Version: ClaudeClaw 1.0.38
Problem
On Slack, long-running agent tasks (e.g. multi-step content generation that takes ~10 min) only post the final assistant message to the thread. While the agent works, the user just sees a silent ephemeral status ("Thinking…", "Reviewing findings…") with no per-step feedback — which reads as "the bot is stuck/dead" on long tasks.
What the code shows
In src/runner.ts, ClaudeClaw runs Claude with --output-format stream-json --verbose and processes the event stream via onChunk callbacks (it even tracks tool calls live). But this live streaming appears to be wired for Telegram — e.g. the comment around the streaming helper: "emitting text chunks via onChunk … (e.g. Telegram streaming)". The Slack adapter seems to only post the final assistant message.
Request — Slack/Telegram parity
Some way to surface progress on Slack during long tasks, e.g.:
- (a) post intermediate assistant messages to the thread as they stream, or
- (b) reflect the agent's latest text/step in the Slack assistant status (
assistant.threads.setStatus),
ideally behind a config flag (e.g. slack.streamProgress: true) so it's opt-in.
Use case: an internal content bot (blog articles, landing pages) where each task runs several minutes and the team wants to follow "writing… / generating images… / opening PR…" live. Happy to test any branch on our deployment. Thanks for ClaudeClaw! 🙏
Version: ClaudeClaw 1.0.38
Problem
On Slack, long-running agent tasks (e.g. multi-step content generation that takes ~10 min) only post the final assistant message to the thread. While the agent works, the user just sees a silent ephemeral status ("Thinking…", "Reviewing findings…") with no per-step feedback — which reads as "the bot is stuck/dead" on long tasks.
What the code shows
In
src/runner.ts, ClaudeClaw runs Claude with--output-format stream-json --verboseand processes the event stream viaonChunkcallbacks (it even tracks tool calls live). But this live streaming appears to be wired for Telegram — e.g. the comment around the streaming helper: "emitting text chunks via onChunk … (e.g. Telegram streaming)". The Slack adapter seems to only post the final assistant message.Request — Slack/Telegram parity
Some way to surface progress on Slack during long tasks, e.g.:
assistant.threads.setStatus),ideally behind a config flag (e.g.
slack.streamProgress: true) so it's opt-in.Use case: an internal content bot (blog articles, landing pages) where each task runs several minutes and the team wants to follow "writing… / generating images… / opening PR…" live. Happy to test any branch on our deployment. Thanks for ClaudeClaw! 🙏