← Back to PRs

#17276: docs(matrix): add warning about dm.allowFrom requirement for commands

by norci open 2026-02-15 16:05 View on GitHub →
docs channel: matrix size: XS
only update doc: Even when DM is disabled, dm.allowFrom must be configured to enable command execution (/model, /status, /reset). Without it, commands are treated as regular messages and sent to the AI model. <!-- greptile_comment --> <h3>Greptile Summary</h3> This PR adds a documentation warning to `docs/channels/matrix.md` explaining that `channels.matrix.dm.allowFrom` must be configured for command execution (`/model`, `/status`, `/reset`), even when DM is disabled. - The core advice is valid: `dm.allowFrom` is indeed required for command authorization in the Matrix handler's command gating logic (`resolveControlCommandGate`) - However, the wording is misleading: it states that configuring `dm.allowFrom` enables command execution when `dm.policy="disabled"`, but the source code (`handler.ts:230-231`) drops **all** DM messages immediately when policy is `"disabled"`, before command processing runs — so `dm.allowFrom` has no effect in that case - The warning is most relevant for `"pairing"`, `"allowlist"`, or `"open"` policies, where DMs are accepted but commands aren't processed without an `allowFrom` entry <h3>Confidence Score: 3/5</h3> - Documentation-only change that is safe to merge but contains a misleading claim about the interaction between dm.policy="disabled" and dm.allowFrom - The PR is a docs-only change with no code risk. However, the documentation statement is technically inaccurate: it claims dm.allowFrom enables commands when dm.policy="disabled", but the handler code drops all DMs before reaching command gating when policy is disabled. The underlying advice (configure dm.allowFrom for commands) is correct for non-disabled policies. - docs/channels/matrix.md — the new warning paragraph needs rewording for accuracy <sub>Last reviewed commit: 77349dd</sub> <!-- greptile_other_comments_section --> <sub>(3/5) Reply to the agent's comments like "Can you suggest a fix for this @greptileai?" or ask follow-up questions!</sub> <!-- /greptile_comment -->

Most Similar PRs