Streams Lark/Feishu platform events as NDJSON via lark-cli event consume subprocesses.
npx skills add https://github.com/U1traVeno/skills --skill lark-event-u1traveno
Please help me install this Agent Skill. Skill: lark-event Source: https://github.com/U1traVeno/skills/tree/main/profiles/lark/lark-event Command: npx skills add https://github.com/U1traVeno/skills --skill lark-event-u1traveno
lark-cli event list and inspect params, output schemas, and jq root paths via lark-cli event schema before consuming. - Bounded or unbounded consumption: Stream events to stdout as NDJSON with --max-events / --timeout bounds, --jq filtering/projection, and --output-dir file capture. - Subprocess contract for AI agents: A stderr ready marker ([event] ready event_key=...), stdin-EOF graceful shutdown, structured JSON error envelopes, and documented exit codes (0-5) make it safe to orchestrate as a child process. - Use Case: Run lark-cli event consume im.message.receive_v1 --as bot to power a chatbot that reacts to incoming messages, or run multiple consumers in parallel for reactions, task updates, and meeting events with fault isolation per key. ## Quick Start Ask the AI to list available Lark event keys, then start consuming im.message.receive_v1 as the bot identity with a one-event sample to inspect the payload shape.High-intent search queries and answers about installing and using this skill.
Run lark-cli event consume im.message.receive_v1 --as bot to stream incoming messages as NDJSON to stdout. Use lark-cli event list to discover other EventKeys and event schema to inspect payload fields before writing jq filters.
Pass a jq expression via --jq, for example --jq 'select(.message_type=="text") | .content'. Check event schema <key> --json first for jq_root_path and field descriptions, since some fields like .content are pre-decoded and fromjson will fail on them.
No, each event consume process takes exactly one EventKey. To listen to N keys, start N subprocesses; they share a single local bus daemon, so overhead is small and each key gets independent filtering and identity settings.
Unbounded runs treat stdin EOF as a graceful shutdown signal, so running with /dev/null stdin or nohup causes immediate exit. Feed stdin a never-ending source like < <(tail -f /dev/null), or use --max-events / --timeout for bounded runs.
kill -9 skips the OAPI unsubscribe call for EventKeys with PreConsume hooks, leaking server-side subscriptions and causing 'subscription already exists' errors or duplicate delivery on restart. Use SIGTERM or close stdin instead.
Yes, the --as user|bot|auto flag selects the identity, but support varies per EventKey. For example task and whiteboard events support both, while VC meeting and minutes events require --as user with specific scopes granted.
yokingmaStreams Feishu/Lark real-time events as NDJSON via lark-cli with bounded run options.
SpaceZephyrCapture real-time Lark platform events via WebSocket long connection with NDJSON output.
yokingmaResolve Lark/Feishu contact names or emails into open_id values.
chenguojun06-starConsume real-time Lark events as NDJSON streams with jq filtering.