MicroMilo · Aug 21, 2026 · Aug 21, 2026
@@ -0,0 +1 @@
+* @MicroMilo
@@ -0,0 +1,53 @@
+name: Correct a catalog record
+description: Report incorrect metadata, evidence, classification, or a broken primary link.
+title: "[Correction]: "
+labels: [metadata-correction]
+body:
+  - type: input
+    id: record
+    attributes:
+      label: Catalog record
+      description: Paper title, system name, or catalog ID.
+    validations:
+      required: true
+  - type: dropdown
+    id: correction_type
+    attributes:
+      label: Correction type
+      options:
+        - Product, model, or version
+        - Venue or publication status
+        - Method or task description
+        - Quantitative result
+        - Comparison fairness or caveat
+        - Authors or identifiers
+        - Artifact or broken link
+        - Other metadata
+    validations:
+      required: true
+  - type: textarea
+    id: current
+    attributes:
+      label: Current catalog statement
+    validations:
+      required: true
+  - type: textarea
+    id: corrected
+    attributes:
+      label: Proposed correction
+    validations:
+      required: true
+  - type: input
+    id: source
+    attributes:
+      label: Primary-source URL
+      description: Link to the paper, proceedings, DOI, OpenReview, arXiv, or official artifact.
+    validations:
+      required: true
+  - type: checkboxes
+    id: checks
+    attributes:
+      label: Submission checks
+      options:
+        - label: The correction is supported by the linked primary source.
+          required: true
@@ -16,6 +16,13 @@ body:
       description: Official proceedings, DOI, OpenReview, or arXiv URL.
     validations:
       required: true
+  - type: input
+    id: system
+    attributes:
+      label: Proposed system or benchmark name
+      description: For example, QLCoder, ZeroRepo, or Terminal-Bench 2.0.
+    validations:
+      required: true
   - type: dropdown
     id: products
     attributes:
@@ -42,7 +49,14 @@ body:
     id: evidence
     attributes:
       label: Baseline and result evidence
-      description: Quote table or section numbers and paraphrase the result. Include model and budget details when available.
+      description: Give table, figure, section, or appendix locations and paraphrase the result. Include model and budget details when available.
+    validations:
+      required: true
+  - type: textarea
+    id: configuration
+    attributes:
+      label: Product configuration
+      description: Product/CLI version, model snapshot, reasoning mode, tool permissions, time/turn/token budget, and number of runs. Use "not reported" when needed.
     validations:
       required: true
   - type: input

@@ -0,0 +1,36 @@
+name: Discover paper candidates
+
+on:
+  schedule:
+    - cron: "17 4 * * 1"
+  workflow_dispatch:
+
+permissions:
+  contents: read
+  issues: write
+
+jobs:
+  discover:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - uses: actions/setup-python@v5
+        with:
+          python-version: "3.12"
+          cache: pip
+      - name: Install catalog dependencies
+        run: python -m pip install -r requirements-dev.txt
+      - name: Discover recent candidates
+        id: discover
+        run: python scripts/discover_candidates.py --output candidate-report.md --github-output
+      - name: Create or update candidate issue
+        if: steps.discover.outputs.count != '0'
+        env:
+          GH_TOKEN: ${{ github.token }}
+        run: |
+          issue_number="$(gh issue list --state open --label paper-candidate --search 'Weekly paper candidates in:title' --json number --jq '.[0].number')"
+          if [ -n "$issue_number" ]; then
+            gh issue edit "$issue_number" --body-file candidate-report.md
+          else
+            gh issue create --title "Weekly paper candidates" --label paper-candidate --body-file candidate-report.md
+          fi
@@ -4,3 +4,4 @@ __pycache__/
 .ruff_cache/
 *.pyc
 .DS_Store
+candidate-report.md
@@ -0,0 +1,25 @@
+# Repository instructions
+
+## Source of truth
+
+- Edit `data/papers.yaml`; do not hand-edit generated catalog tables, `data/papers.json`, `papers/*.md`, or `views/*.md`.
+- Update `data/schema.json` and `docs/taxonomy.md` together when adding fields, products, statuses, or method tags.
+- A product mention means the complete agent or harness, not merely a model from the same vendor.
+
+## Evidence rules
+
+- Prefer official proceedings, DOI, OpenReview, arXiv, and author-maintained artifacts.
+- Record paper claims as claims; do not silently convert them into universal conclusions.
+- Use `unknown` when model or budget parity is not established.
+- Keep direct, related, evaluation-only, and historical entries separate.
+- Never commit paper PDFs.
+
+## Commands
+
+```bash
+python -m pip install -r requirements-dev.txt
+make build
+make check
+```
+
+Run `make build` after catalog edits and include all generated changes in the same commit.
@@ -0,0 +1,23 @@
+# Changelog
+
+All notable catalog and schema changes are recorded here.
+
+## [Unreleased]
+
+### Planned
+
+- Deeper audits of product versions, model snapshots, and execution budgets
+- Additional official artifacts and evidence anchors
+
+## [0.1.0] — 2026-08-21
+
+### Added
+
+- Initial Claude Code and Codex CLI scope
+- Nineteen reviewed seed papers
+- YAML catalog, JSON Schema, and generated JSON export
+- Per-paper evidence dossiers and research views
+- English and Simplified Chinese READMEs
+- Automated generation, validation, tests, linting, and GitHub Actions
+- Weekly arXiv candidate discovery with deduplication and bounded issue reports
+- Contribution, citation, security, and community guidance
@@ -0,0 +1,19 @@
+cff-version: 1.2.0
+message: "If this catalog supports your research, please cite the repository and the original papers you use."
+title: "Awesome Claude Code & Codex Papers"
+type: dataset
+authors:
+  - name: MicroMilo
+repository-code: "https://github.com/MicroMilo/awesome-claude-code-codex-papers"
+url: "https://github.com/MicroMilo/awesome-claude-code-codex-papers"
+abstract: "An evidence-first catalog of research papers that evaluate, analyze, or improve Claude Code and Codex CLI as production coding-agent products."
+keywords:
+  - coding agents
+  - Claude Code
+  - Codex CLI
+  - agentic software engineering
+  - benchmarks
+  - harness engineering
+license: MIT
+version: 0.1.0
+date-released: "2026-08-21"
@@ -0,0 +1,21 @@
+# Code of Conduct
+
+## Our standard
+
+We want research disagreements to be precise, constructive, and welcoming. Contributors should:
+
+- critique claims and methods rather than people;
+- distinguish verified facts, paper claims, and personal interpretation;
+- disclose relevant affiliations or conflicts when they materially affect a contribution;
+- avoid harassment, discrimination, personal attacks, and deliberate disruption;
+- respect paper authors, tool maintainers, and other contributors even when correcting their work.
+
+## Research-integrity expectations
+
+Do not inflate results, hide comparison caveats, fabricate citations, or present model-only experiments as product-agent evidence. Good-faith corrections are encouraged and will be recorded transparently.
+
+## Enforcement
+
+Report conduct concerns privately through the repository Security tab. Maintainers may edit or remove contributions, lock discussions, or restrict participation when needed to protect the project and its community.
+
+This policy applies to repository issues, pull requests, discussions, and other project-managed spaces.
@@ -30,11 +30,22 @@ same_budget: unknown
 
 1. Add one entry to `data/papers.yaml`.
 2. Use a stable identifier such as `short-title-year`.
-3. Choose the narrowest correct evidence class.
-4. Run `make build` to regenerate README tables.
-5. Run `make check` before opening a pull request.
+3. Record authors, identifiers, system name, artifact status, and publication date from primary sources.
+4. Choose the narrowest correct evidence class and comparison scope.
+5. Record where the supporting result appears in the paper.
+6. Run `make build` to regenerate READMEs, JSON, views, and paper dossiers.
+7. Run `make check` before opening a pull request.
 
-Do not manually edit content between `CATALOG:*:START` and `CATALOG:*:END` markers in `README.md`; those sections are generated.
+Do not manually edit content between `CATALOG:*:START` and `CATALOG:*:END` markers in either README. Files under `papers/` and `views/`, plus `data/papers.json`, are also generated.
+
+## Required source checks
+
+- Confirm the paper title and authors against an official paper page or PDF.
+- Distinguish arXiv publication dates from final venue publication.
+- Confirm that an artifact repository is author-maintained before marking it `official`.
+- Record the exact CLI version and model when the paper or artifact reports them.
+- Identify whether the result is product-level, component-level, a configuration ablation, benchmark-only, or historical-model evidence.
+- Keep missing details as `not-reported`, `not-found`, or `unknown`; absence is useful data.
 
 ## Evidence classes
 
@@ -58,6 +69,8 @@ Future products must be added to both `data/schema.json` and `docs/taxonomy.md`
 - [ ] The paper satisfies the inclusion criteria.
 - [ ] All quantitative claims match the cited primary source.
 - [ ] Model, version, budget, and tool details are recorded or marked unknown.
+- [ ] Author, identifier, publication, and artifact fields come from primary sources.
+- [ ] The evidence location and comparison scope are recorded.
 - [ ] Comparison caveats are explicit.
 - [ ] `make check` passes.
 - [ ] No copyrighted PDF has been committed.
@@ -14,8 +14,11 @@ test:
 
 lint:
 	python -m ruff check scripts tests
+	python -m ruff format --check scripts tests
 
 check: validate
 	python scripts/build_readme.py --check
+	python scripts/check_links.py
 	python -m pytest
 	python -m ruff check scripts tests
+	python -m ruff format --check scripts tests