X Builder Digest

Turn curated X/Twitter builder activity plus optional podcast/blog items into concise English and Chinese markdown digests with first-mention person blurbs,...

Security Scan

The skill's code and instructions match its description: it transforms user-provided social-feed JSON into bilingual markdown digests and does not request credentials, external installs, or network access.

Purpose & Capability

Name/description ask for converting social-feed JSON into English/Chinese digests; the included script (scripts/render_digest.py), sample input, and sample outputs implement exactly that functionality and require only a JSON input and an output directory.

Instruction Scope

SKILL.md instructs the agent to accept structured JSON and produce two markdown files; the script only reads the provided input JSON and writes digest-en.md and digest-zh.md. The instructions explicitly forbid including tokens, session files, or local absolute paths in outputs and the script does not attempt to read other files or environment variables.

Install Mechanism

No install spec is present and the script uses only Python standard libraries (json, html, pathlib, sys). Nothing is downloaded or written beyond the digest outputs. This is low-risk for installation.

Credentials

The skill declares no required environment variables, credentials, or config paths and the code does not access os.environ or request secrets. The PEOPLE and ZH_OVERRIDES dictionaries are static and reasonable for producing blurbs.

Persistence & Privilege

The skill is not always-enabled and does not modify agent/system configuration. It writes output files only to the user-specified output directory and does not persist credentials or alter other skills.

Assessment

This skill appears coherent and safe for its stated purpose. Before using it: run it on non-sensitive sample data first; avoid passing JSON that contains private tokens, cookies, or other secrets (the script will blindly include any text from the input in outputs); run in a sandbox or dedicated directory if you are concerned about file writes; and if you want custom person blurbs for many people, consider supplying them from a separate, reviewed JSON rather than embedding untrusted data into the input.

Generate a bilingual digest from structured social-feed JSON.

Use this skill when

• The input is already collected from X/Twitter, podcasts, or blogs
• The user wants English and Chinese markdown outputs
• The user wants first-mention person blurbs
• The user wants Chinese output to be a real summary, not untranslated English text

Do not include in public outputs

• Private chat IDs, bot tokens, cookies, auth headers, session files
• Local absolute paths from the operator machine
• Internal repo names unless the user explicitly wants them kept
• Proprietary watchlists unless they are intentionally bundled as sample data

Workflow

1. Accept a structured JSON payload with keys like stats, x, podcasts, and blogs.
2. Generate two markdown files:
   • digest-en.md
   • digest-zh.md
3. In English output:
   • keep concise source-derived summaries
   • preserve original links
4. In Chinese output:
   • produce actual Chinese summaries, not English text with Chinese headings
   • add a short person blurb on first mention when available
   • preserve original links
5. If the user asks for a more executive style, compress to fewer bullets and stronger prioritization.

File layout

• Main script: scripts/render_digest.py
• Example input: references/sample-input.json
• Example output: references/sample-output-zh.md

Running

Use:

python3 scripts/render_digest.py input.json output_dir

This writes English and Chinese markdown outputs into output_dir.

Notes

• Keep the skill generic. Inputs are user-provided JSON, not hardcoded private data.
• If a team wants custom people blurbs, store them in a simple dictionary inside the script or in a separate JSON file.
• Prefer explicit overrides for high-value names rather than pretending freeform translation is always good.