Agent Continuity & Memory OS

Recover the work, not a convenient story about it. Turn workspace evidence into reliable continuity through validated handoffs, bounded context recovery, and a decision history with reversal triggers.

What this product helps you do

Turn workspace evidence into reliable continuity through validated handoffs, bounded context recovery, and a decision history with reversal triggers.

Exactly what is included

• Session Handoff, Context Recovery, and Decision Ledger filesystem skills
• Offline handoff and decision validators
• A bounded workspace index and evidence-query utility
• Markdown and JSON templates with synthetic completed examples
• Direct filesystem editions plus honest guided editions for Custom GPT and Kimi Agent

How to use it

1. Inventory an approved workspace without following symlinks or common secret paths.
2. Label recovered claims as confirmed, supported, or unknown.
3. Preserve decisions with evidence, consequences, owners, and reversal triggers.
4. Create a validated handoff before changing operators or sessions.
5. Resume with the smallest reversible verification step.

Compatibility

• Codex: Filesystem skill : direct setup. Copy each complete skill folder into the Codex skills directory used by the installation.
• Claude Code: Filesystem skill : direct setup. Copy each complete skill folder into the project or user Claude Code skills directory.
• Kimi Code: Filesystem skill : direct setup. Copy each complete skill folder into the project or user Kimi Code skills directory.
• Custom GPT: Guided configuration : manual setup. Paste the supplied instructions, upload the text-forward knowledge files, and test in Preview.
• Kimi Agent: Guided configuration : manual setup. Use one supplied DOCX source per focused skill and complete the supplied setup path.

Who it is for

Teams using coding agents across long projects, context resets, operator changes, or approval-heavy delivery.

Who should not buy it

Teams expecting invisible cloud memory, automatic transcript ingestion, or unrestricted scanning outside an approved workspace.

Limitations

• The indexer cannot prove intent from a modified timestamp and deliberately excludes likely secret paths.
• No tool can recover evidence that was never recorded or is outside the approved scope.
• The system organises project memory; it does not replace backups, source control, access management, or human approval.

Privacy, safety, and claim boundaries

The supplied scripts run offline and contain no connector, credential field, hidden network call, live sending, dynamic code execution, or destructive command. Synthetic examples replace private business and customer data. Any external action remains separate work requiring explicit authority.

Version and integrity

Version 1.0.0. Archive SHA-256: 9805c42dd36a72c2b964fd552896f70febaa0c4120a2eb4722276e5cf0450203. The archive includes a file manifest, per-file checksums, version history, licence, support boundary, guided editions, fixtures, and tests.

Frequently asked questions

Does this product install itself or connect to my accounts?

No. The filesystem editions are copied into supported skill folders, while Custom GPT and Kimi Agent use guided setup. No included script signs in, calls an external API, sends a message, deploys an automation, or changes a live account.

Do the scripts need an API key?

No. The included validators, indexers, planners, and auditors run offline with Node.js 18 or newer. They read only the files you explicitly provide and make no network request.

Can I use the files with real client or customer data?

Use the minimum data needed, follow your organisation's approved privacy and access rules, and keep credentials and sensitive values out of the files. The examples are synthetic so you can test the workflow before introducing appropriately minimised real data.

Are Claude, Codex, GPT, or Kimi endorsing this product?

No. This is an independent RankThread product. Platform names describe compatibility only; RankThread is not affiliated with or endorsed by Anthropic, OpenAI, or Moonshot AI.

What support is included?

Support covers download integrity, identifying the supplied file, and clarifying the documented setup path. It does not include custom implementation, account administration, legal review, live sending, data migration, or a guaranteed business result.

What is the refund position?

Refund eligibility follows the digital-product terms shown at checkout. This listing adds no separate performance guarantee because the product cannot control platform changes, data quality, approvals, adoption, or commercial outcomes.

Related RankThread resources

• Digital Project Blueprint Kit
• App Development

This independent RankThread product is not affiliated with or endorsed by Anthropic, OpenAI, or Moonshot AI.