* feat(connectors): the backend serves a connector's tool list, cached for 24 hours
The Connectors page opens one app and shows every tool it has. The backend
had no way to read that list.
- `tools/connectors/portal/`: a client for the portal's tool-list route and a
JSON cache under the Hermes home, one file per portal origin and connector.
An entry is fresh for 24 hours. After that the read revalidates with the
stored ETag: 304 keeps the list, 404 deletes the entry, an upstream failure
serves the stored list marked stale, and a 401 never serves the cache.
- `connectors.tools {slug, refresh}`: account-level, routed by `profile`, no
chat session. Errors carry a fixed `reason` from one closed set on the rail.
- Every connector model that is not operation state moves into
`tui_gateway/contracts/connectors.py`. Handlers that no chat session owns
live in `tui_gateway/methods_connectors_account.py`.
The wire model is tolerant: an unknown facet reads as unclassified and one odd
tool never blanks a connector.
* feat(connectors): catalog, accounts and member tool rules by RPC
The Connectors page needs the app catalog, the connected account of one app,
a way to disconnect it, and the member's own on/off rules. None had an RPC.
- `connectors.catalog`: name, description, category and logo of each app.
- `connectors.accounts`, `connectors.accounts.remove`: read the accounts at
the tool gateway and remove one by id.
- `connectors.policy.get`: the rule layers that apply to the member, widest
first. The body is a union on `mode`, so a reader can name who turned a
tool off.
- `connectors.policy.set`: one change, a union on `type` (the tools of one
connector, or one connector on or off), with the revision the user saw. A
stale revision answers `POLICY_CONFLICT`. The backend composes the upstream
write in one pure function, so no renderer learns the upstream rules.
- Bundled MCP manifests can name their hosted twin with `connector:`, so the
page can show one card per app.
* feat(connectors): connect an app without a chat session
Every connector RPC took a `session_id`, and a connect that did not come from
the model's tool call minted a link with no watcher. The Connectors page has
no chat session, and its card must flip to connected by itself.
- `connectors.list`, `connectors.connect`, `connectors.operation.status`,
`connectors.operation.wake` and `connection.respond` take `owner`, a union
on `type`: `session` (today's behaviour and authorization) or `account`
(routed by `profile`, authorized by the live transport like `mcp.*`).
`session_id` is gone from these params; every desktop caller sends `owner`.
- An account connect runs the same operation lifecycle on a background
thread, under the profile's scope, so the watcher reads the account and
settles the operation. A second connect for an app that is already
connecting returns the open operation and mints nothing.
- `connection.update` carries `owner`. An account operation has no session to
address, so its updates go out on the session-less broadcast path.
* feat(mcp-catalog): eighteen more bundled entries name their hosted connector
A bundled MCP entry and a hosted connector for the same app are one card
on the Connectors page only when the manifest names its hosted twin.
Linear and Notion had the field. These entries get it too: airtable,
asana, attio, calendly, dropbox, figma, railway, supabase, todoist,
betterstack, canva, cloudflare, datadog, intercom, neon, sentry, stripe
and vercel. Atlassian maps to two hosted connectors and Prisma Postgres
is not clearly the same app, so both stay without one.
* refactor(connectors): the account handlers share one gate, one params model and one write table
The six account-level handlers each repeated the availability gate, the
auth catch and the catch-all reply. One decorator now owns that, and each
handler validates its params with its contract model instead of a ladder
of isinstance checks. The five connection RPCs share one guard for the
unexpected-failure reply.
The four write composers for the member rules were the same function
with a different list key and polarity. They are one table now.
The owner union lives in contracts/common.py, so the params side and the
event side stop declaring it twice and the import cycle is gone.
An account operation start carries one event and a flag, so the wait for
the sign-in link blocks instead of polling every 50 ms. run_operation
loses its two account-only parameters; drive_operation is the second
entry point.
Tests: four deleted (they exercised pydantic or the mock), three merged
into tables, two added (a client that still sends the old top-level
session_id is refused; all six account RPCs run off the server loop).
The shared reply helper and the HTTP and managed-client fakes move to
one place each. Comments are one line or gone.
* fix(connectors): a missing tool-list route reads as "unavailable", not "connector gone"
The tool-list read treated every 404 as the portal's "this connector is
not in the catalog" answer. It deleted the cache entry and answered
CONNECTOR_NOT_FOUND, so a page would offer to remove an app that is
connected and works. A portal that does not serve the route yet answers
a bare 404 for every app.
Only the portal's own {"error": "connector_not_found"} means the
connector is gone. Any other 404 is now a tool-list outage: the cached
list is served as stale, or the RPC answers TOOLS_UNAVAILABLE.
* fix(connectors): a connect from the page returns to the app after sign-in
The sign-in link carries a return target only when the session's surface
is the desktop. A chat session binds that surface. An account-owned call
has no chat session, so nothing bound it: the link was minted without a
return target and the browser ended on the portal's done page instead of
coming back to Hermes.
Every account-owned call now runs with the process's own surface bound,
next to its profile scope. The operation thread copies that context, so
the first link and every reissued link carry the return target and the
operation id.
* test(connectors): defer the new connector RPC coverage
The tests for the new account RPCs, the portal client, the tool-list cache
and the rule composer leave this PR and come back in one later change, after
the API is settled. The same was done for #111008.
Kept: the edits that existing tests need because the five connection RPCs
now take `owner` instead of `session_id`, and the rename of the managed
client seam.
Removed: six new test files, their two fakes and the gateway conftest, and
the new cases in test_mcp_catalog.py, test_connectors_gateway_client.py,
gateway-rpc.test.ts and notifications.test.ts. Reverting this commit restores
all of them.
* fix(cli): the connection panel hands the tool thread back at once
The classic CLI's connection callback waited on a queue for the user's first
decision. The operation's watcher starts only after the callback returns, and
the watcher is what polls a hosted account, runs the 300-second deadline and
sees Ctrl+C.
For a hosted connector the panel opens on the sign-in link, where the only
key that filled the queue was Cancel. The account was never polled: the user
signed in, the panel never changed, and Esc reported the app as skipped.
Ctrl+C set the interrupt flag but left the thread parked on the queue, so the
turn never ended.
The callback now opens the panel and returns, as the gateway's callback does
for the desktop and the Ink TUI. The panel's actions already reach the
operation through apply_answer on the UI thread, so the queue is removed. An
install with a form still waits for Connect, because the backend starts no
work for a pending row. Ctrl+C now settles the operation as `interrupt`, and
open rows become `not_connected`.
Checked on the e2e rig with the fake tool gateway: hosted connect completes on
the third status read; Ctrl+C ends the turn and the polling stops; an MCP
install with a plain and a secret field still saves config and both values.
* fix(connectors): "run it again" lives in the library, so the classic CLI can use it
Making a new sign-in link for a failed or expired hosted connector was
implemented only in the JSON-RPC layer (`_reissue`). The classic CLI does not
go through JSON-RPC: its Connect button on a failed row called apply_answer,
which does nothing for a hosted operation because it has no MCP runner. The
panel showed "Waiting…" until the deadline.
`tools.connectors.run.reissue(operation, names)` now holds the checks and the
per-kind action, and returns a refusal reason or None. The gateway maps each
reason to the same JSON-RPC error as before. The CLI calls it for a hosted
row; a refusal is shown on the row. MCP rows keep their path, because Connect
on a failed MCP row re-sends the form values.
Checked on the e2e rig: a scripted failed sign-in, then Connect: a second mint
with `reinitiate: true`, a new link with a new connection id, then connected.
* feat(connectors): the account list and disconnect go through the portal
`connectors.accounts` and `connectors.accounts.remove` called the tool
gateway. They now call the portal's account-management routes
(`GET /api/v1/connectors/accounts`, `DELETE /api/v1/connectors/accounts/{id}`),
which apply the organisation membership checks and write the disconnect audit
row. There is no fallback to the gateway when the portal is unavailable, and a
removal is never retried.
The read of ONE account stays on the gateway (`GET v1/connectors/accounts/{id}`):
the portal has no such route, and the operation watcher polls it once per second.
`ConnectorClient.list_accounts` and `delete_account` are removed. The removed
account's reply model carries `connector`, which both services send.
* fix(connectors): the account RPCs answer what the portal really sends
Checked against the portal source and against the staging and production
services.
- Errors are read from the upstream error code, not the HTTP status. A rule
write answered 409 for a stale revision and for a user with no organisation;
both read as "the policy changed". `org_required` is now `ORG_REQUIRED` and
403 `no_access` is `ORG_ACCESS_DENIED` on every account RPC; only a rejected
sign-in is `NEEDS_NOUS_AUTH`. `connectors.list` and `connectors.connect` with
the account owner map these too.
- `connectors.policy.get` and `connectors.policy.set` carry `effective`: the
portal's own result for this user, with its stamp and without provider or
subject ids. Nothing is recomputed locally.
- A rule write needs the revision the user saw: `expected_revision` is required
and must be a revision string; a bad one is refused before any HTTP call.
- A tool row carries `no_auth`; a list without the upstream flag is an invalid
answer, not `false`.
- `connectors.accounts.remove` returns the app of the removed account. An
invalid id is `INVALID_PARAMS`.
- The tool-list cache is per signed-in member (a hash of the token's `sub`),
so two Nous accounts on one profile do not share entries.
- A malformed slug is a local error, not a 404 from a server nobody called.
Live, staging: no revision and a malformed revision refused locally; a good
revision wrote one disabled Gmail tool and returned it in `effective`; the
same revision again answered `POLICY_CONFLICT`; the list row showed the tool;
the restore brought the member rules back to the start. Live, staging and
production, read-only: all 60 tool lists (5483 tools) parse.
* fix(connectors): the operation RPCs match their contract; a settled card cannot start a new link
Found by two adversarial reviews of the RPC layer and its types.
- `connectors.connect` from a chat session with no open operation is refused
(`UNKNOWN_OPERATION`). It used to call `manage_connections` through the tool
registry with no card: it made a link nobody watched, returned a reply
without the required `settled` field, and named an operation that was never
registered. There is one way into an operation: the agent's call, or the
account owner's `connectors.connect`. "Run it again" inside an open
operation is unchanged.
- `connection.update` for a session is routed by session key AND profile; two
profiles with the same key no longer cross-deliver a sign-in link. The event
payload gets the same redaction as the RPC replies.
- `connection.respond` runs on the long-handler pool: an approval can start MCP
OAuth discovery, which blocked every RPC of the gateway while it ran.
- `connectors.list` rows are a closed snake_case model: `connector`, `enabled`,
`connected`, `connection_status`, `status_reason`, `gateway_disabled_tools`.
The last one is display data: the gateway enforces the rules, the backend
only passes the list on. The phantom `name` and `description` are gone, and
the desktop uses the generated types instead of hand-written copies.
- `tools_listing` (model-only data) no longer rides on `connectors.operation.status`.
- `unavailable` is removed from the target states and settle reasons: nothing
produces it. The contract generator now fails when a contract enum and its
domain enum differ.
- `ConnectorErrorReason` is part of the generated TypeScript and OpenRPC.
- The desktop sends `connection.respond` on the socket that holds the session,
as wake and reissue already did.
- Contract violations are logged every time, at error level.
- An account connect whose prepare step is slow returns the live operation
instead of an error while the operation keeps running.
- The MCP-manifest `connector` field leaves this PR (it moves to a later one
on top of the catalog-reader change). `hermes_cli/mcp_catalog.py` and
`optional-mcps/` are untouched by this PR again.
anti-slop: no net-new findings (15 touched files).
* fix(connectors): the model gets no sign-in link wherever a card exists; side agents cannot connect
The flag that tells the model "a connection card exists" was the session
platform (`== "desktop"`). The Ink TUI and the classic CLI also draw a card,
so there a connector call on an unconnected app handed the model the raw
`connect_url` and told it to pass the link to the user.
- The agent turn now declares how a link can reach the user
(`tools/connectors/turn.py`): CARD when the agent was built with a
connection callback, SIDE for a subagent or a background turn, LINK for a
headless run (`-q`, cron, ACP, api_server, messaging). It is set once per
tool batch in the agent loop and read by the connector dispatch path, which
never sees the agent. The session platform decides return-to-app only.
- CARD: the result carries `connect_card_available` and our hint, never the
link and never the gateway's own hint.
- SIDE: subagents (`delegate_tool`), gateway background turns and the classic
CLI `/bg` are built with `side_agent=True`. They hold no `manage_connections`
tool on any path that derives the tool list, and a connector call on an
unconnected app gets no link, only "report this to the main agent".
- LINK is unchanged.
- The hosted path with no card builds a detached operation, as the MCP path
does, so no `connection.update` is emitted for an operation no client asked
for. Names and docstrings that said "off desktop" now say "no card".
- A settled card is dead on the desktop: `reissueConnectionTarget` and
`respondToConnectionRequest` share one guard and send nothing for a settled
or unknown operation.
- The model-facing settled result no longer carries `connection_id`; the model
repeated it to the user.
Shown on the real clients with a real model (rig, fake tool gateway): Ink TUI
and classic CLI get `connect_card_available` and no link, the model opens the
card, the account connects, the retried call succeeds; `-q` still gets the
link; a subagent and a background turn have no `manage_connections` and get
the no-link hint; on the desktop a card settled with Continue has no enabled
control and sends no RPC.
* feat(tools): every call made through tool_search + tool_call shows a real label on all three clients
A bridged call showed as a generic `tool_call` row in the Ink TUI and as
`⚡ tool_call` in the classic CLI, because the display looked the name up in
the tool registry and bridged names are made at run time. The desktop labelled
only batches that were all hosted connector calls, by parsing names itself.
- `tools/tool_labels.py` is the one place that turns a bridged call into a
label: kind, app, action, emoji and text. Hosted: `connectors__gmail__GMAIL_SEND_EMAIL`
→ "Gmail · send email". MCP: "Linear · list issues". A local deferred tool
keeps its own emoji, verb and primary-argument preview. A batch gets exactly
one label per entry, always; an entry with no name gets a generic label.
- Classic CLI: one row per inner call; the duration on the last row; the
failure text on the row of the call that failed. With friendly labels off
it prints what it printed before.
- Gateway: tool start, progress and complete events and stored transcript rows
carry a typed `labels` field. It does not depend on the classic CLI's
display setting. Clients no longer parse tool names.
- Ink TUI: rows from the labels; the verbose trail keeps Args and Result.
- Desktop: `ConnectorExecution` renders hosted, MCP and mixed turns from the
labels, one row per call. The labels reach the row under a key no tool
argument can use. The connect card it drew under a failed tool result is
gone: after `CONNECTION_REQUIRED` the one way in is the agent's own
`manage_connections` call.
- `tool_search` and `tool_describe` rows read "Searching tools · <query>" and
"Reading tool details · N tools".
Shown on the real desktop (video and screenshots), the Ink TUI and the classic
CLI with the rig: hosted rows, MCP rows, a two-entry batch, a failed entry, a
`CONNECTION_REQUIRED` row with no card under it, labels after a reload, and the
desktop rows with the classic CLI setting off.
* fix(connectors): the model can tell "hosted tools unavailable" from "no such tool"; manage_connections routes MCP names correctly
- A failed hosted search or describe used to return nothing, by design, so the
model saw only local tools and told the user that a connected app was
missing. The local results are unchanged; when the hosted leg failed, the
`tool_search` and `tool_describe` results carry
`connectors: {status: "unavailable", reason: "unreachable" | "sign_in_expired"}`
and one hint line. A rejected token is `sign_in_expired`; an entitlement
refusal or a shut gate adds nothing. `tool_describe` no longer lists those
names under `not_found` next to "search again".
- NS-932. The description now says which side a name belongs to: a bare name
is a hosted connector account; `mcp: true` only when the user asks for an MCP
server, a local server or an install, or when the name exists only in the
catalog; connect and reconnect are hosted verbs, install, enable and
authorize are MCP verbs. It names the three clients that draw a card.
- A misrouted target is refused with the call that works. Only when the
gateway does not know the connector (confirmed on that failure path) and the
name is a catalog entry does the target fail with "X is a local MCP server.
Call manage_connections with action install ...". It is a per-target
outcome: other targets of the same call keep their links and their card. A
vendor failure on a name both sides know stays an ordinary failed row. The
MCP side mirrors it, and never for an entry that is only not installed.
- "Do not re-ask after a skip or a timeout" no longer stops the model when the
USER asks for that app again; the description and the settled-result notes
say so. A builder saw the model refuse a direct user request.
Shown on the Ink TUI and the classic CLI with a real model: a dead gateway and
a 401; "connect fxmail" goes hosted; "install the fx-noauth MCP server" goes
MCP; "connect fx-noauth" reaches the MCP install card in one corrective round
with no hosted mint; a two-target call where one is misrouted still connects
the other with exactly one mint.
* fix(tui): the connection card answers every key, shows what is happening, and is dead once settled
Reproduced on the real Ink TUI with the rig, then fixed:
- The keyboard was dead during the sign-in wait: the card kept a `submitting`
flag that the normal OAuth path never cleared, and Esc went through the same
guard. The in-flight state now belongs to the answered row and clears when
that row moves, when any later frame of the operation arrives, or after
five seconds. Esc skips the row in every phase; Ctrl+C interrupts the turn
(the input handler had no branch for this overlay); Shift+arrows scroll the
transcript and the card ignores them; arrow keys no longer move the text
cursor and the field focus at once.
- The card was lost at turn idle: the overlay flag was cleared while the
operation stayed in the store, and a resume dropped the pending card. The
flag survives idle, a resume shows the pending card again, a session switch
clears it.
- States with no branch: `not_connected` and a row with no link fell into the
credential form; `expired` vanished with no note. The title and the row text
now name the action (connect, reconnect, install, enable, authorize); a
failed or expired row with no fields offers Try again / Skip; a failed row
WITH fields reopens the form over the typed draft, with the failure above it.
- A settled card is dead: at settle the overlay closes and one transcript line
per app states the outcome. A settled or dismissed operation id is
remembered, so no replay or resume can reopen its card. Esc in the last
"Finishing…" moment hides the card and still writes the outcome lines.
- A failed `connection.respond` and a browser that did not open are shown on
the card in one sentence.
Also: `tui_gateway/connector_payload.py` redacted the BOOLEAN `secret` flag of
a credential field to the string "[REDACTED]". On the desktop every credential
field therefore rendered as a password and lost its prefilled default. A
boolean is no longer redacted.
* chore(connectors): remove the comments and docstrings this branch added
Deletions only. Kept: tool directives (`# noqa`, `// eslint-disable`, ...),
`// SAFETY:` lines, and the docstrings of the contract models under
`tui_gateway/contracts/`, which become the descriptions in the generated
OpenRPC and TypeScript.
Checked that no code changed: every Python file has the same AST as before
once docstrings and `pass` are ignored (62 files), and every TypeScript file
prints the same with comments stripped by the TypeScript printer (32 files).
The generated contract files are unchanged.
* fix(connectors): a card restored after a reload answers again; every account RPC names auth and org failures
Found by the end-to-end runs on the pushed head.
- Desktop: after a window reload, Continue on the restored card sent nothing.
The answer looked up the backend that holds the session with the runtime
session id, the lookup wants the stored id, and a failed lookup returned
silently. When the lookup gives no owner the answer now goes out on the
window's active socket, which is what main does.
- `connectors.policy.get` answered `POLICY_UNAVAILABLE` for a rejected sign-in,
a refused scope, a non-member and a missing organisation alike: the handler
runs with the gateway's globals and did not import the reason enum, so its
own error mapping raised. `connectors.accounts.remove` caught auth failures
in its generic branch. `org_required` was mapped on `policy.set` only. All
six account RPCs now answer `NEEDS_NOUS_AUTH`, `FORBIDDEN_SCOPE`,
`ORG_ACCESS_DENIED` and `ORG_REQUIRED` for those four upstream answers.
621 lines
30 KiB
Python
621 lines
30 KiB
Python
"""Progressive tool disclosure ("tool search"): MCP/plugin tools and a curated set of
|
|
event-triggered core tools are replaced in the model-visible array by three bridge tools —
|
|
tool_search / tool_describe / tool_call. Invariants: core tools (``toolsets._HERMES_CORE_TOOLS``)
|
|
and session-gated GUI toolsets never defer unless named in ``defer``; ANY deferrable tool
|
|
activates the bridge (the listing scales with budget, not activation); the catalog is
|
|
stateless — rebuilt from the live tool-defs every assembly (a session-keyed one drifts and
|
|
silently drops tools); bridge calls route through ``model_tools.handle_function_call``."""
|
|
|
|
from __future__ import annotations
|
|
|
|
import functools
|
|
import json
|
|
import logging
|
|
import math
|
|
from collections import Counter
|
|
from dataclasses import dataclass
|
|
from typing import Any, Dict, Iterable, List, Optional, Tuple
|
|
|
|
from hermes_cli.config_defaults import DEFAULT_CONFIG
|
|
from tools.registry import tool_error
|
|
from tools.tool_search_catalog import (
|
|
BRIDGE_TOOL_NAMES, CHARS_PER_TOKEN, TOOL_CALL_NAME, TOOL_DESCRIBE_NAME, TOOL_SEARCH_NAME,
|
|
CatalogEntry, _fn, _listing_group_label, _registry_entry, _registry_toolset,
|
|
build_catalog, build_catalog_listing_with_form, search_catalog)
|
|
from tools.tool_search_validation import (
|
|
local_batch_error, normalize_tool_call_entries, not_deferrable_error, validate_deferred_call_args)
|
|
from tools.connectors import CONNECTOR_BATCH_SENTINEL, is_connector_name
|
|
from tools.connectors.search import (
|
|
connections_in_scope, connector_entries_by_group, connectors_unavailable, remote_schemas_for)
|
|
|
|
logger = logging.getLogger("tools.tool_search")
|
|
# Bound the work one bridge call requests. Search is capped at the gateway's
|
|
# own limit: the connector search route answers 7 use_cases per request and
|
|
# returns HTTP 502 for 8 or more (measured 2026-09-09), and one local call
|
|
# maps to one gateway request. Describe has no such remote limit.
|
|
_MAX_QUERIES_PER_CALL = 7
|
|
_MAX_DESCRIBE_NAMES_PER_CALL = 10
|
|
|
|
|
|
@dataclass(frozen=True)
|
|
class ToolSearchConfig:
|
|
"""Resolved, validated tool-search configuration for a single assembly."""
|
|
enabled: str # "auto" | "on" | "off" — "auto" is an alias of "on" today
|
|
# Listing budget as % of context; does NOT gate activation, only bounds how much
|
|
# the embedded manifest may consume before it degrades (full -> names -> bare).
|
|
threshold_pct: float # 0..100
|
|
search_default_limit: int
|
|
max_search_limit: int
|
|
listing: str = "auto" # "auto"/"on" = embed the manifest when it fits; "off" = bare bridge
|
|
listing_max_tokens: int = 4000 # budget = min(this, threshold_pct% of context)
|
|
# None = curated default; an explicit list replaces it wholesale ([] = defer no core tools).
|
|
defer_tools: Optional[frozenset] = None
|
|
|
|
@property
|
|
def effective_defer_tools(self) -> frozenset:
|
|
return _DEFAULT_DEFERRED_TOOLS if self.defer_tools is None else self.defer_tools
|
|
|
|
@classmethod
|
|
def from_raw(cls, raw: Any) -> "ToolSearchConfig":
|
|
"""Build from a raw dict / legacy bool / None; every field is clamped and unknown
|
|
values fall back to safe defaults — a config typo must not break the agent."""
|
|
if not isinstance(raw, dict): # legacy bool / None
|
|
raw = {"enabled": "off" if raw is False else "auto"}
|
|
max_search_limit = _clamped_int(raw.get("max_search_limit"), 25, 1, 50)
|
|
defer_raw = raw.get("defer")
|
|
if defer_raw is not None and not isinstance(defer_raw, (list, tuple, set)):
|
|
# Loud, then the curated default: a scalar here means the user tried to shrink the
|
|
# tool surface and got nothing — never silently ignore it (#116404).
|
|
logger.warning(
|
|
"tools.tool_search.defer is %r, expected a YAML list of tool names "
|
|
"(e.g. [todo_list, computer_use]; [] keeps every tool eager) - "
|
|
"using the curated default set.", defer_raw)
|
|
defer_raw = None
|
|
return cls(
|
|
enabled=_tri_state(raw.get("enabled", "auto")),
|
|
threshold_pct=max(0.0, min(100.0, _safe_float(raw.get("threshold_pct"), 5.0))),
|
|
search_default_limit=_clamped_int(
|
|
raw.get("search_default_limit"), 5, 1, max_search_limit),
|
|
max_search_limit=max_search_limit,
|
|
listing=_tri_state(raw.get("listing", "auto")),
|
|
listing_max_tokens=_clamped_int(raw.get("listing_max_tokens"), 4000, 200, 60000),
|
|
defer_tools=(frozenset(str(n).strip() for n in defer_raw if str(n).strip())
|
|
if isinstance(defer_raw, (list, tuple, set)) else None))
|
|
|
|
|
|
_TRI_STATE_ALIASES = {"true": "on", "1": "on", "yes": "on", "false": "off", "0": "off", "no": "off"}
|
|
|
|
|
|
def _tri_state(value: Any) -> str:
|
|
"""Normalize an ``auto``/``on``/``off`` setting (bool-ish aliases accepted)."""
|
|
text = str(value).strip().lower()
|
|
return _TRI_STATE_ALIASES.get(text, text if text in ("auto", "on", "off") else "auto")
|
|
|
|
|
|
def _clamped_int(value: Any, fallback: int, lo: int, hi: int) -> int:
|
|
"""``int(value)`` (or ``fallback`` when unparseable) clamped to ``[lo, hi]``."""
|
|
try:
|
|
value = int(value)
|
|
except (TypeError, ValueError):
|
|
value = fallback
|
|
return max(lo, min(hi, value))
|
|
|
|
|
|
def _safe_float(value: Any, fallback: float) -> float:
|
|
try:
|
|
return float(value)
|
|
except (TypeError, ValueError):
|
|
return fallback
|
|
|
|
|
|
def _config_from_loader(loader_name: str) -> ToolSearchConfig:
|
|
"""Tool-search config via ``hermes_cli.config.<loader_name>`` (defaults on any failure)."""
|
|
try:
|
|
import hermes_cli.config as _cfg_mod
|
|
tools_cfg = (getattr(_cfg_mod, loader_name)() or {}).get("tools")
|
|
tools_cfg = tools_cfg if isinstance(tools_cfg, dict) else {}
|
|
return ToolSearchConfig.from_raw(tools_cfg.get("tool_search"))
|
|
except Exception as e:
|
|
logger.debug("Failed to load tool-search config: %s", e)
|
|
return ToolSearchConfig.from_raw(None)
|
|
|
|
|
|
load_config = functools.partial(_config_from_loader, "load_config")
|
|
load_config_readonly = functools.partial(_config_from_loader, "load_config_readonly") # no copy
|
|
|
|
|
|
def _core_tool_names() -> frozenset[str]:
|
|
"""Names that never defer by default (lazy: ``toolsets`` imports ``tools.registry``)."""
|
|
try:
|
|
from toolsets import _HERMES_CORE_TOOLS
|
|
return frozenset(_HERMES_CORE_TOOLS)
|
|
except Exception:
|
|
return frozenset()
|
|
|
|
|
|
# Session-gated GUI toolsets: off ``_HERMES_CORE_TOOLS`` so non-GUI clients never pay
|
|
# their schema; once enabled they stay direct unless the deferral list names them.
|
|
_DIRECT_SURFACE_TOOLSETS = frozenset({"desktop_ui", "project"})
|
|
|
|
# Event-triggered tools deferred BY DEFAULT (a catalog stub suffices). Keep the curated
|
|
# list in DEFAULT_CONFIG so config discovery and runtime behavior cannot drift. An explicit
|
|
# ``defer`` list replaces this wholesale ([] = everything eager). ``clarify`` is deliberately
|
|
# absent: A/B showed deferring it collapsed structured-clarify usage (18/18 -> 7/18) — the
|
|
# ask-the-user affordance must be ambient, a stub is not enough.
|
|
_DEFAULT_DEFERRED_TOOLS = frozenset(DEFAULT_CONFIG["tools"]["tool_search"]["defer"])
|
|
|
|
|
|
def is_deferrable_tool_name(name: str, defer_tools: Optional[frozenset] = None) -> bool:
|
|
"""True if a tool is *eligible* for deferral: named in ``defer_tools`` (curated set or
|
|
user override), OR an MCP tool, OR neither core nor a session-gated GUI surface (i.e. a
|
|
plugin tool). Bridge names never defer."""
|
|
if name in BRIDGE_TOOL_NAMES:
|
|
return False
|
|
if defer_tools is not None and name in defer_tools:
|
|
return True
|
|
if name in _core_tool_names():
|
|
return False
|
|
toolset = _registry_toolset(name) # None (unregistered/malformed) never defers
|
|
return toolset is not None and (
|
|
toolset.startswith("mcp-") or toolset not in _DIRECT_SURFACE_TOOLSETS)
|
|
|
|
|
|
def _tool_def_names(tool_defs: Iterable[Dict[str, Any]]) -> Iterable[str]:
|
|
"""Function names of a tool-defs list (``""`` for a nameless def)."""
|
|
return (_fn(td).get("name", "") for td in tool_defs)
|
|
|
|
|
|
def classify_tools(tool_defs: List[Dict[str, Any]], defer_tools: Optional[frozenset] = None,
|
|
) -> Tuple[List[Dict[str, Any]], List[Dict[str, Any]]]:
|
|
"""Split a tool-defs list into (visible, deferrable); bridge tools are dropped (re-added
|
|
after classification)."""
|
|
visible: List[Dict[str, Any]] = []
|
|
deferrable: List[Dict[str, Any]] = []
|
|
for td, name in zip(tool_defs, _tool_def_names(tool_defs)):
|
|
if name not in BRIDGE_TOOL_NAMES:
|
|
(deferrable if is_deferrable_tool_name(name, defer_tools) else visible).append(td)
|
|
return visible, deferrable
|
|
|
|
|
|
def _deferrable_in(tool_defs: List[Dict[str, Any]]) -> List[Dict[str, Any]]:
|
|
"""Deferrable subset of pre-assembly ``tool_defs`` under the read-only user config."""
|
|
return classify_tools(tool_defs, load_config_readonly().effective_defer_tools)[1]
|
|
|
|
|
|
def estimate_tokens_from_schemas(tool_defs: Iterable[Dict[str, Any]]) -> int:
|
|
"""Token cost via the chars/4 rule (order-of-magnitude precision suffices)."""
|
|
def _chars(td: Dict[str, Any]) -> int:
|
|
try:
|
|
return len(json.dumps(td, ensure_ascii=False, separators=(",", ":")))
|
|
except (TypeError, ValueError):
|
|
return len(str(td))
|
|
return int(math.ceil(sum(map(_chars, tool_defs)) / CHARS_PER_TOKEN))
|
|
|
|
|
|
def should_activate(config: ToolSearchConfig, deferrable_tokens: int,
|
|
context_length: Optional[int], *, connections_granted: bool = False) -> bool:
|
|
"""``"off"`` never activates; ``"on"``/``"auto"`` activate whenever any deferrable tool
|
|
exists ("auto" is reserved for a future budget-gated mode — do not distinguish them
|
|
without that design). ``context_length`` is kept for caller compatibility."""
|
|
if config.enabled == "off":
|
|
return False
|
|
if deferrable_tokens > 0:
|
|
return True
|
|
return connections_granted
|
|
|
|
|
|
def listing_token_budget(config: ToolSearchConfig, context_length: Optional[int]) -> int:
|
|
"""``min(listing_max_tokens, threshold_pct% of context)``; unknown context uses a 10K
|
|
percentage leg (5% of a typical 200K window)."""
|
|
pct_leg = (int(context_length * (config.threshold_pct / 100.0))
|
|
if context_length and context_length > 0 else 10_000)
|
|
return max(0, min(config.listing_max_tokens, pct_leg))
|
|
|
|
|
|
def _bridge_schema(name: str, description: str, properties: Dict[str, Any],
|
|
required: List[str]) -> Dict[str, Any]:
|
|
"""One OpenAI-style function schema (key order is part of the frozen bytes)."""
|
|
return {"type": "function", "function": {
|
|
"name": name, "description": description,
|
|
"parameters": {"type": "object", "properties": properties, "required": required}}}
|
|
|
|
|
|
_CONNECTIONS_HINT = (
|
|
" Names starting with `connectors__` are tools of remote connector accounts "
|
|
"(Gmail, Linear, Notion, ...); `manage_connections` checks whether an account "
|
|
"is connected and gets the authorization link when it is not.")
|
|
|
|
|
|
def _search_description(deferred_count: int, listing: Optional[str], listing_form: str,
|
|
connections_granted: bool = False) -> str:
|
|
"""tool_search bridge description with the listing embedded (framing per ``listing_form``).
|
|
``connections_granted`` adds the one sentence that ties ``connectors__`` names to
|
|
``manage_connections``; without that tool in the session the sentence would name a tool
|
|
the model cannot call."""
|
|
desc = (
|
|
(f"Search {deferred_count} additional tools that are loaded on demand. "
|
|
if deferred_count else "Search remote connector tools (email, calendars, issue trackers, and more). ")
|
|
+ "Takes a list of queries searched in parallel against the same "
|
|
"catalog; send one query per distinct capability you need. Returns "
|
|
"matching tool names grouped per query plus a shared map with each "
|
|
"tool's description. Follow with "
|
|
f"`{TOOL_DESCRIBE_NAME}` to load full parameter schemas, "
|
|
f"then `{TOOL_CALL_NAME}` to invoke. Tools listed at the top of this "
|
|
"system prompt are already available and do not need to be searched."
|
|
+ (_CONNECTIONS_HINT if connections_granted else ""))
|
|
if not listing:
|
|
return desc
|
|
if listing_form == "groups":
|
|
return desc + (
|
|
"\n\nThe servers below are connected and their tools ARE available "
|
|
"through this bridge. For any request in these domains, search "
|
|
"here FIRST — do not claim the capability is unavailable and do "
|
|
"not substitute a generic tool (terminal/browser) without "
|
|
"searching.\n\n" + listing)
|
|
desc += (
|
|
"\n\nEvery deferred capability is listed below. If a tool name "
|
|
"appears here, do NOT claim it is unavailable — load it with "
|
|
f"`{TOOL_DESCRIBE_NAME}` (skip `{TOOL_SEARCH_NAME}` when you "
|
|
"already see the exact name).")
|
|
if listing_form == "mixed":
|
|
desc += (
|
|
" For servers marked 'names not listed', the tools exist "
|
|
f"too — find them with `{TOOL_SEARCH_NAME}` before "
|
|
"concluding anything is missing.")
|
|
return desc + "\n\n" + listing
|
|
|
|
|
|
def bridge_tool_schemas(deferred_count: int, listing: Optional[str] = None,
|
|
listing_form: str = "", connections_granted: bool = False) -> List[Dict[str, Any]]:
|
|
"""Bridge tool schemas injected in place of deferred tools; kept short — every byte is paid
|
|
every turn. ``listing`` is embedded in the tool_search description; per-tool forms say
|
|
"skip search when you see the exact name", "groups" says search is mandatory."""
|
|
return [
|
|
_bridge_schema(
|
|
TOOL_SEARCH_NAME,
|
|
_search_description(deferred_count, listing, listing_form, connections_granted),
|
|
{
|
|
"queries": {
|
|
"type": "array",
|
|
"items": {"type": "string"},
|
|
"description": "Search queries, each a few keywords describing one capability (e.g. ['create github issue', 'send slack message']). Searched in parallel; results come back grouped per query. A single string is accepted and treated as one query.",
|
|
},
|
|
"limit": {
|
|
"type": "integer",
|
|
"description": "Maximum number of matches per query. Defaults to 5 and is clamped to the configured maximum (25 by default).",
|
|
},
|
|
},
|
|
["queries"],
|
|
),
|
|
_bridge_schema(
|
|
TOOL_DESCRIBE_NAME,
|
|
f"Load the full JSON schemas for tools returned by `{TOOL_SEARCH_NAME}`. "
|
|
f"Required before `{TOOL_CALL_NAME}` if a tool's parameters are unknown. "
|
|
"Batch every schema you need into one call.",
|
|
{
|
|
"names": {
|
|
"type": "array",
|
|
"items": {"type": "string"},
|
|
"description": "Exact tool names (as returned by tool_search). A single string is accepted and treated as one name.",
|
|
},
|
|
},
|
|
["names"],
|
|
),
|
|
_bridge_schema(
|
|
TOOL_CALL_NAME,
|
|
"Invoke deferred tools. Takes `calls`, an array of {name, arguments} "
|
|
"— one entry per invocation; a single call is an array of one. "
|
|
"Local tools require one entry per tool_call. Only connectors__ names "
|
|
"may be batched together; mixed and multi-local batches are rejected. "
|
|
"Connector entries execute individually with results in input order. "
|
|
f"Argument shapes match each tool's schema (see `{TOOL_DESCRIBE_NAME}`). "
|
|
"Policy, hooks, and approvals run as for directly-listed tools.",
|
|
{
|
|
"calls": {
|
|
"type": "array",
|
|
"items": {
|
|
"type": "object",
|
|
"properties": {
|
|
"name": {"type": "string", "description": "Exact tool name to invoke."},
|
|
"arguments": {"type": "object", "description": "Arguments matching the tool schema."},
|
|
},
|
|
"required": ["name", "arguments"],
|
|
},
|
|
"description": "One local invocation, or one or more connector invocations. Never mix local and connector tools.",
|
|
},
|
|
},
|
|
["calls"],
|
|
),
|
|
]
|
|
|
|
|
|
@dataclass
|
|
class AssemblyResult:
|
|
"""Outcome of one assembly (tests and observability)."""
|
|
tool_defs: List[Dict[str, Any]]
|
|
activated: bool
|
|
deferred_count: int = 0
|
|
deferred_tokens: int = 0
|
|
threshold_tokens: int = 0
|
|
# 0 = passthrough; 1 = bridge + per-tool listing; 2 = bare bridge / server summary only.
|
|
tier: int = 0
|
|
listing_form: str = "none" # "full" | "names" | "mixed" | "groups" | "none"
|
|
|
|
|
|
def assemble_tool_defs(tool_defs: List[Dict[str, Any]], *, context_length: Optional[int] = None,
|
|
config: Optional[ToolSearchConfig] = None) -> AssemblyResult:
|
|
"""Tool-defs the model should see: passthrough when inactive, else deferrable tools
|
|
replaced by the three bridge tools. Idempotent — existing bridge tools are stripped first."""
|
|
config = config or load_config()
|
|
incoming = [td for td, name in zip(tool_defs, _tool_def_names(tool_defs))
|
|
if name not in BRIDGE_TOOL_NAMES]
|
|
visible, deferrable = classify_tools(incoming, config.effective_defer_tools)
|
|
connections_granted = connections_in_scope(incoming)
|
|
if not deferrable:
|
|
if should_activate(config, 0, context_length, connections_granted=connections_granted):
|
|
return AssemblyResult(tool_defs=incoming + bridge_tool_schemas(0, connections_granted=connections_granted),
|
|
activated=True, tier=2)
|
|
return AssemblyResult(tool_defs=incoming, activated=False)
|
|
deferrable_tokens = estimate_tokens_from_schemas(deferrable)
|
|
if not should_activate(config, deferrable_tokens, context_length):
|
|
return AssemblyResult(
|
|
tool_defs=incoming, activated=False, deferred_count=len(deferrable),
|
|
deferred_tokens=deferrable_tokens,
|
|
threshold_tokens=int((context_length or 0) * (config.threshold_pct / 100.0)), tier=0)
|
|
listing, listing_form = None, "none"
|
|
listing_budget = listing_token_budget(config, context_length)
|
|
if config.listing != "off":
|
|
listing, listing_form = build_catalog_listing_with_form(
|
|
deferrable, max_tokens=listing_budget)
|
|
bridge = bridge_tool_schemas(len(deferrable), listing=listing, listing_form=listing_form,
|
|
connections_granted=connections_granted)
|
|
tier = 1 if listing_form in ("full", "names", "mixed") else 2
|
|
logger.info(
|
|
"tool_search activated (tier %d): %d core/visible tools kept, %d deferred "
|
|
"(~%d tokens), listing %s (budget ~%d tokens)",
|
|
tier, len(visible), len(deferrable), deferrable_tokens, listing_form, listing_budget)
|
|
return AssemblyResult(
|
|
tool_defs=visible + bridge, activated=True, deferred_count=len(deferrable),
|
|
deferred_tokens=deferrable_tokens, threshold_tokens=listing_budget,
|
|
tier=tier, listing_form=listing_form)
|
|
|
|
|
|
def is_bridge_tool(name: str) -> bool:
|
|
return name in BRIDGE_TOOL_NAMES
|
|
|
|
|
|
def _clip_description(text: str, cap: int = 500) -> str:
|
|
"""Cap a record description, marking the cut so it reads as deliberate.
|
|
|
|
A bare slice ends mid-word ("apply exponential bac") and looks like
|
|
corruption; the ellipsis says "there is more — tool_describe has it".
|
|
500 keeps 9 in 10 vendor connector descriptions whole and every first
|
|
sentence (measured p90 575, first-sentence max 329 over 353 tools).
|
|
"""
|
|
return text if len(text) <= cap else text[:cap] + "…"
|
|
|
|
|
|
def _shared_tool_record(entry: CatalogEntry) -> Dict[str, Any]:
|
|
"""One record for the shared ``tools`` map (per-query groups carry names only);
|
|
``required`` lets the model attempt a trivial call without a ``tool_describe`` round-trip."""
|
|
try:
|
|
required = entry.schema["function"]["parameters"]["required"]
|
|
except (TypeError, KeyError, AttributeError):
|
|
required = []
|
|
return {"source": entry.source, "source_name": entry.source_name,
|
|
"description": _clip_description(entry.description or ""),
|
|
"required": [r[:64] for r in (required if isinstance(required, list) else [])
|
|
if isinstance(r, str)][:32]}
|
|
|
|
|
|
def _available_source_summary(catalog: List[CatalogEntry]) -> List[Dict[str, Any]]:
|
|
"""Deterministic ``[{name, tool_count}]`` of connected sources (attached to empty query
|
|
groups so a lexical miss is not read as a missing capability)."""
|
|
counts = Counter(_listing_group_label(entry.source_name) for entry in catalog)
|
|
return [{"name": name, "tool_count": counts[name]} for name in sorted(counts)]
|
|
|
|
|
|
def _string_list_arg(args: Dict[str, Any], key: str, *, dedupe: bool, max_items: int,
|
|
retry_hint: str) -> Tuple[Optional[List[str]], Optional[str]]:
|
|
"""Read a list-of-strings bridge argument -> ``(items, error_json)``. A bare string (a
|
|
common model slip) is a one-item list; rejects non-lists, all-blank lists, > ``max_items``."""
|
|
raw = args.get(key)
|
|
raw = [raw] if isinstance(raw, str) else raw
|
|
if not isinstance(raw, list):
|
|
return None, tool_error(f"{key} is required and must be an array of strings")
|
|
out: List[str] = []
|
|
for item in raw:
|
|
text = str(item or "").strip()
|
|
if text and (not dedupe or text not in out):
|
|
out.append(text)
|
|
if not out:
|
|
return None, tool_error(f"{key} is required and must contain at least one non-empty string")
|
|
if len(out) > max_items:
|
|
return None, tool_error(f"too many {key}: {len(out)} > max {max_items}. {retry_hint}")
|
|
return out, None
|
|
|
|
|
|
def dispatch_tool_search(args: Dict[str, Any], *, current_tool_defs: List[Dict[str, Any]],
|
|
config: Optional[ToolSearchConfig] = None,
|
|
connector_search: Optional[Any] = None) -> str:
|
|
config = config or load_config()
|
|
queries, err = _string_list_arg(args, "queries", dedupe=False, max_items=_MAX_QUERIES_PER_CALL,
|
|
retry_hint="Retry with fewer, more targeted queries.")
|
|
if err:
|
|
return err
|
|
raw_limit = args.get("limit")
|
|
limit = (config.search_default_limit if raw_limit is None
|
|
else _clamped_int(raw_limit, config.search_default_limit, 1, config.max_search_limit))
|
|
catalog = build_catalog(_deferrable_in(current_tool_defs))
|
|
remote_entries: List[List[CatalogEntry]] = [[] for _ in queries]
|
|
hosted_failure: Optional[str] = None
|
|
if connections_in_scope(current_tool_defs):
|
|
remote_entries, hosted_failure = connector_entries_by_group(
|
|
queries, connector_search=connector_search)
|
|
results: List[Dict[str, Any]] = []
|
|
tools_map: Dict[str, Dict[str, Any]] = {}
|
|
available_sources = _available_source_summary(catalog) if catalog else []
|
|
for position, query in enumerate(queries):
|
|
corpus = catalog + remote_entries[position]
|
|
hits = search_catalog(corpus, query, limit=limit)
|
|
for h in hits:
|
|
tools_map.setdefault(h.name, _shared_tool_record(h))
|
|
matches = [h.name for h in hits]
|
|
group: Dict[str, Any] = {"query": query, "matches": matches}
|
|
if not matches and catalog:
|
|
group["available_sources"] = available_sources
|
|
group["hint"] = (
|
|
"This query returned no lexical matches, but the sources above "
|
|
"are connected and their tools remain available. Retry "
|
|
"tool_search with the service name plus a concrete action or "
|
|
"object before concluding the capability is unavailable.")
|
|
results.append(group)
|
|
remote_count = sum(1 for name in tools_map if is_connector_name(name))
|
|
payload: Dict[str, Any] = {"queries": queries, "total_available": len(catalog) + remote_count,
|
|
"results": results, "tools": tools_map}
|
|
if hosted_failure:
|
|
payload["connectors"] = connectors_unavailable(hosted_failure, verb="searched")
|
|
return json.dumps(payload, ensure_ascii=False)
|
|
|
|
|
|
def dispatch_tool_describe(args: Dict[str, Any], *, current_tool_defs: List[Dict[str, Any]],
|
|
config: Optional[ToolSearchConfig] = None,
|
|
connector_describe: Optional[Any] = None) -> str:
|
|
config = config or load_config_readonly()
|
|
names, err = _string_list_arg(
|
|
args, "names", dedupe=True, max_items=_MAX_DESCRIBE_NAMES_PER_CALL,
|
|
retry_hint="Retry with fewer names per call.")
|
|
if err:
|
|
return err
|
|
deferrable = _deferrable_in(current_tool_defs)
|
|
by_name = {name: _fn(td) for td, name in zip(deferrable, _tool_def_names(deferrable)) if name}
|
|
remote_schemas, hosted_failure = remote_schemas_for(names, current_tool_defs, connector_describe)
|
|
|
|
tools: Dict[str, Dict[str, Any]] = {}
|
|
not_found: List[str] = []
|
|
undescribed: List[str] = []
|
|
errors: Dict[str, str] = {}
|
|
for name in names:
|
|
fn = by_name.get(name)
|
|
remote_fn = remote_schemas.get(name)
|
|
if fn is not None:
|
|
tools[name] = {"description": fn.get("description", ""),
|
|
"parameters": fn.get("parameters", {})}
|
|
elif isinstance(remote_fn, dict):
|
|
tools[name] = {"description": str(remote_fn.get("description", "")),
|
|
"parameters": remote_fn.get("parameters", {})}
|
|
elif is_connector_name(name):
|
|
(undescribed if hosted_failure else not_found).append(name)
|
|
elif _registry_entry(name) is not None and not is_deferrable_tool_name(
|
|
name, load_config_readonly().effective_defer_tools):
|
|
# Registered but bridge/core/GUI-surface: a real name, wrong door.
|
|
errors[name] = not_deferrable_error(name)
|
|
else:
|
|
not_found.append(name)
|
|
result: Dict[str, Any] = {"tools": tools}
|
|
if not_found:
|
|
result["not_found"] = not_found
|
|
result["hint"] = "Names in not_found are not currently available. Re-run tool_search to refresh."
|
|
if errors:
|
|
result["errors"] = errors
|
|
if hosted_failure:
|
|
result["connectors"] = connectors_unavailable(hosted_failure, verb="described", names=undescribed)
|
|
return json.dumps(result, ensure_ascii=False)
|
|
|
|
|
|
def scoped_deferrable_names(tool_defs: List[Dict[str, Any]]) -> frozenset[str]:
|
|
"""Deferrable names in the *pre-assembly* ``tool_defs`` of the session scope — the
|
|
universe ``tool_call`` may reach. Gates bridge dispatch AND the executor unwrap so a
|
|
restricted session cannot invoke an out-of-scope tool via the bridge."""
|
|
defer_tools = load_config_readonly().effective_defer_tools
|
|
return frozenset(n for n in _tool_def_names(tool_defs)
|
|
if n and is_deferrable_tool_name(n, defer_tools))
|
|
|
|
|
|
def resolve_underlying_call(args: Dict[str, Any]) -> Tuple[Optional[str], Dict[str, Any], Optional[str]]:
|
|
"""Parse a ``tool_call`` invocation into (underlying_name, args, error_msg).
|
|
|
|
Used by:
|
|
* the dispatcher in ``model_tools.handle_function_call``,
|
|
* the display layer (so the activity feed shows the underlying tool),
|
|
* the trajectory recorder.
|
|
|
|
A connector-only batch resolves
|
|
to ``(CONNECTOR_BATCH_SENTINEL, {"calls": [...]}, None)``: the batch is
|
|
one dispatch unit owned by the ``model_tools`` bridge branch, and the
|
|
sentinel is what planners/display layers see. A single local entry keeps
|
|
the historical single-tool contract unchanged.
|
|
|
|
On parse error, returns ``(None, {}, error_message)``.
|
|
"""
|
|
entries, err = normalize_tool_call_entries(args)
|
|
if err:
|
|
return None, {}, err
|
|
|
|
if len(entries) > 1 and any(not is_connector_name(e["name"]) for e in entries):
|
|
return None, {}, local_batch_error(entries)
|
|
if is_connector_name(entries[0]["name"]):
|
|
return CONNECTOR_BATCH_SENTINEL, {"calls": entries}, None
|
|
|
|
name = entries[0]["name"]
|
|
raw_args = entries[0]["arguments"]
|
|
if not is_deferrable_tool_name(name, load_config_readonly().effective_defer_tools):
|
|
return None, {}, not_deferrable_error(name)
|
|
return name, raw_args, None
|
|
|
|
|
|
__all__ = [
|
|
"TOOL_SEARCH_NAME", "TOOL_DESCRIBE_NAME", "TOOL_CALL_NAME", "BRIDGE_TOOL_NAMES",
|
|
"ToolSearchConfig", "CatalogEntry", "AssemblyResult", "load_config", "is_deferrable_tool_name",
|
|
"classify_tools", "estimate_tokens_from_schemas", "should_activate", "build_catalog",
|
|
"build_catalog_listing_with_form", "listing_token_budget", "search_catalog",
|
|
"bridge_tool_schemas", "assemble_tool_defs", "is_bridge_tool", "dispatch_tool_search",
|
|
"dispatch_tool_describe", "resolve_underlying_call", "scoped_deferrable_names",
|
|
"validate_deferred_call_args", "normalize_tool_call_entries",
|
|
"CONNECTOR_BATCH_SENTINEL", "is_connector_name"]
|
|
|
|
|
|
# ---- BEGIN PLUGIN-COMPAT (revert-scheduled; see COMPAT_MANIFEST.md) ----
|
|
# Names external plugins imported from this module before the Sep 2026 decomposition.
|
|
# Internal code MUST NOT use these (scripts/check_compat_pointers.py fails CI if it does).
|
|
# The whole block is removed by reverting the commit that added it.
|
|
from typing import Literal # noqa: F401,E402
|
|
import copy # noqa: F401,E402
|
|
from dataclasses import field # noqa: F401,E402
|
|
import re # noqa: F401,E402
|
|
import snowballstemmer # noqa: F401,E402
|
|
import threading # noqa: F401,E402
|
|
|
|
def build_catalog_listing(
|
|
deferrable: List[Dict[str, Any]],
|
|
*,
|
|
max_tokens: int = 4000,
|
|
) -> Optional[str]:
|
|
"""Render a skills-style manifest of the deferred catalog.
|
|
|
|
One line per tool — ``name: short description`` — grouped under a
|
|
heading per source (MCP server / plugin toolset), exactly like the
|
|
bundled-skills listing in the system prompt:
|
|
|
|
github tools: (44)
|
|
- create_issue: Open a new issue in a GitHub repository.
|
|
- merge_pull_request: Merge an open pull request.
|
|
...
|
|
|
|
Ordering is deterministic (groups and tools sorted by name) so the
|
|
rendered block is byte-stable across assemblies of the same catalog —
|
|
this keeps the request prefix cacheable across turns.
|
|
|
|
Token-budget fallbacks (cheap chars/4 estimate, same rule as the
|
|
activation gate):
|
|
1. full listing (names + short descriptions)
|
|
2. names-only listing, still grouped
|
|
3. server-level summary — one line per MCP server / plugin toolset
|
|
(name + tool count), so the model always knows WHICH domains are
|
|
reachable through the bridge even when per-tool names don't fit
|
|
4. ``None`` — only when the summary itself exceeds the budget
|
|
"""
|
|
text, _form = build_catalog_listing_with_form(deferrable, max_tokens=max_tokens)
|
|
return text
|
|
# ---- END PLUGIN-COMPAT ----
|