# NousResearch/hermes-agent · Memory and cron context trust boundary fixes

## Open now

- 4 issues, 4 PRs open on GitHub.
- #26714: Telegram replies to cron deliveries can revive stale tasks via quoted cron context
- #44963: Make memory Write Gate approvals explicit and staged
- #63469: \[Bug\]: orchestrator trusts stale memory over canonical policy and corrupts multi-profile model routing
- #64681: \[Bug\]: quoted/ambiguous text can become a permanent model ban without confirmation
- PR #26725: fix(gateway): redact cron reply context in live turns
- PR #63544: fix(memory): demote memory context from authoritative to informational background (#63469)
- PR #72988: fix(memory): stage unconfirmed policy claims instead of silently persisting
- PR #104611: Improve memory Write Gate review flow — re-land on current main

These issues and fixes address cases where unconfirmed or stale context—from memory, cron outputs, or quoted replies—is treated as authoritative, potentially corrupting routing decisions or persisting policy claims without explicit approval. The work spans reducing trust in background context and adding explicit staging or redaction before such inputs affect live sessions.

## TL;DR

- 8 member(s): 4 open issue(s), 0 closed issue(s), 4 open PR(s), 0 merged PR(s), 0 PR(s) closed without merging.
- No fix has been merged to main; one or more PRs were attempted (see below).
- **Recommended:** review the prior/competing PR(s) below. A fix was attempted but not merged; your case may extend or revive that work.

## Maintainer resolution verdict

- Internal-tracker verdict over this cluster's issue members: `unresolved`. This is distinct from the fix status above. Weigh it yourself.

## How this cluster was grouped

- Link evidence (`STRUCTURAL_PARTIAL`): some members are connected by structural evidence (shared changed files, or a closing/superseded reference); the rest are joined by an automated relevance judgement over their text.
- Members outside that structural core may belong to a different underlying problem.
- Pairwise link density `0.18`: the share of member pairs that carry a link, counting a structural link in full and a relevance-only link at 0.36. `1.00` means every pair is linked.

## Open related issues

- #26714: Telegram replies to cron deliveries can revive stale tasks via quoted cron context
- #44963: Make memory Write Gate approvals explicit and staged
- #63469: \[Bug\]: orchestrator trusts stale memory over canonical policy and corrupts multi-profile model routing
- #64681: \[Bug\]: quoted/ambiguous text can become a permanent model ban without confirmation

## Prior & competing PRs

- **#26725: fix(gateway): redact cron reply context in live turns**. open
  - maintainer direction: category `uncategorized`; implication `needs_llm`
  - files: `gateway/run.py`, `tests/gateway/test_reply_to_injection.py`
- **#63544: fix(memory): demote memory context from authoritative to informational background (#63469)**. open
  - maintainer direction: category `duplicate_of`; implication `duplicate`; refs `5291925d6629385dc563cfe4de747f8bdb8bf405`, `#63544`, `#63544`
  - files: `agent/memory_manager.py`
- **#72988: fix(memory): stage unconfirmed policy claims instead of silently persisting**. open
  - maintainer direction: category `partial_fix`; implication `needs_llm`
  - files: `tests/tools/test_memory_tool.py`, `tools/memory_tool.py`
- **#104611: Improve memory Write Gate review flow — re-land on current main**. open
  - maintainer direction: category `uncategorized`; implication `needs_llm`
  - files: `gateway/slash_commands.py`, `hermes_cli/cli_commands_mixin.py`, `hermes_cli/commands.py`, `hermes_cli/write_approval_commands.py`, `tests/tools/test_write_approval.py`, `tools/memory_tool.py`, `tools/write_approval.py`, `website/docs/user-guide/configuration.md`, `website/docs/user-guide/features/memory.md`

## Fix status

- No merged fix on record.
- Subsystem: `agent`