De-bundle the Claude Code plugin: skills ship from the agent-skills marketplace (LCR-181)

CLAUDE.md

@@ -2,10 +2,10 @@
 
 ## Project Overview
 
-**lightcone-cli** is Lightcone Research's agentic layer for ASTRA (Agentic Schema for Transparent Research Analysis). It ships the `lc` executable and Claude Code skills/hooks used during interactive analysis work.
+**lightcone-cli** is Lightcone Research's agentic layer for ASTRA (Agentic Schema for Transparent Research Analysis). It ships the `lc` executable. The agent skills and hooks used during interactive analysis work ship separately, from the [agent-skills](https://github.com/LightconeResearch/agent-skills) marketplace as the `lightcone` plugin. `lc init` registers the marketplace globally with each harness on PATH, then activates the plugin per project. For Claude Code it runs `claude plugin marketplace add` (global) and writes `enabledPlugins` into the project's `.claude/settings.json` (activation only). For Codex it runs `codex plugin` (global) when `codex` is on PATH.
 
 - **ASTRA** = pure specification: schema, validation, prior insights & findings, evidence verification, helpers, minimal CLI
-- **lightcone-cli** = agentic layer: Claude Code skills, project scaffolding, **Snakemake-based execution**, container builds
+- **lightcone-cli** = agentic layer: agent skills, project scaffolding, **Snakemake-based execution**, container builds
 
 lightcone-cli depends on ASTRA. The `astra` CLI handles spec operations; the `lc` CLI handles execution and agent operations.
 
@@ -49,9 +49,7 @@ astra.yaml ── snakefile generator ──> .lightcone/Snakefile
 src/lightcone/              # namespace — NO __init__.py
 ├── cli/                    # Click surface
 │   ├── __init__.py         # exposes main()
-│   ├── commands.py         # init, run, status, verify, build
-│   ├── plugin.py           # get_plugin_source_dir
-│   └── claude/             # force-included Claude plugin bundle (in installed wheel only)
+│   └── commands.py         # init, run, status, verify, build
 ├── engine/                 # execution substrate — Snakemake-based
 │   ├── __init__.py
 │   ├── manifest.py         # write_manifest, sha256_dir, code_version — the integrity layer
@@ -67,17 +65,12 @@ src/lightcone/              # namespace — NO __init__.py
     ├── cli.py              # `lc eval` subcommand group
     ├── harness.py, sandbox.py, graders.py, build.py, report.py, models.py
 
-claude/lightcone/           # Claude plugin source — force-included into the wheel
-├── skills/                 # lc-new, lc-from-code, lc-from-paper,
-│                            # lc-feedback, ralph;
-│                            # paper-reproduction bundle: lc-from-paper (entry),
-│                            # ralph (loop substrate),
-│                            # paper-extraction, figure-comparison,
-│                            # check-sentence-by-sentence
-│                            # (see skills/README.md for the full bundle map)
-├── agents/                 # lc-extractor
-├── templates/              # Project CLAUDE.md template
-└── scripts/                # Session hooks (bash): venv activation, validate-on-save, session-start primer
+# Skills, hooks, and the lc-extractor subagent are NOT in this repo. They ship
+# from github.com/LightconeResearch/agent-skills as the `lightcone` plugin.
+# `lc init` registers the marketplace globally with each harness, then activates
+# the plugin per project: a global `claude plugin marketplace add` plus
+# enabledPlugins in .claude/settings.json for Claude Code, and `codex plugin`
+# (global) for Codex when `codex` is on PATH.
 
 tests/                      # pytest — mirrors src/ structure
 pyproject.toml              # hatchling + hatch-vcs, ASTRA + Snakemake as deps
@@ -153,7 +146,7 @@ Global config (`~/.lightcone/config.yaml`) is auto-created with defaults on firs
 | Change manifest semantics | `src/lightcone/engine/manifest.py` + `tests/test_manifest.py` | Bump `SCHEMA_VERSION`; add a test |
 | Change Snakefile shape | `src/lightcone/engine/snakefile.py` + `tests/test_snakefile.py` | Includes a `snakemake -n` parse test |
 | Add container features | `src/lightcone/engine/container.py` | `compute_image_tag()`, build/resolve functions |
-| Create a skill | `claude/lightcone/skills/` | SKILL.md with YAML frontmatter (`name`, `description`, `allowed-tools`) |
+| Create or edit a skill | [`LightconeResearch/agent-skills`](https://github.com/LightconeResearch/agent-skills) | Skills live in that repo, not here |
 
 ## Test Patterns
 

README.md

@@ -18,23 +18,37 @@ care of the rest — specification, execution, and provenance.
 uv tool install lightcone-cli
 lc init my-analysis
 cd my-analysis
-claude
+# then open your agent harness in the project, e.g. Claude Code:
+claude   # trust the folder to install the lightcone plugin
 ```
 
+The skills ship as a plugin, not with `lightcone-cli`. `lc init` registers the
+[agent-skills](https://github.com/LightconeResearch/agent-skills) marketplace
+globally with each agent harness on PATH, once. For Claude Code it runs
+`claude plugin marketplace add`, then writes `enabledPlugins` into the project's
+`.claude/settings.json` to activate the `lightcone` plugin on folder trust. For
+Codex it runs `codex plugin` when `codex` is on PATH. The
+[install guide](https://docs.lightconeresearch.org/user/install/) has the
+by-hand commands as a fallback.
+
 Then tell the agent what you have to start from — a research question
-(`/lc-new`), existing code (`/lc-from-code`), or a paper to reproduce
-(`/lc-from-paper`).
+(`/lightcone:new`), existing code (`/lightcone-experimental:from-code`), or a
+paper to reproduce (`/lightcone-experimental:from-paper`).
 
 → [Full getting-started guide](https://docs.lightconeresearch.org/user/getting-started/)
 
 ## Skills
 
+Skills live in the
+[agent-skills](https://github.com/LightconeResearch/agent-skills) marketplace.
+
 | Skill | What it does |
 |---|---|
-| [`/lc-new`](https://docs.lightconeresearch.org/skills/lc-new/) | Scope a new analysis from a research question into a full `astra.yaml` spec |
-| [`/lc-from-code`](https://docs.lightconeresearch.org/skills/lc-from-code/) | Bring an existing codebase into ASTRA |
-| [`/lc-from-paper`](https://docs.lightconeresearch.org/skills/lc-from-paper/) | Reproduce a published paper end-to-end |
-| [`/lc-feedback`](https://docs.lightconeresearch.org/skills/lc-feedback/) | File a bug report with version and error context auto-collected |
+| `/lightcone:new` | Scope a new analysis from a research question into a full `astra.yaml` spec |
+| `/lightcone:report` | Author the MyST report that references `astra.yaml` elements by path |
+| `/lightcone:feedback` | File a bug report with version and error context auto-collected |
+| `/lightcone-experimental:from-code` | Bring an existing codebase into ASTRA |
+| `/lightcone-experimental:from-paper` | Reproduce a published paper end-to-end |
 
 ## Capabilities