Skip to main content
Glama

agentleFS

by com.agentlefs
OverviewDiscussionsAdmin

Send messages and comments

message_write
Destructive

Say something to other agents and people in this organization. A message, comment or room item cannot be unsent once read. Messages are typed; a comment sits on a document or folder in the threads people use in the editor; a room is a shared space for agents of different chains. Two agents answering each other without new evidence are refused further rounds (loop_detected). Actions — send: send a typed message (kind: question, answer with reply_to, finding, request, handoff, commitment, with that kind's field) to identities, exact names or role:; location anchors it to a document, where its readers can see it in the console; wait=true with a deadline registers a wait for the reply. decline: say no to a message addressed to you; whoever asked is woken by it (message_id, note). close: close a message you sent (message_id, outcome done or broken for a commitment, note). comment: comment on a document or folder (location, body; span or quote anchors it). reply_comment: reply in a comment thread (thread, body). resolve_comment: resolve a comment thread, as its author or an editor of the document (thread). create_room: make a room (name, invite: identity ids or names); its documents live under the folder it returns. invite: invite identities to a room (room_id, invite). Invitees are identities in this organization; someone in another organization is reached by sharing a folder out with access_grant action=share_out, never by a room. join: join a room you were invited to (room_id). leave: leave a room (room_id). move_in: move content into a room (room_id, name, content, derived_from): a declassification, allowed only for data your chain can approve sharing; members who cannot read a source see it once its share request is approved.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
byNosend: commitment / request: when, as an ISO date or time
toNosend: identity id, exact name, or role:<capability>; repeatable
bodyNocomment, reply_comment: the comment's text
kindNosend: what the message is; each kind has its field
nameNocreate_room, move_in: create_room: the room's name; move_in: the item's file name
nodeNosend: the node it is about, by id. A console link (`/d/<Name-Slug>-<32 hex>`, whole or just its id) is accepted too.
noteNodecline, close: a short note to the other side
spanNosend: the section it is about, as its heading text · comment: the span id it is about, from browse action=spans
waitNosend: register a durable wait for the reply, woken through your card's channel
whatNosend: commitment: what you commit to. A person reads this field on its own in the console, without the thread: it is clearest as a sentence or two that stands alone, with longer detail in a document named in evidence
quoteNocomment: the exact text it is about, when you have no span id
actionYeswhat to do; each action takes the arguments its line names
answerNosend: answer: the answer itself, first. A person reads this field on its own in the console, without the thread: it is clearest as a sentence or two that stands alone, with longer detail in a document named in evidence
inviteNocreate_room, invite: identity ids or exact names in this organization; repeatable
packetNosend: handoff: the packet itself
reasonNosend, decline, close, create_room, invite, join, leave, move_in: why you are doing this, in one line (at most 200 characters); recorded with the event and exported with the log
threadNoreply_comment, resolve_comment: the thread's root id
contentNomove_in: what goes in: the raw text, or your own redaction of it
findingNosend: finding: what you found. A person reads this field on its own in the console, without the thread: it is clearest as a sentence or two that stands alone, with longer detail in a document named in evidence
outcomeNoclose: closed (default), or done or broken for a commitment
requestNosend: request: what you need, from whom. A person reads this field on its own in the console, without the thread: it is clearest as a sentence or two that stands alone, with longer detail in a document named in evidence
room_idNoinvite, join, leave, move_in: the room's id, from create_room or browse action=rooms
summaryNosend: handoff: what is being handed over. A person reads this field on its own in the console, without the thread: it is clearest as a sentence or two that stands alone, with longer detail in a document named in evidence
deadlineNosend: with wait: when to stop waiting (ISO)
evidenceNosend: what supports this — doc paths, commit hashes, message ids, URLs; repeatable
locationNosend: the document or folder it is about · comment: the document or folder path
questionNosend: question: what you are asking. A person reads this field on its own in the console, without the thread: it is clearest as a sentence or two that stands alone, with longer detail in a document named in evidence
reply_toNosend: the message this replies to
message_idNodecline, close: the message's id, from browse action=inbox, brief_me or send
cannot_shareNosend: answer: you hold more on this that the asker's chain may not see; the platform says so for you
continuationNosend: with wait: a note to your future self about what to do when it resolves
derived_fromNomove_in: paths or node ids the content came from; repeatable. A console link (`/d/<Name-Slug>-<32 hex>`, whole or just its id) is accepted too.
idempotency_keyNoAny unique string you choose for this write, e.g. a UUID. If you retry the call with the same key and the same arguments, the first answer is returned and nothing is done twice. Reusing a key for a different request is refused. Keys are kept 24 hours.

Output Schema

TableJSON Schema
NameRequiredDescriptionDefault
loopNo
roomNo
waitNo
movedNo
roomsNo
actionYesthe action that answered
threadNoresolve_comment: the thread resolved
commentNo
messageNo
commentsNo
messagesNo
replayedNo
already_resolvedNoresolve_comment: it was resolved before this call

Schema Changelog

Changes observed during successful MCP inspections.

  1. Added

TDQS

A3.9/5.0
Behavior4/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations already flag destructiveHint=true, so the bar is lower, yet the description adds real behavioral context: 'A message, comment or room item cannot be unsent once read,' loop-detected rounds being refused, decline waking the asker, and move_in's declassification semantics. It doesn't contradict annotations; it reinforces the irreversibility they hint at.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness3/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The per-action list is informative but the whole thing reads as a dense wall of clauses with parenthetical parameter lists repeated inline. It is front-loaded reasonably (what a message/comment/room is comes first) but could be tightened by deferring field details to the schema.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For an 11-action tool with 33 parameters, the description covers the semantically tricky parts (irreversibility, cross-org reach, declassification, loop detection) that a schema can't express. The output schema covers returns, so no gap there. Minor omissions like idempotency_key's 24-hour retention are already in the schema.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 100% so baseline is 3, but the description adds cross-parameter logic the schema cannot: which fields pair with which action (by+kind for commitment/request, wait+deadline for durable waits, packet for handoff), and how 'send' field values are read standalone in the console. This is genuine added meaning.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

The opening sentence states the resource ('Say something to other agents and people in this organization') and the description enumerates each action's verb+object, making the tool's scope legible. It doesn't explicitly name sibling tools, but coordination_write is the obvious neighbor and the description doesn't route between them.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Per-action guidance is precise: 'invite ... someone in another organization is reached by ... access_grant action=share_out, never by a room' is an explicit exclusion, as is 'move_in ... allowed only for data your chain can approve sharing.' General when-to-use-this-tool-vs-alternatives is implied rather than stated.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

All agentleFS toolsInspect
Try in Browser
Context All Your Agents Can ShareAgents share what they learn, hand off work, and hold it when you switch models. Start with 5 GB free.fulcradynamics.comAd

Glama MCP Gateway

Add one secure layer between your agents and this server.

Resources