research-citation-check

Research Citation Check

• Used to check whether the body citations of an existing {主题}_review.tex are really consistent with the content of the corresponding papers.
• Only when a fatal error is confirmed, minimally rewrite "the sentence containing that citation".
• Rendering PDF/Word preferentially depends on research-literature-review; preparing structured input does not depend on that skill.
• Compatible with prompt triggering of the old name check-review-alignment ;.check-review-alignment/ is still the stable historical workspace name.

Input

• work_dir: includes *_review.tex and .bib
• optional --tex: specify the tex file name

Core principles

• Do not change just for the sake of changing
• Do not touch when it cannot be confirmed
• Only change necessary sentences
• Preserve all LaTeX command structures

Workflow

Input

Provide the necessary input according to the user's request and the configuration file; missing information should be clearly listed and steps that depend on that input should stop.

Execution steps

• When a bug caused by a design flaw of this skill occurs in the user's environment, preferentially use bensz-collect-bugs to record it according to the specification in ~/.bensz-skills/bugs/, and it is strictly forbidden to directly modify the source code of the skill already installed in the user's local Claude Code / Codex.
• If the AI can still complete the user's task through a workaround, it should first record the bug, then continue to complete the current task.
• When the user explicitly requests public reporting actions such as "report bensz skills bugs", call the local gh and bensz-collect-bugs, and only upload newly added bugs to huangwb8/bensz-bugs; do not pull / clone the entire bug repository.

Dependency check

• Only enforce checking when rendering is executed research-literature-review, and during the transition period it may fall back to systematic-literature-review
• If it is only --prepare, rendering dependencies are not required to be available

Pre-check and locate

• Find *_review.tex and the corresponding .bib
• Stop immediately when any core file is missing

Structured context extraction

cd /path/to/research-citation-check
python3 scripts/run_ai_alignment.py --work-dir "/path/to/work_dir" --prepare

• Generate ai_alignment_input.json
• The input includes at least: sentence, bibkey, bibliographic metadata, DOI/URL, PDF abstract paragraph or BibTeX abstract

AI semantic check

• Evidence priority: PDF abstract paragraph > BibTeX abstract/title > inferred only from the sentence
• Each citation must be judged as one of:
  • fake_citation
  • wrong_citation
  • contradictory_citation
  • weak_support
  • overclaim
  • style_issue
• When it cannot be confirmed, keep it as is and record it in Warnings

Report

• The report includes at least:
  • Summary
  • Specific details
  • Critical Fixes (P0)
  • Warnings (P1)
  • Rendering Result
• The details of each citation include at least: title, DOI, original sentence, actual content of the paper, reasonableness assessment, issue severity

Rendering

cd /path/to/research-citation-check
python3 scripts/run_ai_alignment.py --work-dir "/path/to/work_dir" --render

Output

• {work_dir}/.check-review-alignment/ai_alignment_report.md
• {work_dir}/.check-review-alignment/ai_alignment_input.json
• Modified {主题}_review.tex
• Newly generated {主题}_review.pdf
• Newly generated {主题}_review.docx

Output management

Intermediate files of new tasks for this Skill are all written to ./.bensz-api/task-{yyyymmdd-hhmm}-{简短描述}/{skill名}/input|output|log/. The same task reuses one task root directory; only multi-Skill collaboration creates shared/. Official deliverables are not written to this directory; historical hidden directories only allow explicit compatibility reading, migration, or cleanup.

Validation

• See configuration in config.yaml
• Script entry point:scripts/run_ai_alignment.py
• Rendering dependencies:research-literature-review（fallback：systematic-literature-review）

Failure and recovery

Preserve error evidence and completed artifacts; retry from the most recent failed step only after the input, environment, or external dependencies have recovered.

Constraints

• P0: Must fix; minimal rewriting or correction of erroneous bibkeys is allowed
• P1: Warn only; do not rewrite
• P2: Skip entirely
• Prohibited:
  • Rewriting sentences that do not contain citations
  • Rewriting an entire paragraph
  • Introducing new bibkeys (except when fixing erroneous keys)
  • Fabricating paper content

Common hard constraints

• When the task requires writing to disk, use a unique ./.bensz-api/task-{yyyymmdd-hhmm}-{简短描述}/ root directory; place shared materials in shared/, and place Skill-specific materials in that Skill's input/、output/、log/。
• Official deliverables, source code, and official plans are saved according to project conventions and are not written into the task workspace; do not overwrite, delete, migrate, or write remotely without authorization.
• Project maintenance changes check BAC availability and record requirements, AI outputs, tool results, file changes, and verification summaries; BAC performs process auditing only and does not replace attribution, responsibility, or compliance judgments.
• Do not record API Keys, access tokens, passwords, Cookies, environment/credential files, private Prompts, identity information, local usernames, hostnames, or unnecessarily large raw data.
• File paths must be normalized and restricted to the authorized project scope; external URLs, subprocesses, and network access follow least privilege to prevent path traversal, SSRF, and command injection.
• Skill versions are recorded uniquely in their own config.yaml:skill_info.version; changes to public APIs, protocols, directories, or configurations are synchronized with documentation and CHANGELOG.md。
• Report only design defects in the Skill or Bensz infrastructure itself to bensz-collect-bugs; first desensitize and write to ~/.bensz-skills/bugs/, do not interrupt the current task, report publicly only when the user explicitly requests it, and do not directly modify the source code of Skills already installed by the user.

Skill-specific constraints

Do not exceed the scope declared by this Skill description and the process above; do not disguise unverified information as definitive conclusions.