update SKILLS

Author: GiggleLiu

.claude/CLAUDE.md

@@ -4,7 +4,7 @@
 Rust library for NP-hard problem reductions. Implements computational problems with reduction rules for transforming between equivalent formulations.
 
 ## Skills
-- [issue-to-pr](skills/issue-to-pr/SKILL.md) -- Convert a GitHub issue into a PR with an implementation plan. Validates the issue against the appropriate checklist, then dispatches to `add-model` or `add-rule`.
+- [issue-to-pr](skills/issue-to-pr/SKILL.md) -- Convert a GitHub issue into a PR with an implementation plan. One item per PR: `[Rule]` issues require both models to exist on `main`; never bundle model + rule in the same PR.
 - [add-model](skills/add-model/SKILL.md) -- Add a new problem model. Can be used standalone (brainstorms with user) or called from `issue-to-pr`.
 - [add-rule](skills/add-rule/SKILL.md) -- Add a new reduction rule. Can be used standalone (brainstorms with user) or called from `issue-to-pr`.
 - [review-implementation](skills/review-implementation/SKILL.md) -- Review implementation completeness by dispatching parallel subagents (structural + quality) with fresh context. Auto-detects new models/rules from git diff. Called automatically at the end of `add-model`/`add-rule`, after each `executing-plans` batch, or standalone via `/review-implementation`.

.claude/skills/issue-to-pr/SKILL.md

@@ -54,16 +54,19 @@ LABELS=$(gh issue view <number> --json labels --jq '[.labels[].name] | join(",")
 - If `Good` is NOT in the labels → **STOP**: "Issue #N has not passed check-issue. Please run `/check-issue <N>` first."
 - If `Good` is present → continue to step 4.
 
-### 3.5. Check for Related Open PRs
+### 3.5. Model-Existence Guard (for `[Rule]` issues only)
 
-When creating a `[Rule]` PR, check if there's already an open `[Model]` PR for the source or target problem:
+For `[Rule]` issues, parse the source and target problem names from the title (e.g., `[Rule] BinPacking to ILP` → source=BinPacking, target=ILP). Verify that **both** models already exist in the codebase on `main`:
 
 ```bash
-gh pr list --state open --search "[Model]" --json title,number,headRefName
+grep -r "struct SourceName" src/models/
+grep -r "struct TargetName" src/models/
 ```
 
-- If an open Model PR exists for the source/target problem, the Rule PR should **base its branch on that Model branch** (not `main`), to avoid duplicating CLI registration, export regeneration, etc.
-- If both a `[Model]` and `[Rule]` issue exist for the same problem, prefer implementing them in the **same PR** to avoid redundant work. Base the branch on `main`, implement the model first, then the rule.
+- If **both** models exist → continue to step 4.
+- If either model is missing → **STOP**. Comment on the issue: "Blocked: model `<name>` does not exist in main yet. Please implement it first (or file a `[Model]` issue)."
+
+**One item per PR:** Do NOT implement a missing model as part of a `[Rule]` PR. Each PR should contain exactly one model or one rule, never both. This avoids bloated PRs and repeated implementation when the model is needed by multiple rules.
 
 ### 4. Research References
 
@@ -254,5 +257,5 @@ Run /review-pipeline to process Copilot comments, fix CI, and run agentic tests.
 | Not verifying facts from issue | Use WebSearch/WebFetch to cross-check claims |
 | Branch already exists on retry | Check with `git rev-parse --verify` before `git checkout -b` |
 | Dirty working tree | Verify `git status --porcelain` is empty before branching |
-| Redundant Model+Rule PRs | Check for related open PRs (Step 3.5); combine or base on existing branch |
+| Bundling model + rule in one PR | Each PR must contain exactly one model or one rule — STOP and block if model is missing (Step 3.5) |
 | Plan files left in PR | Delete plan files before final push (Step 7c) |

.claude/skills/project-pipeline/SKILL.md

@@ -45,16 +45,17 @@ Filter items where `status == "Ready"`. Partition into `[Model]` and `[Rule]` bu
 
 #### 0b. Gather Context for Ranking
 
-1. **Existing problems:** Call `list_problems` (MCP tool) to get all problems currently in the reduction graph.
+1. **Existing problems:** Grep for problem struct definitions in the codebase: `grep -r "^pub struct" src/models/ | sed 's/.*pub struct \([A-Za-z]*\).*/\1/'` to get all problem names currently implemented on `main`.
 2. **Pending rules:** From the full project board JSON, collect all `[Rule]` issues that are in "Ready" or "In Progress" status. Parse their source/target problem names (e.g., `[Rule] BinPacking to ILP` → source=BinPacking, target=ILP).
 
 #### 0c. Check Eligibility
 
-**Rule issues require both source and target models to exist.** For each `[Rule]` issue, parse the source and target problem names (e.g., `[Rule] BinPacking to ILP` → source=BinPacking, target=ILP). Check that both appear in the `list_problems` output (existing models) OR in a `[Model]` issue in the current Ready/In Progress columns.
+**Rule issues require both source and target models to exist on `main`.** For each `[Rule]` issue, parse the source and target problem names (e.g., `[Rule] BinPacking to ILP` → source=BinPacking, target=ILP). Check that both appear in the existing problems list (from Step 0b grep).
 
-- If both models exist → **eligible**
-- If a missing model has a `[Model]` issue in Ready → eligible only in `--all` mode (the Model will be processed first); in single-issue mode, **skip this Rule** and mark it `[blocked]`
-- If a missing model has no `[Model]` issue at all → **ineligible**, mark it `[blocked]` with reason
+- If both models exist in the codebase → **eligible**
+- If either model is missing from the codebase → **ineligible**, mark it `[blocked]` with reason (e.g., "model X not yet implemented on main")
+
+Do NOT consider pending `[Model]` issues as satisfying the dependency — only models already merged to `main` count. This prevents bundling model + rule in the same PR.
 
 All `[Model]` issues are always eligible (no dependency check needed).
 
@@ -96,7 +97,7 @@ Ready issues (ranked):
 
 **If a specific issue number was provided:** verify it is in the Ready column. If it is blocked, STOP with a message explaining which model is missing.
 
-**If `--all`:** proceed with all eligible issues in ranked order (highest score first). **Dependency ordering override:** if a `[Model]` issue and a `[Rule]` that depends on it are both eligible, the Model MUST be processed before that Rule regardless of score. Blocked rules are skipped.
+**If `--all`:** proceed with all eligible issues in ranked order (highest score first). Models before Rules at same score. Blocked rules are skipped. After each issue is processed, re-check eligibility for remaining rules (a just-merged Model may unblock them).
 
 **Otherwise (no args):** pick the highest-scored eligible (non-blocked) issue and proceed immediately (no confirmation).
 
@@ -207,10 +208,10 @@ Completed: 2/4 | In Review: 3 | Returned to Ready: 1
 | Mistake | Fix |
 |---------|-----|
 | Issue not in Ready column | Verify status before processing; STOP if not Ready |
-| Picking a Rule whose model doesn't exist | Hard constraint: both source and target models must exist in codebase or be in a Ready Model issue (for `--all` mode) |
+| Picking a Rule whose model doesn't exist | Hard constraint: both source and target models must exist on `main` — pending Model issues do NOT count |
 | Missing project scopes | Run `gh auth refresh -s read:project,project` |
 | Forgetting to move back to Ready on total failure | Only move to In Review if a PR exists |
-| Processing Rules before their Model dependencies | In `--all` mode, ensure Models that unblock pending rules come before those rules regardless of score |
+| Processing Rules before their Model dependencies | In `--all` mode, re-check eligibility after each issue — a just-merged Model may unblock rules |
 | Scoring a variant as "related" | Weighted/unweighted variants or graph-subtype specializations of existing problems score 0 on C2 |
 | Not syncing main between batch issues | Each issue gets a fresh worktree from `origin/main` |
 | Worktree left behind on failure | Always clean up with `git worktree remove` in Step 5 |