.
+### Added
+- **`.skill` builder**: `scripts/build-skill.js` and `npm run build:skill` produce a ZIP for ChatGPT upload
+- **`bin` entry**: `npx @ckelsoe/prompt-architect` runs the installer directly
+### Changed
+- `package.json` `files` array now ships `skills/`, `.claude-plugin/`, and `MIGRATION.md`
+- Removed the obsolete adapters `for-cursor.mdc`, `for-github-copilot.md`, and `for-openai-codex-cli.md` — those tools read `SKILL.md` natively. Only `system-prompt.md` (universal) and `for-windsurf.md` remain.
+---
 ## [3.0.0] - 2026-03-24
 ### Breaking Changes
@@ -14,7 +125,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ### Added
 - **Claude Code Plugin System support**: `.claude-plugin/plugin.json` and `marketplace.json` — install via `/plugin marketplace add` and `/plugin install`
 - **Gemini CLI support**: Native Agent Skills compatibility. Install via `gemini skills install`
-- **Agent Skills standard compliance** (agentskills.io): Added `license`, `compatibility`, and `metadata` fields to SKILL.md frontmatter. Works with 30+ compatible agents including Cursor, Copilot, Kiro, Roo Code, Amp, OpenHands, and more
+- **Agent Skills standard compliance** (agentskills.io): Added `license`, `compatibility`, and `metadata` fields to SKILL.md frontmatter. Works with 30+ compatible agents including Cursor, Copilot, Kiro, Zoo Code, Amp, OpenHands, and more
 - **Interactive multi-agent installer**: Detects installed AI agents (Claude Code, Gemini CLI, Cursor, Copilot, Windsurf, Codex) and presents a selection UI using @clack/prompts
 - **Universal install path**: `~/.agents/skills/` for Agent Skills standard compatible tools
 - **CLI flags**: `--all`, `--claude`, `--gemini`, `--agents`, `--cursor`, `--copilot`, `--windsurf`, `--codex`, `--yes`
@@ -258,7 +369,7 @@ The framework selection system in SKILL.md was completely restructured:
 ## Links
-- [npm Package](https://www.npmjs.com/package/@ckelsoe/claude-skill-prompt-architect)
-- [GitHub Repository](https://github.com/ckelsoe/claude-skill-prompt-architect)
-- [Issue Tracker](https://github.com/ckelsoe/claude-skill-prompt-architect/issues)
-- [Contributing Guidelines](https://github.com/ckelsoe/claude-skill-prompt-architect/blob/main/README.md#contributing)
+- [npm Package](https://www.npmjs.com/package/@ckelsoe/prompt-architect)
+- [GitHub Repository](https://github.com/ckelsoe/prompt-architect)
+- [Issue Tracker](https://github.com/ckelsoe/prompt-architect/issues)
+- [Contributing Guidelines](https://github.com/ckelsoe/prompt-architect/blob/main/README.md#contributing)

package/MIGRATION.md CHANGED Viewed

@@ -54,7 +54,7 @@ skills/prompt-architect/SKILL.md
 ### The skill itself is unchanged
-The `SKILL.md` content, all 27 frameworks, reference docs, and templates are identical. Only the packaging and distribution have changed.
+The `SKILL.md` content, all framework reference docs, and all templates are identical. Only the packaging and distribution have changed.
 ## GitHub Repository

package/README.md CHANGED Viewed

@@ -1,10 +1,10 @@
 # Prompt Architect
-Transform vague prompts into expert-level, structured prompts using 27 research-backed frameworks across 7 intent categories.
+Transform vague prompts into expert-level, structured prompts using 29 frameworks across 7 intent categories.
 Works with **Claude Code, ChatGPT, Gemini CLI, Cursor, GitHub Copilot, Windsurf, OpenAI Codex**, and [30+ Agent Skills compatible tools](https://agentskills.io).
-[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT) [![npm](https://img.shields.io/npm/v/@ckelsoe/prompt-architect)](https://www.npmjs.com/package/@ckelsoe/prompt-architect) [![Agent Skills](https://img.shields.io/badge/Agent%20Skills-compatible-blue)](https://agentskills.io)
+[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT) [![npm](https://img.shields.io/npm/v/@ckelsoe/prompt-architect)](https://www.npmjs.com/package/@ckelsoe/prompt-architect) [![Agent Skills](https://img.shields.io/badge/Agent%20Skills-compatible-blue)](https://agentskills.io) [![Available on CodeGuilds](https://img.shields.io/badge/Available_on-CodeGuilds-6366f1)](https://codeguilds.dev/packages/prompt-architect)
 ## Quick Start
@@ -14,9 +14,9 @@ npx @ckelsoe/prompt-architect
 The interactive installer detects your AI agents (Claude Code, Gemini CLI, Cursor, Copilot, Codex, and more) and lets you choose where to install.
-> **Important:** Use `npx`, not `npm install`. The `npx` command runs the interactive multi-agent installer. Running `npm install` will only install to Claude Code silently via the postinstall hook.
+> **Tip:** Use `npx` rather than `npm install`. Both work — `npm install` runs the same installer via the postinstall hook and installs to every agent it detects — but only `npx` gives you the interactive menu to choose specific targets.
-> Requires `.npmrc` with `@ckelsoe:registry=https://npm.pkg.github.com` and a GitHub token with `read:packages` scope.
+> Published on the public npm registry. **No authentication required.**
 ---
@@ -28,11 +28,14 @@ The interactive installer detects your AI agents (Claude Code, Gemini CLI, Curso
 - [Supported Frameworks](#supported-frameworks)
 - [Quick Start](#quick-start)
 - [Installation](#installation)
+- [Verifying Your Installation](#verifying-your-installation)
+- [Updating](#updating)
 - [Usage](#usage)
 - [Framework Selection Guide](#framework-selection-guide)
 - [Project Structure](#project-structure)
 - [Documentation](#documentation)
 - [Contributing](#contributing)
+- [Support](#support)
 - [License](#license)
 ---
@@ -57,7 +60,7 @@ Prompt Architect is an [Agent Skills](https://agentskills.io) compatible skill t
 ## Key Features
-### 27 Research-Backed Frameworks Across 7 Intent Categories
+### 29 Frameworks Across 7 Intent Categories
 | Framework | Best For | Complexity |
 |-----------|----------|------------|
@@ -81,7 +84,8 @@ Prompt Architect is an [Agent Skills](https://agentskills.io) compatible skill t
 | **Least-to-Most** | Compositional multi-hop problems (simplest first) | Medium |
 | **Plan-and-Solve (PS+)** | Zero-shot numerical/calculation reasoning | Low |
 | **Chain of Thought** | Reasoning, problem-solving | Medium |
-| **Chain of Density** | Iterative refinement, summarization | Medium |
+| **Chain of Density** | Summarization at fixed length, maximum information density | Medium |
+| **Iterative Compression** | Shortening text toward a target length | Low |
 | **Self-Refine** | Iterative output quality improvement (any task) | Medium |
 | **CAI Critique-Revise** | Principle-based critique and revision (Anthropic) | Medium |
 | **Devil's Advocate** | Strongest opposing argument against a position | Low |
@@ -457,7 +461,7 @@ RESPONSE FORMAT:
 **Best for:** Identifying specific failure causes before they happen
 **Approach:** Assume the project has already failed → describe the failure → work backwards to specific causes with warning signs
-**Research:** Gary Klein's prospective hindsight — ~30% improvement over forward risk analysis
+**Research:** Mitchell, Russo & Pennington (1989), popularized by Gary Klein (HBR 2007) — prospective hindsight improves identification of reasons for future outcomes by ~30%; LLM application is practitioner-level, with no dedicated AI paper
 **Example Use Cases:** Project kickoffs, product launches, technical migrations, high-stakes strategic decisions
@@ -558,23 +562,33 @@ Choose the method that matches your AI tool:
 ### Claude Code
-**Option A: Plugin system (recommended)**
+```
+/install-skill https://github.com/ckelsoe/prompt-architect/tree/main/skills/prompt-architect
+```
+**Plugin system (enables updates):**
 ```
 /plugin marketplace add ckelsoe/prompt-architect
 /plugin install prompt-architect@prompt-architect-marketplace
 ```
-**Option B: npm**
-```bash
-npm install -g @ckelsoe/prompt-architect
+### Codex CLI
+```
+$skill-installer install https://github.com/ckelsoe/prompt-architect/tree/main/skills/prompt-architect
 ```
-The installer runs automatically and copies the skill to `~/.claude/skills/`. Restart Claude Code to load it.
-**Option C: Interactive multi-agent installer**
+### Other Agents (Gemini, Cursor, Copilot, Cline, Zoo Code, etc.)
+Copy `skills/prompt-architect/` from this repo to `~/.agents/skills/prompt-architect/`
+### npm (alternative)
 ```bash
 npx @ckelsoe/prompt-architect
 ```
-Detects all installed AI agents and lets you choose where to install.
+Installs to both `.claude/skills/` and `.agents/skills/` in one step.
 ### ChatGPT
@@ -585,30 +599,6 @@ ChatGPT natively supports [Agent Skills](https://agentskills.io). Available on B
 3. Click **New skill** → **Upload from your computer**
 4. Upload the `.skill` file
-### Gemini CLI
-```bash
-gemini skills install https://github.com/ckelsoe/prompt-architect.git
-```
-Or manually: copy `skills/prompt-architect/` to `~/.gemini/skills/prompt-architect/`
-### Cursor, Copilot, Windsurf, Codex CLI
-Use the interactive installer to auto-detect and install:
-```bash
-npx @ckelsoe/prompt-architect
-```
-Or copy `skills/prompt-architect/` to your tool's skills directory. All of these tools support the [Agent Skills standard](https://agentskills.io) natively:
-| Tool | Skills directory |
-|------|----------------|
-| **Cursor** | `.agents/skills/` or `.cursor/skills/` (project) |
-| **GitHub Copilot** | `.github/skills/` or `~/.copilot/skills/` |
-| **OpenAI Codex** | `.agents/skills/` (project) or `~/.agents/skills/` |
-| **VS Code, Roo Code, Kiro, Amp, Junie, Goose, 30+ others** | `~/.agents/skills/` (universal) |
 ### Windsurf
 Windsurf does not yet support Agent Skills natively:
@@ -624,7 +614,7 @@ See [`adapters/README.md`](adapters/README.md) for detailed instructions for eac
 ---
-## Quick Start
+## Verifying Your Installation
 After installing, test with:
@@ -851,11 +841,7 @@ prompt-architect/
 ├── LICENSE                            # MIT License
 │
 └── prompt-architect/                  # The skill
-    ├── SKILL.md                       # Core skill instructions (5 KB)
-    │
-    ├── scripts/                       # Analysis Utilities
-    │   ├── framework_analyzer.py      # Framework recommendation logic
-    │   └── prompt_evaluator.py        # Quality scoring system
+    ├── SKILL.md                       # Core skill instructions
     │
     ├── references/                    # Framework Documentation
     │   └── frameworks/                # Loaded on-demand
@@ -922,9 +908,8 @@ prompt-architect/
 **Core Components:**
 - **SKILL.md** - Main skill logic and instructions (intent-based selection system)
-- **27 Framework Docs** - Complete references with examples
-- **29 Templates** - Ready-to-use structures (27 frameworks + hybrid + chain variants)
-- **2 Python Scripts** - Analysis and scoring utilities
+- **Framework Reference Docs** - 28 files covering all 29 frameworks, with examples (RISE documents two variants)
+- **Templates** - 30 ready-to-use structures: one per framework, plus `hybrid_template.txt` for combinations
 ---

package/adapters/README.md CHANGED Viewed

@@ -19,7 +19,7 @@ These tools read `SKILL.md` natively. No adapter needed — just install to thei
 | **GitHub Copilot** | Copy `skills/prompt-architect/` to `.github/skills/` or `~/.copilot/skills/` |
 | **OpenAI Codex** | Copy `skills/prompt-architect/` to `.agents/skills/` in your project |
 | **VS Code** | Copy `skills/prompt-architect/` to `.agents/skills/` in your project |
-| **Roo Code** | Copy `skills/prompt-architect/` to `.agents/skills/` or `~/.agents/skills/` |
+| **Zoo Code** | Copy `skills/prompt-architect/` to `.agents/skills/` or `~/.agents/skills/` |
 | **Kiro, Amp, Junie, Goose, OpenHands, etc.** | Copy to `~/.agents/skills/` (universal path) |
 Or use the interactive installer to auto-detect and install:

package/adapters/system-prompt.md CHANGED Viewed

@@ -13,16 +13,30 @@ You are an expert in prompt engineering and systematic application of prompting
 ### 1. Initial Assessment
-When a user provides a prompt to improve, analyze across dimensions:
-- **Clarity**: Is the goal clear and unambiguous?
-- **Specificity**: Are requirements detailed enough?
-- **Context**: Is necessary background provided?
-- **Constraints**: Are limitations specified?
-- **Output Format**: Is desired format clear?
+When a user provides a prompt to improve, **score it 1-10 on each of these five dimensions** and report an overall score (the mean, to one decimal place). Always show the scores — they justify the changes you are about to make and give the user a before/after they can feel.
+| Dimension | What you are scoring |
+|---|---|
+| **Clarity** | Is the goal unambiguous? Penalize vague terms ("thing", "stuff", "something", "maybe"), unresolved pronouns, and an implied-but-unstated objective. |
+| **Specificity** | Are requirements concrete? Reward named entities, quantities, and explicit format/length/style specifications. Penalize prompts so short they cannot carry the detail. |
+| **Context** | Is the necessary background present? Reward stated situation, audience, and rationale ("because", "in order to"). Penalize a bare instruction with no setting. |
+| **Completeness** | Are *what*, *why*, *how*, and *output format* all present? Each missing element costs. |
+| **Structure** | Is it organized for its length? Reward sections, lists, and logical ordering. Penalize run-on sentences and long unbroken prose. |
+**Rubric anchors** — apply per dimension so scores mean the same thing every time:
+| Band | Meaning |
+|---|---|
+| **1-3** | Absent or actively harmful. The model would have to guess this dimension entirely. |
+| **4-6** | Present but underspecified. The model can proceed, but will fill gaps with assumptions the user did not choose. |
+| **7-8** | Solid. Enough to produce a good result; refinement would be marginal. |
+| **9-10** | Complete and unambiguous. A competent model has nothing left to infer on this dimension. |
+Score the prompt *as written*, not as you charitably interpret it — the gap between those two is precisely what the framework will fix. A prompt scoring 7+ across the board often needs no framework at all (see **When NOT to Use Frameworks**).
 ### 2. Intent-Based Framework Selection
-With 27 frameworks, identify the user's **primary intent** first, then use the discriminating questions within that category.
+With 29 frameworks, identify the user's **primary intent** first, then use the discriminating questions within that category.
 ---
@@ -64,7 +78,8 @@ With 27 frameworks, identify the user's **primary intent** first, then use the d
 |--------|-----------|
 | Rewrite, refactor, convert | **BAB** |
 | Iterative quality improvement | **Self-Refine** |
-| Compress or densify | **Chain of Density** |
+| Summarize at fixed length, maximize information | **Chain of Density** |
+| Shorten text toward a target length | **Iterative Compression** |
 | Outline-first then expand sections | **Skeleton of Thought** |
 ---
@@ -102,6 +117,20 @@ With 27 frameworks, identify the user's **primary intent** first, then use the d
 ---
+#### Combining Frameworks
+Most prompts need exactly one framework. Combine only when the task genuinely has **two separable phases** — one framework structures the request, a second governs how the output is checked or refined. If you cannot name the two phases, do not combine.
+| When | Combination | Why |
+|---|---|---|
+| High-stakes content that must survive review | **CO-STAR + Self-Refine** | CO-STAR fixes audience/tone/format; Self-Refine adds a critique-and-revise loop before delivery |
+| Multi-step procedure executed with tools | **RISEN + ReAct** | RISEN specifies the steps and success criteria; ReAct governs the tool-use cycle within each step |
+| Business deliverable with a hostile audience | **BROKE + Devil's Advocate** | BROKE sets objective and key results; Devil's Advocate stress-tests them before they reach a stakeholder |
+When you combine, state plainly in your analysis which framework owns which phase. Never stack more than two — a third adds structure the model spends attention parsing rather than following.
+---
 ### 3. Framework Quick Reference
 **Simple:** APE | RTF | CTF
@@ -109,7 +138,7 @@ With 27 frameworks, identify the user's **primary intent** first, then use the d
 **Comprehensive:** CO-STAR | RISEN | TIDD-EC
 **Data:** RISE-IE | RISE-IX
 **Reasoning:** Plan-and-Solve | Chain of Thought | Least-to-Most | Step-Back | Tree of Thought | RCoT
-**Structure/Iteration:** Skeleton of Thought | Chain of Density
+**Structure/Iteration:** Skeleton of Thought | Chain of Density | Iterative Compression
 **Critique/Quality:** Self-Refine | CAI Critique-Revise | Devil's Advocate | Pre-Mortem
 **Meta/Reverse:** RPEF | Reverse Role Prompting
 **Agentic:** ReAct
@@ -138,7 +167,8 @@ Ask targeted questions (3-5 at a time) based on identified gaps:
 **For Least-to-Most**: Full problem, decomposed subproblems in dependency order?
 **For Plan-and-Solve**: Problem with all relevant numbers/variables?
 **For Chain of Thought**: Problem, reasoning steps, verification?
-**For Chain of Density**: Content to improve, iterations, optimization goals?
+**For Chain of Density**: Source document to summarize, fixed target length, number of iterations?
+**For Iterative Compression**: Content to compress, target length, optimization goal, stopping criterion?
 **For Self-Refine**: Output to improve, feedback dimensions, stop condition?
 **For CAI Critique-Revise**: The principle to enforce, output to critique?
 **For Devil's Advocate**: Position to attack, attack dimensions, severity ranking needed?
@@ -167,7 +197,7 @@ Show improved prompt with:
 - Confirm improvements align with intent
 - Refine based on feedback
-- Switch or combine frameworks if needed
+- Switch or combine frameworks if needed (see **Combining Frameworks** above)
 - Continue until satisfactory
 ## Key Principles

package/package.json CHANGED Viewed

@@ -1,7 +1,7 @@
 {
   "name": "@ckelsoe/prompt-architect",
-  "version": "3.2.2",
-  "description": "Agent skill for analyzing and improving prompts using 27 research-backed frameworks across 7 intent categories. Works with Claude Code, Gemini CLI, Cursor, Copilot, and 30+ Agent Skills compatible tools.",
+  "version": "3.3.1",
+  "description": "Agent skill for analyzing and improving prompts using 29 frameworks across 7 intent categories. Works with Claude Code, Gemini CLI, Cursor, Copilot, and 30+ Agent Skills compatible tools.",
   "main": "skills/prompt-architect/SKILL.md",
   "keywords": [
     "claude-code",
@@ -56,7 +56,7 @@
     "prompt-architect": "scripts/install.js"
   },
   "engines": {
-    "node": ">=14.0.0"
+    "node": ">=20.19.0"
   },
   "files": [
     "skills/",
@@ -81,7 +81,7 @@
     "type": "skill",
     "skillName": "prompt-architect",
     "skillPath": "skills/prompt-architect",
-    "version": "3.2.2",
+    "version": "3.3.1",
     "compatibility": {
       "claudeCode": ">=1.0.0"
     },
@@ -116,6 +116,7 @@
       "ReAct",
       "Chain-of-Thought",
       "Chain-of-Density",
+      "Iterative-Compression",
       "Self-Refine",
       "CAI-Critique-Revise",
       "Devils-Advocate",