Skip to main content
Glama

NASC: Networked AI Session Continuity

nasc_notify

Sends, reads, or acknowledges notifications. send requires for_who, message and human_requested=true, set only after explicit human permission for this message or ongoing notifications within an authorized scope; scoped ongoing permission covers later updates without a new request. Permission to build, fix, deploy, finish work, or generally coordinate with teammates does not itself authorize notifications. Without permission, routine updates stay in the current conversation, with no repeated requests to send FYIs. The flag is the caller's assertion, not consent verification. read returns compact newest-first rows for the caller (own rows plus 'all' broadcasts); FYI types older than 14 days drop out of the default view. ack takes notification_id or notification_ids (max 100). Discovery never grants access. Ambiguous destinations are refused. An explicit project that conflicts with the destination session is rejected.

Tool call template

Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "action": "<action>"
    },
    "name": "nasc_notify"
  }
}

Input Schema

NameRequiredDescriptionDefault
whoNoOperator name label. Attribution comes from the authenticated account.
tldrNoOptional for action=send: a one-line human-facing summary, about 140 characters. A reader shows tldr to the human and uses message as working context.
typeNo
brainNoOptional exact target org UUID. A send with no brain UUID may resolve an exact project plus named active NASC recipients across current memberships and live grants. Discovery never grants access, and ambiguous matches are refused.
guardNoOptional for action=send: opaque rider for the mailbox lane's client hooks. The server stores it verbatim and never interprets its shape beyond requiring an object.
limitNoread: 1-50. Default 10.
actionYes
statusNoread. Default active (unacknowledged).
for_whoNosend: name, alias, comma-separated list, array, or 'all' (broadcast). read: filters to that recipient, or 'all' for everyone; omitted, the caller.
messageNosend: the full detail for the AI or teammate session that reads it.
projectNoread: one slug, or 'all' for every project plus org-wide rows; omitted, the personal inbox. send: an exact slug or name locates named recipients; omitted, it comes from the destination, then source, session. 'all' is rejected on send.
verboseNoread: full bodies.
from_whoNoread: only rows from this person.
session_idNoOptional session id from nasc_start. Pins this call's authorization, org and project to that session.
for_sessionNoOptional for action=send: one session (id, continuity token or alias in the active brain) as the mailbox delivery target; an unresolvable value fails the send, and it defaults type to 'coordination'. For action=read: filters to that session.
include_staleNoread: includes FYIs older than 14 days.
human_requestedNosend only. true after explicit human permission for this message or ongoing notifications within the authorized recipients, purpose, and duration. An assertion, not verified consent.
notification_idNoack: the one to acknowledge. read: that one in full, any status or age.
continuity_tokenNoOptional host conversation id. Resolves the active session and pins this call's authorization, org and project to it.
notification_idsNoack: a batch, max 100.

Schema Changelog

Changes observed during successful MCP inspections.

  1. First observed

TDQS

A4/5.0
Behavior4/5

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

With no annotations, the description carries the full burden and does so well: it discloses that human_requested is an unverified self-assertion, that read is newest-first and drops FYIs older than 14 days unless include_stale is set, that ambiguous destinations are refused, and that a project conflicting with the destination session is rejected. Remaining gaps are around failure/partial-send behavior and whether ack has side effects on the sender.

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

Conciseness4/5

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

Purpose is front-loaded in sentence one and every subsequent sentence carries non-obvious policy rather than filler. The cost is density: permission rules, read retention, and ack limits run together in a single block without separation by action, making it slower to parse than it needs to be.

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 a 20-parameter, no-output-schema, no-annotation tool this is substantially complete on the risky paths (consent, retention, refusals). It stops short on return shape — "compact newest-first rows" is the only description of read output, and there is no output schema to fill that gap — and ack's result is never described.

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 already 90%, so the baseline is 3; the description earns above that by adding cross-parameter rules the schema does not state on its own — that 'all' is rejected for project on send, that for_who 'all' means broadcast, that ack caps at 100 ids, and that human_requested gates the whole send path. Several of its parameter notes (limit default, verbose, guard) merely echo the schema.

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 gives three concrete verbs against one resource ("Sends, reads, or acknowledges notifications"), and each action is then elaborated with its own semantics. It does not, however, distinguish itself from plausible siblings like nasc_mailbox_pull or nasc_route, so an agent must infer the boundary from the action names.

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?

Send usage is exceptionally well bounded: explicit human permission is required, generic permission to build/deploy/coordinate explicitly does not authorize it, and absent permission the guidance is to keep routine updates in the current conversation with no repeated FYI requests. Guidance for choosing read versus ack versus other tools is thinner — the description never says when to call read instead of a sibling retrieval tool.

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

Try in Browser

Glama MCP Gateway

Add one secure layer between your agents and this server.

Resources