ASRagab · Jul 29, 2026 · Jul 29, 2026 · Jul 29, 2026
diff --git a/.agents/plugins/marketplace.json b/.agents/plugins/marketplace.json
@@ -0,0 +1,20 @@
+{
+  "name": "optimize-anything",
+  "interface": {
+    "displayName": "Optimize Anything"
+  },
+  "plugins": [
+    {
+      "name": "optimize-anything",
+      "source": {
+        "source": "local",
+        "path": "./"
+      },
+      "policy": {
+        "installation": "AVAILABLE",
+        "authentication": "ON_INSTALL"
+      },
+      "category": "Productivity"
+    }
+  ]
+}
diff --git a/.codex-plugin/plugin.json b/.codex-plugin/plugin.json
@@ -0,0 +1,24 @@
+{
+  "name": "optimize-anything",
+  "version": "0.5.0",
+  "description": "Optimize prompts and other text artifacts with measured evaluator feedback",
+  "author": {
+    "name": "optimize-anything contributors"
+  },
+  "repository": "https://github.com/ASRagab/optimize-anything",
+  "license": "MIT",
+  "keywords": ["optimization", "prompts", "evaluator", "gepa"],
+  "skills": "./skills/",
+  "interface": {
+    "displayName": "Optimize Anything",
+    "shortDescription": "Optimize prompts with measured feedback",
+    "longDescription": "Build an evaluation contract, improve prompts with the bundled optimize-anything runtime, and apply only accepted results.",
+    "developerName": "optimize-anything contributors",
+    "category": "Productivity",
+    "capabilities": ["Interactive", "Write"],
+    "defaultPrompt": [
+      "Optimize this prompt and verify the result.",
+      "Improve the prompt in this file safely."
+    ]
+  }
+}
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -8,13 +8,24 @@
 - Centralized project-owned model defaults across the CLI, evaluator generation, judge flows, and operational scripts
 - Preserved proposer precedence as explicit CLI model, then `OPTIMIZE_ANYTHING_MODEL`, then the project default
 
+### Prompt optimization workflow
+- Added the `optimize-prompt` skill for inline prompts, files, embedded prompt regions, and independent batches
+- Separated fast prompt-quality scoring from rigorous task-output evaluation with deterministic hard gates and held-out acceptance
+- Kept repository sources unchanged during search and limited accepted edits to the recorded prompt destination
+
+### Plugin distribution
+- Added a self-locating locked-runtime launcher used by every packaged Claude command
+- Added native Codex plugin and marketplace metadata over the same canonical skill tree
+- Aligned Python, Claude, and Codex release metadata at 0.5.0
+
 ### Provider compatibility
 - Let providers apply their own sampling defaults unless a judge temperature is explicitly supplied
 - Removed hard-coded sampling parameters from generated judge and composite evaluators
 
 ### Documentation and verification
 - Modernized runnable examples, command guidance, protocol docs, skills, and integration tooling for current model identifiers
 - Added drift coverage for model defaults, CLI help and resolution, generated evaluators, and judge request payloads
+- Added offline launcher, evaluator, workflow-fixture, documentation, manifest, and release-version contracts
 - Synchronized package, plugin, and marketplace release versions
 
 ## v0.4.0 - 2026-07-27

diff --git a/README.md b/README.md
@@ -163,9 +163,59 @@ commands, expected report fields, acceptance criteria, and troubleshooting.
 - `analyze`
 - `validate`
 
-## Claude Code Plugin
+## Agent Plugins
 
-optimize-anything is also a Claude Code plugin with guided slash commands and skills.
+The Claude Code plugin and Codex plugin share the same `skills/` tree and
+bundled locked runtime. Plugin users need `uv` and Python 3.10 or newer, but do
+not need a global `optimize-anything` command. The standalone CLI remains a
+separate installation choice.
+
+### Prompt Optimization Workflow
+
+Claude Code users invoke `$optimize-prompt`; Codex users invoke the namespaced
+`$optimize-anything:optimize-prompt`. Both accept prompt text, a standalone
+file, an embedded prompt region, or a list of independent prompt files.
+
+| Evidence mode | Evaluation | What it proves |
+|---|---|---|
+| **Fast mode** | Scores the prompt text for clarity, constraints, and task fitness | Prompt-quality evidence only |
+| **Rigorous mode** | Runs the candidate on representative inputs, then scores task outputs | Task-performance evidence for the tested examples |
+| **Composite mode** | Runs deterministic hard constraints before the rigorous judge | No subjective score can override a failed gate |
+
+Rigorous datasets use JSONL records with `input`, optional `expected`,
+`criteria`, and `hard_constraints`. See
+[`prompt-execution-dataset.md`](skills/optimize-prompt/references/prompt-execution-dataset.md)
+for representative examples, the default system-prompt adapter, custom adapter
+guidance, and expected cost controls. Use explicit proposer, target, and judge
+models; bound calls with a small dataset, budget, and early stopping.
+
+The workflow captures the baseline and writes search output to a temporary or
+run directory. It accepts only a positive comparable score delta with all hard
+constraints satisfied, plus required held-out acceptance in rigorous mode.
+Independent prompt files receive separate decisions. Coupled components require
+an explicit structured-candidate adapter and are not optimized as unrelated
+files.
+
+Inline example:
+
+```text
+$optimize-prompt Improve this prompt in fast mode and return the accepted result:
+"Summarize this."
+```
+
+The response includes the complete accepted prompt, evidence mode, and score
+delta; it does not write a repository file.
+
+Embedded repository example:
+
+```text
+$optimize-prompt Optimize SYSTEM_PROMPT in src/agent.py with the examples in
+evals/prompt.jsonl. Apply it only if held-out acceptance passes.
+```
+
+The workflow optimizes a captured copy, replaces only `SYSTEM_PROMPT`, preserves
+the source representation, and runs the cheapest relevant parse or targeted
+test. Rejected candidates leave the file unchanged.
 
 ### Plugin Regression Workflow
 
@@ -182,25 +232,36 @@ uv run python scripts/check.py --with-plugin
 Requirements:
 - `claude` CLI installed and authenticated
 - `OPENAI_API_KEY` set in the shell that launches the command
-- `ANTHROPIC_API_KEY` set in the shell that launches the command
+- `ANTHROPIC_API_KEY` for `validate` or the full scenario set
 
-The harness runs three real scenarios (`analyze`, `validate`, `quick`), saves Claude JSON outputs plus stderr logs, and fails if Claude does not execute the expected workflow or the optimized artifact is not written.
+The harness runs existing CLI scenarios plus bounded prompt inline-return and
+repository-apply scenarios. `--dry-run` verifies prompt and artifact wiring
+without credentials or model calls.
 
-### Installation
+### Claude Code Plugin
 
 ```bash
-# In Claude Code
-/plugin install ASRagab/optimize-anything
+/plugin marketplace add ASRagab/optimize-anything
+/plugin install optimize-anything@optimize-anything
 ```
 
-Or clone and install locally:
+For a local clone, replace the repository name in the first command with its
+absolute path. Restart Claude Code after installation or update, then invoke
+`$optimize-prompt` or an existing slash command.
+
+### Codex Plugin
 
 ```bash
-git clone https://github.com/ASRagab/optimize-anything.git
-cd optimize-anything
-/plugin install .
+codex plugin marketplace add ASRagab/optimize-anything
+codex plugin add optimize-anything@optimize-anything
 ```
 
+For local development, pass the clone path to `codex plugin marketplace add`.
+Start a new Codex thread after install or update, then invoke
+`$optimize-anything:optimize-prompt` so skill discovery refreshes.
+See [install.md](install.md) for update, removal, verification, and the optional
+standalone CLI path.
+
 ### Slash Commands
 
 | Command | Description |
@@ -217,8 +278,9 @@ cd optimize-anything
 
 ### Skills
 
-The plugin includes three skills that Claude Code can invoke automatically:
+Both plugins include four skills that the host can invoke:
 
+- **optimize-prompt** — Build the rubric, choose fast or rigorous evidence, optimize outside the source, and return or safely apply an accepted prompt
 - **optimization-guide** — Full workflow walkthrough covering modes, configuration, budget, and result interpretation
 - **generate-evaluator** — Choose the right evaluator pattern (judge, command, composite) and generate a script
 - **evaluator-patterns** — Library of ready-to-use evaluator templates for prompts, code, docs, and agent instructions

diff --git a/SKILL.md b/SKILL.md
@@ -27,8 +27,14 @@ Or skip evaluator setup entirely — the guided workflow handles it:
 /optimize-anything:quick my-prompt.txt "make it clearer and more specific"
 ```
 
+For inline prompts, repository prompt regions, or representative-example
+evaluation, invoke `$optimize-prompt` in Claude Code or the namespaced
+`$optimize-anything:optimize-prompt` in Codex. It uses the bundled runtime,
+keeps search output outside the source, and applies only an accepted result.
+
 ## Available Skills
 
+- **optimize-prompt** — Optimize inline prompts, files, embedded regions, or independent batches with fast prompt-quality or rigorous task-output evidence
 - **generate-evaluator** — Choose the right evaluator pattern (judge, command, composite) and generate a script
 - **optimization-guide** — Full workflow walkthrough covering optimization modes, configuration, budget, and result interpretation
 - **evaluator-patterns** — Library of ready-to-use evaluator templates for prompts, code, docs, and agent instructions

diff --git a/commands/analyze.md b/commands/analyze.md
@@ -10,13 +10,13 @@ Use an LLM to discover relevant quality dimensions for a given artifact and opti
 ## Usage
 
 ```bash
-optimize-anything analyze SEED_FILE --judge-model openai/gpt-5.6-luna --objective "Quality"
+"${CLAUDE_PLUGIN_ROOT}/scripts/run-optimize-anything" analyze SEED_FILE --judge-model openai/gpt-5.6-luna --objective "Quality"
 ```
 
 ## Example
 
 ```bash
-optimize-anything analyze my-prompt.txt \
+"${CLAUDE_PLUGIN_ROOT}/scripts/run-optimize-anything" analyze my-prompt.txt \
   --judge-model openai/gpt-5.6-luna \
   --objective "Score for clarity and persuasiveness"
 ```
@@ -25,7 +25,7 @@ optimize-anything analyze my-prompt.txt \
 After dimension discovery, proceed directly to optimization using the returned `intake_json`:
 
 ```bash
-optimize-anything optimize my-prompt.txt \
+"${CLAUDE_PLUGIN_ROOT}/scripts/run-optimize-anything" optimize my-prompt.txt \
   --judge-model openai/gpt-5.6-luna \
   --objective "Score for clarity and persuasiveness" \
   --intake-json '<paste intake_json from analyze>' \

diff --git a/commands/budget.md b/commands/budget.md
@@ -10,13 +10,13 @@ Analyze a seed artifact and recommend an appropriate iteration budget based on i
 ## Usage
 
 ```
-optimize-anything budget SEED_FILE
+"${CLAUDE_PLUGIN_ROOT}/scripts/run-optimize-anything" budget SEED_FILE
 ```
 
 ## Example
 
 ```
-optimize-anything budget my-prompt.txt
+"${CLAUDE_PLUGIN_ROOT}/scripts/run-optimize-anything" budget my-prompt.txt
 ```
 
 See [README](../README.md) for full flag documentation.
diff --git a/commands/compare.md b/commands/compare.md
@@ -9,7 +9,7 @@ Compare two artifacts with the same scoring setup by composing existing `score`
 
 ## Procedure
 1. Score the original artifact:
-   - `optimize-anything score <original> --judge-model <model> --objective "..." [--intake-json ...]`
+   - `"${CLAUDE_PLUGIN_ROOT}/scripts/run-optimize-anything" score <original> --judge-model <model> --objective "..." [--intake-json ...]`
 2. Score the optimized artifact with the exact same evaluator setup.
 3. Present side-by-side:
    - Overall score for each artifact

diff --git a/commands/explain.md b/commands/explain.md
@@ -10,13 +10,13 @@ Display the optimization plan that would be executed for the given seed artifact
 ## Usage
 
 ```
-optimize-anything explain SEED_FILE --evaluator-command bash eval.sh --objective "your goal"
+"${CLAUDE_PLUGIN_ROOT}/scripts/run-optimize-anything" explain SEED_FILE --evaluator-command bash eval.sh --objective "your goal"
 ```
 
 ## Example
 
 ```
-optimize-anything explain prompt.txt \
+"${CLAUDE_PLUGIN_ROOT}/scripts/run-optimize-anything" explain prompt.txt \
   --evaluator-command bash evaluators/clarity.sh \
   --budget 50
 ```

diff --git a/commands/intake.md b/commands/intake.md
@@ -10,13 +10,13 @@ Normalize and validate an intake specification, filling in defaults for quality
 ## Usage
 
 ```
-optimize-anything intake --intake-json '{"artifact_class": "prompt"}'
+"${CLAUDE_PLUGIN_ROOT}/scripts/run-optimize-anything" intake --intake-json '{"artifact_class": "prompt"}'
 ```
 
 ## Example
 
 ```
-optimize-anything intake --intake-file intake.json
+"${CLAUDE_PLUGIN_ROOT}/scripts/run-optimize-anything" intake --intake-file intake.json
 ```
 
 See [README](../README.md) for full flag documentation.
diff --git a/commands/optimize.md b/commands/optimize.md
@@ -4,6 +4,10 @@ description: Guided optimization workflow with mode selection and evaluator setu
 ---
 Run optimization using a deterministic, guided workflow.
 
+For inline prompts, embedded prompt regions, independent prompt batches, or
+task-output evaluation with representative examples, invoke the shared
+`$optimize-prompt` skill instead.
+
 ## Step 1: Identify the artifact
 - If the user provided a file argument, use it directly.
 - Otherwise ask: **"What file should I optimize?"**
@@ -29,7 +33,7 @@ Present these options and ask the user to choose one unless they already specifi
 - If evaluator is already specified, proceed.
 - If no evaluator is specified:
   1. Run `analyze` first:
-     - `optimize-anything analyze <file> --judge-model <model> --objective "<objective>"`
+     - `"${CLAUDE_PLUGIN_ROOT}/scripts/run-optimize-anything" analyze <file> --judge-model <model> --objective "<objective>"`
   2. If analyze fails (API key missing, model unavailable): ask the user for their preferred model, or suggest using `--evaluator-command` with a custom script instead.
   3. Ask: **"Should we use LLM judge directly, or do you want a custom evaluator?"**
   4. If custom evaluator is needed, invoke evaluator generation workflow.

diff --git a/commands/quick.md b/commands/quick.md
@@ -9,7 +9,7 @@ Run a no-questions-asked fast optimization.
 
 ## Behavior (do not ask follow-up questions)
 1. Run analysis to discover dimensions:
-   - `optimize-anything analyze <file> --judge-model openai/gpt-5.6-luna --objective "<objective>"`
+   - `"${CLAUDE_PLUGIN_ROOT}/scripts/run-optimize-anything" analyze <file> --judge-model openai/gpt-5.6-luna --objective "<objective>"`
    - If analyze fails, skip dimension discovery and run optimize with `--judge-model` directly using the objective as-is.
 2. Run optimization using LLM judge with:
    - `--judge-model openai/gpt-5.6-luna`

diff --git a/commands/score.md b/commands/score.md
@@ -10,17 +10,17 @@ Score a single artifact file using a command evaluator, HTTP evaluator, or LLM j
 ## Usage
 
 ```
-optimize-anything score SEED_FILE --evaluator-command bash eval.sh
-optimize-anything score SEED_FILE --judge-model openai/gpt-5.6-luna --objective "Score clarity"
+"${CLAUDE_PLUGIN_ROOT}/scripts/run-optimize-anything" score SEED_FILE --evaluator-command bash eval.sh
+"${CLAUDE_PLUGIN_ROOT}/scripts/run-optimize-anything" score SEED_FILE --judge-model openai/gpt-5.6-luna --objective "Score clarity"
 ```
 
 ## Example
 
 ```
-optimize-anything score my-prompt.txt \
+"${CLAUDE_PLUGIN_ROOT}/scripts/run-optimize-anything" score my-prompt.txt \
   --evaluator-command bash evaluators/clarity.sh
 
-optimize-anything score my-prompt.txt \
+"${CLAUDE_PLUGIN_ROOT}/scripts/run-optimize-anything" score my-prompt.txt \
   --judge-model openai/gpt-5.6-luna \
   --objective "Score for persuasiveness"
 ```

diff --git a/commands/validate.md b/commands/validate.md
@@ -11,7 +11,7 @@ Use multiple LLM judges to verify that a quality improvement is not provider-spe
 
 ## Usage
 ```bash
-optimize-anything validate <file> \
+"${CLAUDE_PLUGIN_ROOT}/scripts/run-optimize-anything" validate <file> \
   --providers openai/gpt-5.6-luna anthropic/claude-sonnet-5 gemini/gemini-3.6-flash \
   --objective "Score for clarity and constraint adherence"
 ```