x-twitter-source-evidence

Files (1)

• SKILL.md

SKILL.md (first lines)

# X/Twitter Source Evidence

Use this skill when a task needs grounded X/Twitter context before analysis,
documentation, product research, support triage, or content planning.

## Workflow

1. Define the evidence question before calling tools:
   - Which account, tweet, reply thread, keyword, follower set, or media asset matters?
   - Which output will consume the evidence?
   - Is the task read-only, or does it ask to post, reply, message, upload, monitor, or run a giveaway?
2. Prefer read-only collection first:
   - Search tweets and tweet replies.
   - Inspect tweet URLs or tweet IDs.
   - Look up public user profiles.
   - Export follower or following context when the workflow needs audience evidence.
   - Collect media metadata or download links only when the task needs media review.
3. Use TweetClaw through OpenClaw or the Xquik MCP/API when available for structured X/Twitter source data.
4. Keep credentials in approved runtime configuration. Do not paste API keys, cookies, OAuth tokens, or account credentials into files or prompts.
5. Produce a source packet:
   - Query, URLs, IDs, account handles, and time window.
   - Retrieved tweets, replies, user profiles, follower summaries, media references, and direct-message context when explicitly requested.
   - Gaps, rate-limit notes, and assumptions.
   - Links back to original public sources where available.

## Boundaries

- Do not post tweets, post tweet replies, send direct messages, upload media, follow, unfollow, like, retweet, schedule, or run giveaways from this skill.
- If the user asks for a write-like action, switch to the approved TweetClaw/OpenClaw flow and request explicit user approval at the tool boundary.
- Keep drafting, scoring, scheduling, analytics, and publishing decisions in the calling workflow. This skill supplies evidence only.
- Avoid claims that a source packet is complete. State what was searched and what was not searched.

## Output Shape

Return concise evidence that another agent or reviewer can audit:

```markdown
## X/Twitter Source Packet

- Objective:
- Sources searched:
- Queries and accounts:
- Evidence:
- Gaps:
- Recommended next step:
```

From ReactSphere/awesome-copilot-agents, MIT.

More from ReactSphere/awesome-copilot-agents