Continuity Framework
ReviewAudited by ClawScan on May 10, 2026.
Overview
The skill is coherent and mostly local, but it creates long-lived AI memory about users, relationships, and sessions without clear retention, review, or deletion controls.
Review before installing if you do not want an agent to maintain persistent memories about you or your conversations. If you use it, keep the memory directory private, avoid feeding sensitive transcripts, disable heartbeat reflection unless explicitly desired, and periodically inspect or delete files under the configured memory directory.
Findings (3)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
Personal or sensitive details from conversations could be saved and reused in later sessions, including relationship inferences that the user may not have explicitly confirmed.
The skill explicitly stores relationship, identity, and reflection memory with long-term or permanent persistence. The provided artifacts do not show clear retention limits, user review gates, or deletion controls for these memories.
`relationship` | Connection dynamics | Long-term ... `moment` | Significant episodes | Permanent ... `identity.md` # Self-model and growth narrative ... `relationships/` ... `reflections/`
Use only with a trusted memory directory, review the generated memory files regularly, and require confirmation before saving inferred, speculative, relationship, or identity memories.
If enabled in heartbeat automation, the agent may update memory and generate follow-up questions after a session becomes idle.
The skill proposes a heartbeat-triggered post-session reflection workflow. This is disclosed and purpose-aligned, but it can cause memory updates after the user has stopped actively interacting.
**Trigger**: Heartbeat after conversation idle > 30 minutes; **Action**: Run continuity reflect; **Output**: Updated memories + questions for next session
Enable heartbeat reflection only if you want autonomous post-session memory updates, and keep it disabled or manual for sensitive conversations.
Users have less provenance information to verify who maintains the code or how the CLI is intended to be installed.
The skill includes a script and CLI-oriented documentation, but the registry metadata does not provide source provenance, a homepage, or an install contract.
Source: unknown; Homepage: none; Install specifications: No install spec — this is an instruction-only skill; Code file presence: scripts/continuity.py
Inspect the included script before use and avoid granting it broader automation until the source and installation path are clear.