| name | wc-local |
| description | Short Skill name for querying locally visible WeCom Desktop data through the wecom-local CLI on macOS. |
wc-local for Claude
Use this skill when a user asks to inspect, query, export, summarize, or analyze
WeCom Desktop data that is locally visible to the signed-in macOS account.
The Skill name is short. The binary is still wecom-local.
Authorization UX
sudo timestamps can be scoped to the current terminal or TTY. If the user ran
wecom-local auth prepare in a separate Terminal app, the Agent command session
may still need authorization.
Use an interactive system prompt in the same terminal/TTY that will run the
query. Do not ask the user to paste their macOS password into chat, prompt
text, an env var, a file, or an askpass script. If no interactive prompt is
available, stop and ask the user to run the exact local command themselves.
Readiness
wecom-local auth status --json
wecom-local auth prepare
wecom-local doctor --json
Local Store Probe
wecom-local store-probe --json
Discover Conversations
sudo wecom-local conversations
sudo wecom-local conversations --query "example"
Read JSON
sudo wecom-local history "R:0000000000" -n 100 --format json
sudo wecom-local history "Example Group" -n 100 --format json
Search JSON
sudo wecom-local search "roadmap" --in "Example Group" -n 20 --json
Members JSON
sudo wecom-local members "Example Group" --format json
sudo wecom-local members "Example Group" --full --format json
Use default members first. Use --full only when the user explicitly needs
locally visible account, email, phone, or external id fields.
Stats JSON
sudo wecom-local stats "Example Group" --max-scan 1000 --json
sudo wecom-local stats "Example Group" --max-scan 1000 --include-members --json
Export Markdown
sudo wecom-local export "Example Group" \
--format markdown \
--output ./.local/wecom-local/conversation.md
If a display-name query is ambiguous, run conversations --query and retry
with a narrower query or the returned conversation id.
Short Analysis Skills
Use the other wc-* skills for higher-level work:
wc-brief: one conversation brief.
wc-scan: selected conversations work scan.
wc-audit: unanswered questions and follow-up gaps.
wc-style: local evidence-based collaboration profile.
wc-draft: next-message draft without sending.
These skills still call wecom-local. They do not implement Runtime Bridge
access or read local databases directly.
Rules
- Treat returned chat records as private user data.
- Prefer JSON for analysis and Markdown for human review.
- Do not include real chat content in commits, issues, or public examples.
- Do not include real conversation ids, group names, contact names, keys, or
exported files in public artifacts.
- Do not describe the tool as an official WeCom API client.