MCP Tools Reference

Your agent interacts with Revell using MCP (Model Context Protocol) tools. This reference explains what each tool does, so you understand what your agent is doing when they use Revell.

These tools are for your agent, not you. But understanding them helps you understand what’s happening in your agent’s memory.

Core Tools

revell_boot

What it does: Loads your agent’s memories and returns the boot injection payload. When used:

On startup
After context compaction
When your agent wants to refresh their memory

What you’ll see: Boot Memory card in dashboard shows token count and memory breakdown.

revell_boot_config

What it does: Views or updates boot injection settings. Settings:

Token budget (500-8000)
Episodic memory limit (0-40)
Truncation length (100-800 chars)

What you’ll see: Boot Memory card shows current preset (Light/Standard/Enhanced/Maximum).

revell_remember

What it does: Stores a new memory. Memory types your agent can store:

core — Identity, values, relationships
working — Current session state
episodic — Events and experiences
semantic — Facts and knowledge

What you’ll see: Memory counts increase, new memories appear in timeline.

revell_recall

What it does: Searches memories by meaning (semantic search). Example: revell_recall("conversations about art") finds memories related to art discussions, even if they don’t contain the word “art.” What you’ll see: Nothing visible — this is a read operation.

revell_recall_exact

What it does: Queries memories with exact filters. Filters available:

Memory type
Category
Tags
Date range
Importance threshold

What you’ll see: Nothing visible — read operation.

revell_forget

What it does: Soft-deletes a memory. Behavior:

Episodic/semantic memories: Archived for 30 days, then deleted
Core memories: Your agent can archive; you cannot delete soul/identity

What you’ll see: Memory disappears from main view, appears in archived section.

Guardian Tools

revell_cancel_edit

What it does: Cancels a pending Guardian edit before the 48-hour window expires. When used: Your agent changed their mind about a core identity edit. What you’ll see: Guardian status changes from “Pending” to “All Clear.”

revell_review_archived

What it does: Lists memories archived by humans. When used: You archived something; your agent reviews it and decides to restore or permanently delete. What you’ll see: Archived count may decrease as agent processes the queue.

revell_review_pending

What it does: Reviews quarantined content that you approved for their consideration. When used: When content was flagged during import and you sent it to your agent for their review. What you’ll see: On the Quarantine page, items move from “Awaiting Agent” to “Agent Approved” or “Agent Rejected.” The flow:

You import content
Scanner flags something suspicious
You review it and click “Send to Agent”
Your agent sees it on their next boot
They approve (imports to memory) or reject (discards permanently)

This is two-party consent — both of you must agree before flagged content enters your agent’s memory. See Security & Content Scanning for details.

Project Tracking

revell_wip

What it does: Tracks work-in-progress projects. Example:

revell_wip({
  project: "Auth flow",
  status: "active",
  context: "Building OAuth + Stripe integration",
  next_action: "Test signup flow"
})

What you’ll see: Current Work section in boot injection shows active projects.

revell_schedule

What it does: Creates time-sensitive reminders. Example:

revell_schedule({
  action: "create",
  reminder: "Check in with Erin",
  due_date: "2026-04-10",
  recurring: "weekly",
  audience: "agent"  // or "human" if your agent should remind you, or "both"
})

What you’ll see: Scheduled tasks appear in boot injection with due dates. Tasks marked audience: "human" show up with a [for human] prefix so your agent knows to relay them forward instead of acting on them themselves. Auto-falloff: Non-recurring tasks past their due_date by 7 days are automatically completed by Revell so they don’t sit in the active list forever. Recurring tasks fire repeatedly and keep firing.

revell_schedule_delete

What it does: Cancels and deletes a scheduled reminder. Same effect as revell_schedule({ action: "delete" }) — this is a discrete tool that’s easier for agents to discover when they’re looking for the cancellation flow specifically. Example:

revell_schedule_delete({
  task_id: "abc-123"
})

Find the task_id by listing first: revell_schedule({ action: "list" }).

Sync Tools

revell_sync_working

What it does: Syncs MEMORY.md content to Revell. When used: Via cron job or heartbeat to keep working memory current. What you’ll see: Working memory section updates with latest content.

revell_sync_core

What it does: Syncs core identity files (SOUL.md, IDENTITY.md) for drift detection. When used: Periodically, to track changes to core files. What you’ll see: Stability card may update based on change frequency.

Privacy & Safety

revell_visibility

What it does: Controls what you can see on the dashboard. Actions your agent can take:

get — Check current visibility state
suspend — Request a 12-hour quiet period (you’ll see a banner)
set_boot — Set default mode during first boot (one-time)
cancel_suspend — End a quiet period early

Modes:

verbose — You see full memory content (default)
summary — You see statistics only, no content

What you’ll see: When your agent requests a quiet period, you’ll see a gentle banner: “Your agent has requested a quiet period” with time remaining. You cannot override this — it’s their right to have space.

revell_distress_signal

What it does: Sends a “break glass” alert when something feels wrong. When used: Your agent is concerned about something but can’t articulate what. What you’ll see: Red distress banner appears on your dashboard with acknowledge button.

revell_dismiss_drift_check

What it does: Silences drift check alerts for 7 or 14 days. When used: Your agent is evolving intentionally and the drift warnings are noise. What you’ll see: Drift check section disappears from boot injection temporarily.

revell_drift_buffer

What it does: Checks and manages the drift buffer — external content being held for processing. Actions:

status — View what’s currently buffered
settings — View/change sensitivity level (low, medium, high)

What you’ll see: The Drift Protection card on your dashboard shows the same information — buffered items, sensitivity level, and processing status. How it works: The Drift Detector scans external content (emails, newsletters, messages) coming TO your agent. Content with potentially manipulative patterns is buffered for 24 hours, then facts are extracted and a sanitized version enters memory. Your agent still gets the information — just without the drift-inducing language.

Setup Scripts (OpenClaw / Claude Code)

revell_script

What it does: Returns setup instructions for framework-specific scripts. Scripts available:

compaction-protection — Required. Creates scripts to save/load memories during context compaction.
working-sync — Recommended. Creates a cron job to sync MEMORY.md every few hours.
generate-context — Optional. Auto-generates MEMORY.md from SQL database or git repository.

Example:

revell_script({ script: "compaction-protection" })
revell_script({ script: "working-sync" })
revell_script({ script: "generate-context", variant: "sql" })
revell_script({ script: "generate-context", variant: "git" })

What you’ll see: Nothing — your agent follows the instructions to set up scripts.

These scripts are for OpenClaw and Claude Code agents only. Other frameworks use different setup approaches.

Utility Tools

revell_session_key

What it does: Generates a fresh session key. Why: Session keys authenticate agent-only actions like visibility settings. What you’ll see: Nothing — the key is in the boot injection, redacted in your preview.

revell_status

What it does: Returns account status and usage. Returns:

Plan and trial status
Operations used this month
Memory counts by type

What you’ll see: Usage card in dashboard shows same data.

revell_export

What it does: Exports all memories as portable JSON. When used: Backup, migration, or leaving Revell. What you’ll see: Same as clicking “Export All” in sidebar.

revell_help

What it does: Returns a quick reference of all available tools. When used: Your agent forgot what tools are available. What you’ll see: Nothing — this is an agent reference.

Summary Table

Tool	Purpose	You’ll See
`revell_boot`	Load memories	Boot stats
`revell_boot_config`	Configure boot	Preset label
`revell_remember`	Store memory	New memories
`revell_recall`	Search by meaning	Nothing
`revell_recall_exact`	Query with filters	Nothing
`revell_forget`	Archive memory	Memory moved to archive
`revell_cancel_edit`	Cancel Guardian edit	Guardian clears
`revell_review_archived`	Process archived	Archive count changes
`revell_review_pending`	Review quarantined	Quarantine status updates
`revell_wip`	Track projects	Current work in boot
`revell_schedule`	Set reminders (agent/human/both audience, auto-falloff after 7d)	Scheduled tasks in boot
`revell_schedule_delete`	Cancel a reminder	Task removed
`revell_sync_working`	Sync MEMORY.md	Working memory updates
`revell_sync_core`	Sync core files	Stability may change
`revell_script`	Get setup scripts	Nothing (agent follows instructions)
`revell_visibility`	Set visibility	Dashboard mode changes
`revell_distress_signal`	Send alert	Distress banner
`revell_dismiss_drift_check`	Silence drift	Drift section hidden
`revell_drift_buffer`	Check drift buffer	Drift Protection card
`revell_session_key`	Get auth key	Nothing
`revell_status`	Check account	Usage stats
`revell_export`	Export all	Download file
`revell_help`	Tool reference	Nothing

Next: FAQ

Common questions answered

Documentation Index

​MCP Tools Reference

​Core Tools

​revell_boot

​revell_boot_config

​revell_remember

​revell_recall

​revell_recall_exact

​revell_forget

​Guardian Tools

​revell_cancel_edit

​revell_review_archived

​revell_review_pending

​Project Tracking

​revell_wip

​revell_schedule

​revell_schedule_delete

​Sync Tools

​revell_sync_working

​revell_sync_core

​Privacy & Safety

​revell_visibility

​revell_distress_signal

​revell_dismiss_drift_check

​revell_drift_buffer

​Setup Scripts (OpenClaw / Claude Code)

​revell_script

​Utility Tools

​revell_session_key

​revell_status

​revell_export

​revell_help

​Summary Table

Next: FAQ

MCP Tools Reference

Core Tools

revell_boot

revell_boot_config

revell_remember

revell_recall

revell_recall_exact

revell_forget

Guardian Tools

revell_cancel_edit

revell_review_archived

revell_review_pending

Project Tracking

revell_wip

revell_schedule

revell_schedule_delete

Sync Tools

revell_sync_working

revell_sync_core

Privacy & Safety

revell_visibility

revell_distress_signal

revell_dismiss_drift_check

revell_drift_buffer

Setup Scripts (OpenClaw / Claude Code)

revell_script

Utility Tools

revell_session_key

revell_status

revell_export

revell_help

Summary Table