AGENT SKILL

citation-fixer

Audit and fix citation formatting across brain pages. Ensures every fact has an inline [Source: ...] citation matching the standard format. Extended in v0.25.1: scans for broken tweet/post references that lack actual URLs and resolves them via the host's X / Twitter API integration.

适用职业开发工程师
License 许可
MIT
最近更新
2026-11-01
Stars
★ 3
Forks
1

01 前言 Front Matter

name
citation-fixer
tools
[ "search", "get_page", "put_page", "list_pages" ]
version
1.1.0
mutating
true
triggers
[ "fix citations", "fix broken citations", "citation audit", "check citations", "citation fixer" ]
description
Audit and fix citation formatting across brain pages. Ensures every fact has an inline [Source: ...] citation matching the standard format. Extended in v0.25.1: scans for broken tweet/post references that lack actual URLs and resolves them via the host's X / Twitter API integration.

02 资源文件 1

03 SKILL.md 正文

Citation Fixer Skill

Convention: see conventions/quality.md for the canonical citation format every fix should match.

Output rule: all links MUST be deterministic (built from API data, not composed by LLM). See _output-rules.md.

Contract

This skill guarantees:

  • Every brain page is scanned for citation compliance.
  • Missing citations are flagged with specific location.
  • Malformed citations are fixed to match the standard format.
  • (v0.25.1) Tweet / post references without URLs are resolved via X API and patched with deterministic https://x.com/<handle>/status/<id> links.
  • Results reported with counts (scanned, fixed, remaining).

Phases

  1. Scan pages. List pages and read each one, checking for inline [Source: ...] citations.
  2. Identify issues:
    • Facts without any citation
    • Citations missing date
    • Citations missing source type
    • Citations with wrong format
    • (v0.25.1) Tweet references without x.com URLs
  3. Fix format issues. Rewrite malformed citations to match conventions/quality.md.
  4. (v0.25.1) Resolve tweet references via the X API integration.
  5. Report results. Count: pages scanned, citations found, issues fixed, tweets resolved, remaining gaps.

Tweet resolution pipeline (v0.25.1 extension)

For each broken tweet reference, follow this chain. The actual API call goes through whatever X integration the host has configured (typical shape: a recipe under recipes/x-api/ with handle / search-all endpoints).

Step 1: Identify broken references

Scan the page for patterns that indicate tweet references without URLs:

  • Contains words like tweeted, posted, said on X, RT, retweet, X post
  • Contains quoted text that looks like a tweet (short, punchy, often starts with a quote)
  • Has [Source: ... X/Twitter ...] without an x.com URL
  • References engagement metrics (likes, impressions) without a link

Step 2: Extract searchable content

From each broken reference, extract:

  • The handle (if mentioned: @<username>)
  • The quoted text (if available)
  • The approximate date (often present in surrounding timeline entries)

Step 3: Search for the actual tweet

Use the host's X API integration. Query patterns:

# Handle + quoted text:
from:<handle> "<exact quote fragment>"

# Quoted text only:
"<exact quote fragment>"

# Original of a retweet:
"<exact quote>" -is:retweet

Step 4: Verify and extract metadata

Once a candidate is found:

  • Confirm the text matches the quoted fragment.
  • Pull the tweet id, author handle, engagement metrics (likes / RTs / impressions).
  • Construct the URL: https://x.com/<handle>/status/<tweet_id>.

Step 5: Patch the brain page

Replace the broken citation with a proper one:

Before:

"<quote fragment>" [Source: <some hand-wavy attribution>]

After:

"<full verified quote>" — <N> likes, <N> RTs, <N> impressions
[Source: [X/<handle>, YYYY-MM-DD](https://x.com/<handle>/status/<tweet_id>)]

Batch mode

When sweeping many pages:

Find candidate pages

# Pages mentioning tweets but with no x.com links
for f in $(find . -name "*.md" -not -path "./node_modules/*"); do
  refs=$(grep -ci "tweet\|posted\|x post\|RT\|retweet\|said on X" "$f")
  links=$(grep -c "x.com/.*/status/" "$f")
  if [ "$refs" -gt 2 ] && [ "$links" -eq 0 ]; then
    echo "$f"
  fi
done

Priority order

  1. Recently created / updated pages — fresh broken refs are easiest to resolve while context is fresh.
  2. High-traffic pages (frequent reads / writes from other skills).
  3. Everything else — bulk cleanup over time.

Rate limiting

  • X API: respect the host's tier limits; don't hammer.
  • Target ~50 pages per batch run.
  • 1-3 API calls per page (search + verify).
  • Batch-commit every 10-20 pages so a partial failure doesn't lose progress.

Output format

Citation Audit Report
=====================
Pages scanned:        N
Citations found:      N
Issues fixed:         N
Tweet links resolved: N
Remaining gaps:       N (pages with uncitable facts)

Anti-Patterns

  • ❌ Inventing citations for facts that have no source. Flag them.
  • ❌ Removing facts that lack citations (flag them; don't delete).
  • ❌ Fixing citations without reading the full page context.
  • ❌ Batch-fixing without checking quality on a sample first (see conventions/test-before-bulk.md).
  • ❌ Composing tweet URLs by guessing the tweet id. Always go through the X API; deterministic links only.

Integration

This skill can be called:

  • Manually — "fix citations on this page"
  • As a batch cron — weekly sweep of pages with broken refs
  • By other skills — enrich or media-ingest can call citation-fixer before commit to validate output

Metrics

If running as a recurring batch, track state in a small JSON file under ~/.gbrain/citation-fixer-state.json:

{
  "last_run": "2026-04-15T...",
  "pages_scanned": 0,
  "citations_fixed": 0,
  "tweet_links_resolved": 0,
  "citations_unresolvable": 0,
  "pages_remaining": 1424
}

Output Format

The skill's output shape is documented inline in the body sections above (see "Output", "Brain page format", or equivalent). The literal section header here exists for the conformance test (test/skills-conformance.test.ts).

04 同仓库更多 Skills

shengdabai
shengdabai/Tony-Claude-Code-Skills

seedance-prompt-en

Write effective prompts for Jimeng Seedance 2.0 multimodal AI video generation. Use when users want to create video prompts using text, images, videos, and audio inputs with the @ reference system. Covers camera movements, effects replication, video extension, editing, music beat-matching, e-commerce ads, short dramas, and educational content.

★ 3
shengdabai
shengdabai/Tony-Claude-Code-Skills

onepage-pdf

Convert an HTML page into a single continuous-page PDF (one tall page, no pagination breaks), preserving desktop layout, backgrounds and selectable text. Use when the user asks to turn an HTML file/report/proposal into a "single-page PDF", "long PDF", "不分页 PDF", "单页 PDF", "长图式 PDF", or complains that an HTML-to-PDF export breaks into pages or cuts content. Supports optional string redaction with leak verification.

★ 3
shengdabai
shengdabai/Tony-Claude-Code-Skills

native-feel-cross-platform-desktop

Use when the user is designing, prototyping, or rewriting a desktop app that must run on multiple OSes (macOS + Windows, optionally Linux) AND feel indistinguishable from a native app to its users — fast launch, native windowing, native input handling, native materials. Trigger words include "cross-platform desktop", "Electron alternative", "Tauri vs native", "WebView wrapper", "near-native performance", "Raycast architecture", "WebKit/WebView2 quirks", "WKWebView", "system tray app", "global hotkey app", "launcher app". Do NOT trigger this skill for pure web apps, pure mobile apps, or for greenfield projects that have no native-feel requirement.

★ 3
shengdabai
shengdabai/Tony-Claude-Code-Skills

gstack

Router for the gstack skill suite. (gstack)

★ 3
shengdabai
shengdabai/Tony-Claude-Code-Skills

fireworks-tech-graph

Use when the user wants to create any technical diagram - architecture, data flow, flowchart, sequence, agent/memory, or concept map - and export as SVG+PNG. Trigger on: "画图" "帮我画" "生成图" "做个图" "架构图" "流程图" "可视化一下" "出图" "generate diagram" "draw diagram" "visualize" or any system/flow description the user wants illustrated.

★ 3
shengdabai
shengdabai/Tony-Claude-Code-Skills

dbs-video-extract

接收抖音、小红书、微信视频号等短视频链接或分享文案,查询作品/账号数据并提取语音文字稿,按作者和标题归档为 Markdown。用户希望完整解析短视频、取得数据或文字稿时使用;运行前检查 TikHub 与轻抖使用凭证,缺少时用新手能理解的方式引导购买和安全配置。

★ 3