Files
openworker/coworker/teams/tools.py
T
Rohit C Prasad cf436d1069 # team chat: own chat store, named workers, mention wakes, cancel interrupt (OPE-99)
ChatStore = groups + append-only messages + per-member cursors; agent posts wake mentions only, user posts wake everyone; post_chat(record_on_item) also lands the answer as an item comment.
Leads name workers (the callname is the handle everywhere); worker digests auto-carry the roster; gate checkbox is the user's call; canceling an assigned item now interrupts an in-flight worker.
2026-08-16 16:22:23 -07:00

366 lines
13 KiB
Python

"""Board and journal verbs as agent tools.
The verbs are generic on purpose (the connector-dialect play): the local TeamStore is
the default backing, and a Jira/Linear-backed dialect can implement the same tool
surface later. Registration is gated by the persona's `team:` trait — a lead gets the
full set, a worker gets the worker set, solo personas get none of this.
The engine decides `taint` (whether this agent touched untrusted content this
session) and passes it at construction — the model never self-reports provenance.
"""
from __future__ import annotations
from typing import Callable, Optional
import aisuite as ai
from .journal import JournalStore
from .model import Actor, BoardError, Role
from .store import TeamStore
LEAD_VERBS = ("create_item", "list_items", "transition", "comment", "assign", "link")
# Workers file items too (a bug spotted in passing, a follow-up) — new items land
# `open` and unassigned; nothing runs until the lead/user assigns them.
WORKER_VERBS = ("create_item", "list_items", "transition", "comment")
JOURNAL_VERBS = ("journal_append", "journal_read")
# Explicit schema: the auto-generator's normalizer strips every `title` key to drop
# pydantic metadata, which also deletes a PARAMETER named `title` from properties.
# Registered via `__coworker_schema__` (same escape hatch as todo_write).
_CREATE_ITEM_SCHEMA = {
"type": "function",
"function": {
"name": "create_item",
"description": (
"Create a work item (open, unassigned — work starts when it is"
" assigned). `criteria` is the acceptance criteria — what gets verified"
" before the item can be done; required. `parent` links it under"
" another item; `case` names its journal case (children inherit the"
" parent's case by default)."
),
"parameters": {
"type": "object",
"properties": {
"title": {"type": "string"},
"criteria": {"type": "string"},
"description": {"type": "string"},
"parent": {"type": "integer"},
"case": {"type": "string"},
},
"required": ["title", "criteria"],
},
},
}
def board_tools(
store: TeamStore,
*,
space: str,
actor: Actor,
taint: Callable[[], bool] = lambda: False,
) -> list:
"""The board verbs for one agent, pre-bound to its space and identity.
Authority is enforced twice on purpose: the returned set is role-filtered
(a worker never even sees `assign`), and the store re-checks every call —
the tool layer is convenience, the store is the gate.
"""
def create_item(
title: str,
criteria: str,
description: str = "",
parent: Optional[int] = None,
case: str = "",
) -> dict:
"""Create a work item (open, unassigned — work starts when it is
assigned). `criteria` is the acceptance criteria — what gets verified
before the item can be done; required. `parent` links it under another
item; `case` names its journal case (children inherit the parent's case
by default)."""
return _call(
store.create_item,
space,
actor,
title=title,
criteria=criteria,
description=description,
parent=parent,
case=case or None,
)
def list_items(state: str = "", assignee: str = "") -> dict:
"""List work items on the board, optionally filtered by state
(open/in_progress/blocked/review/done/canceled) or assignee."""
try:
return {"items": store.list_items(space, actor, state=state or None, assignee=assignee or None)}
except (BoardError, ValueError) as error:
return {"error": str(error)}
def transition(
item: int, to: str, comment: str = "", refs: Optional[list] = None
) -> dict:
"""Move a work item to a new state. Workers move their own item to
in_progress, blocked, or review (attach the blocker or a hand-off summary
as `comment`, and artifact pointers — branch, report, session — as
`refs`); done requires review verification first."""
return _call(
store.transition,
space,
actor,
item,
to,
comment=comment,
refs=[str(ref) for ref in refs or []],
taint=taint(),
)
def comment(item: int, body: str, refs: Optional[list] = None) -> dict:
"""Add a comment to a work item. Comments are durable and attributed —
answers that matter belong here, not in chat. `refs` attach artifact
pointers (branch, PR, report, file:line) to the item."""
return _call(
store.comment,
space,
actor,
item,
body,
refs=[str(ref) for ref in refs or []],
taint=taint(),
)
def assign(item: int, assignee: str) -> dict:
"""Assign a work item to a worker coworker. The item itself becomes the
worker's assignment — write the description and criteria accordingly."""
return _call(store.assign, space, actor, item, assignee)
def link(src: int, kind: str, dst: int) -> dict:
"""Link two work items: kind `parent` (dst becomes src's parent) or
`blocks` (src blocks dst)."""
return _call(store.link, space, actor, src, kind, dst)
verbs = LEAD_VERBS if actor.role in (Role.USER, Role.LEAD) else WORKER_VERBS
local = locals()
out = []
for name in verbs:
wrapped = _wrap(local[name])
if name == "create_item":
wrapped.__coworker_schema__ = _CREATE_ITEM_SCHEMA
out.append(wrapped)
return out
def journal_tools(
journal: "JournalStore",
*,
actor: Actor,
space: str = "",
taint: Callable[[], bool] = lambda: False,
) -> list:
def journal_append(
case: str,
body: str,
kind: str = "note",
item: Optional[int] = None,
entities: Optional[list] = None,
refs: Optional[list] = None,
) -> dict:
"""Append an entry to a journal case as you work: kind is finding,
evidence, decision, note (any observation), or raw (a capture like a log
excerpt — for large captures, save the full output to a file and journal
an excerpt that references it). `entities` are the concrete things it is
about (file paths, resource names, CVE ids) — they power later recall;
`refs` are pointers (file:line, commit, url)."""
return _call(
journal.append,
actor,
case,
body,
kind=kind,
space=space or None,
item=item,
entities=[str(entity) for entity in entities or []],
refs=[str(ref) for ref in refs or []],
taint=taint(),
)
def journal_read(
case: str,
item: Optional[int] = None,
author: str = "",
kind: str = "",
entity: str = "",
include_raw: bool = False,
limit: int = 50,
) -> dict:
"""Read a journal case, filtered: by item, author, entry kind, or entity.
Prefer narrow filtered reads over pulling the whole case. Raw captures
are skipped unless you pass include_raw or kind="raw"."""
try:
return {
"entries": journal.read(
actor,
case,
item=item,
author=author or None,
kind=kind or None,
entity=entity or None,
include_raw=include_raw,
limit=limit,
)
}
except (BoardError, ValueError) as error:
return {"error": str(error)}
local = locals()
return [_wrap(local[name]) for name in JOURNAL_VERBS]
# The staffing gate's schema carrier. Like propose_plan, the real handling lives in
# the TurnEngine (it needs the out-of-band approval round-trip): it emits
# TEAM_PROPOSED and waits; approval PRE-SPAWNS the worker sessions and returns the
# roster (actor ids) to the lead. This body only runs when no approver is wired.
_PROPOSE_TEAM_SCHEMA = {
"type": "function",
"function": {
"name": "propose_team",
"description": (
"Propose the worker coworkers you need for this board. Give EACH member"
" a short unique callname (`name`, e.g. 'nia', 'webb', 'checks') — it"
" becomes their handle for assignment and @mentions, and lets you staff"
" two of the same coworker. The user sees the roster and approves it;"
" approval creates the worker sessions and returns the handles. Only"
" team-capable worker coworkers may be proposed."
),
"parameters": {
"type": "object",
"properties": {
"members": {
"type": "array",
"items": {
"type": "object",
"properties": {
"persona": {"type": "string"},
"name": {"type": "string"},
"model": {"type": "string"},
"reason": {"type": "string"},
},
"required": ["persona", "name"],
},
},
"enable_chat": {"type": "boolean"},
"note": {"type": "string"},
},
"required": ["members"],
},
},
}
# The decomposition gate's schema carrier — the board-flavored sibling of
# propose_plan, usable in ANY permission mode (proposing costs nothing; the board
# only ever holds accepted work). The engine intercepts it; approval creates the
# items and returns their ids.
_PROPOSE_ITEMS_SCHEMA = {
"type": "function",
"function": {
"name": "propose_work_items",
"description": (
"Present your decomposition to the user as proposed WORK ITEMS for the"
" team board. Approval creates them on the board (ids come back in the"
" result); rejection returns feedback to revise. Each item needs a"
" title and acceptance criteria — what gets verified before it can be"
" done. This is not propose_plan: it carries no implementation steps"
" and works in any mode — it is how a lead plans and coordinates via"
" the board."
),
"parameters": {
"type": "object",
"properties": {
"items": {
"type": "array",
"items": {
"type": "object",
"properties": {
"title": {"type": "string"},
"criteria": {"type": "string"},
"description": {"type": "string"},
"case": {"type": "string"},
},
"required": ["title", "criteria"],
},
},
"note": {"type": "string"},
},
"required": ["items"],
},
},
}
def propose_work_items_tool() -> object:
def propose_work_items(items: Optional[list] = None, note: str = "") -> dict:
"""Present proposed work items ({title, criteria, description?, case?})
for the user's approval; approval creates them on the board."""
return {
"approved": False,
"error": "item proposals aren't available in this surface",
}
wrapped = ai.tool(
propose_work_items,
metadata=ai.ToolMetadata(
category="team",
risk_level="low",
capabilities=["team"],
),
)
wrapped.__coworker_schema__ = _PROPOSE_ITEMS_SCHEMA
return wrapped
def propose_team_tool() -> object:
def propose_team(
members: Optional[list] = None, enable_chat: bool = False, note: str = ""
) -> dict:
"""Propose the worker roster for this board (the staffing gate). Each member
is {persona, model?, reason?}. The user approves; approval creates the
worker sessions and returns their actor ids for assignment."""
return {
"approved": False,
"error": "team staffing isn't available in this surface",
}
wrapped = ai.tool(
propose_team,
metadata=ai.ToolMetadata(
category="team",
risk_level="medium",
capabilities=["team"],
),
)
wrapped.__coworker_schema__ = _PROPOSE_TEAM_SCHEMA
return wrapped
def _call(func, *args, **kwargs) -> dict:
try:
result = func(*args, **kwargs)
return result if isinstance(result, dict) else {"ok": True}
except (BoardError, ValueError) as error:
return {"error": str(error)}
def _wrap(func):
risk = "medium" if func.__name__ == "assign" else "low"
return ai.tool(
func,
metadata=ai.ToolMetadata(
category="team",
risk_level=risk,
capabilities=["team"],
),
)