File size: 21,990 Bytes
fbd5681 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 | """Abstract base class for pluggable context engines.
A context engine controls how conversation context is managed when
approaching the model's token limit. The built-in ContextCompressor
is the default implementation. Third-party engines (e.g. LCM) can
replace it via the plugin system or by being placed in the
``plugins/context_engine/<name>/`` directory.
Selection is config-driven: ``context.engine`` in config.yaml.
Default is ``"compressor"`` (the built-in). Only one engine is active.
The engine is responsible for:
- Deciding when compaction should fire
- Performing compaction (summarization, DAG construction, etc.)
- Optionally exposing tools the agent can call (e.g. lcm_grep)
- Tracking token usage from API responses
Lifecycle:
1. Engine is instantiated and registered (plugin register() or default)
2. on_session_start() called when a conversation begins
3. update_from_response() called after each API response with usage data
4. should_compress() checked after each turn
5. compress() called when should_compress() returns True
6. on_session_end() called at real session boundaries (CLI exit, /reset,
gateway session expiry) — NOT per-turn
"""
from abc import ABC, abstractmethod
from typing import Any, Dict, List, Optional
from agent.redact import redact_sensitive_text
MEMORY_CONTEXT_MAX_CHARS = 6_000
_MEMORY_CONTEXT_HEAD_CHARS = 4_000
_MEMORY_CONTEXT_TAIL_CHARS = 1_500
_MEMORY_CONTEXT_TRUNCATION_MARKER = "\n...[memory provider context truncated]...\n"
def sanitize_memory_context(memory_context: str) -> str:
"""Prepare provider context for a context-engine/LLM egress boundary."""
sanitized = redact_sensitive_text(
memory_context.strip(),
force=True,
redact_url_credentials=True,
)
if len(sanitized) <= MEMORY_CONTEXT_MAX_CHARS:
return sanitized
return (
sanitized[:_MEMORY_CONTEXT_HEAD_CHARS]
+ _MEMORY_CONTEXT_TRUNCATION_MARKER
+ sanitized[-_MEMORY_CONTEXT_TAIL_CHARS:]
)
def automatic_compaction_status_message(
engine: Any,
*,
phase: str,
default_message: str,
**context: Any,
) -> str | None:
"""Resolve host-visible status for an automatic compaction event.
Engines can suppress routine automatic status with
``emit_automatic_compaction_status = False`` or customize it by defining
``get_automatic_compaction_status_message(...)``. Empty strings and
``None`` mean "do not emit a lifecycle status".
"""
if not getattr(engine, "emit_automatic_compaction_status", True):
return None
formatter = getattr(engine, "get_automatic_compaction_status_message", None)
if callable(formatter):
message = formatter(
phase=phase,
default_message=default_message,
**context,
)
else:
message = default_message
if message is None:
return None
message = str(message).strip()
return message or None
class ContextEngine(ABC):
"""Base class all context engines must implement."""
# -- Identity ----------------------------------------------------------
@property
@abstractmethod
def name(self) -> str:
"""Short identifier (e.g. 'compressor', 'lcm')."""
# -- Token state (read by run_agent.py for display/logging) ------------
#
# Engines MUST maintain these. run_agent.py reads them directly.
last_prompt_tokens: int = 0
last_completion_tokens: int = 0
last_total_tokens: int = 0
threshold_tokens: int = 0
context_length: int = 0
compression_count: int = 0
# -- Compaction parameters (read by run_agent.py for preflight) --------
#
# These control the preflight compression check. Subclasses may
# override via __init__ or property; defaults are sensible for most
# engines.
#
# protect_first_n semantics (since PR #13754): count of non-system head
# messages always preserved verbatim, IN ADDITION to the system prompt
# which is always implicitly protected. Default 3 keeps the
# historical "system + first 3 non-system messages" head shape.
threshold_percent: float = 0.75
protect_first_n: int = 3
protect_last_n: int = 6
# User-visible lifecycle status for automatic host-triggered compaction.
# Alternative engines that treat compaction as routine background
# maintenance can set this false to keep successful automatic passes silent;
# warnings, errors, and explicit manual commands should still surface.
emit_automatic_compaction_status: bool = True
# -- Core interface ----------------------------------------------------
@abstractmethod
def update_from_response(self, usage: Dict[str, Any]) -> None:
"""Update tracked token usage from an API response.
Called after every LLM call with a normalized usage dict. The legacy
keys ``prompt_tokens``, ``completion_tokens``, and ``total_tokens``
are always present. Newer hosts also include canonical buckets:
``input_tokens``, ``output_tokens``, ``cache_read_tokens``,
``cache_write_tokens``, and ``reasoning_tokens``. Engines should
treat those fields as optional for compatibility with older hosts.
"""
@abstractmethod
def should_compress(self, prompt_tokens: int = None) -> bool:
"""Return True if compaction should fire this turn."""
def should_compress_info(self, prompt_tokens: int = None) -> "tuple[bool, str | None]":
"""Return ``(should_compress, reason)``.
The base implementation is backward-compatible: engines that only
implement ``should_compress`` get ``(should_compress(prompt_tokens),
None)``. Concrete engines with richer block reasons (e.g. a
summary-LLM cooldown or an anti-thrashing guard) override this to
surface a human-readable reason so callers can warn the user instead
of silently skipping compression. Added for the silent-overflow
warning fix (#62625) so plugin engines don't raise AttributeError.
"""
return self.should_compress(prompt_tokens), None
@abstractmethod
def compress(
self,
messages: List[Dict[str, Any]],
current_tokens: Optional[int] = None,
focus_topic: Optional[str] = None,
force: bool = False,
memory_context: str = "",
) -> List[Dict[str, Any]]:
"""Compact the message list and return the new message list.
This is the main entry point. The engine receives the full message
list and returns a (possibly shorter) list that fits within the
context budget. The implementation is free to summarize, build a
DAG, or do anything else — as long as the returned list is a valid
OpenAI-format message sequence.
Args:
focus_topic: Optional topic string from manual ``/compress <focus>``.
Engines that support guided compression should prioritise
preserving information related to this topic. Engines that
don't support it may simply ignore this argument.
force: Whether a user-requested compression should bypass an
engine-owned cooldown. Engines without cooldowns may ignore it.
memory_context: Text returned by memory providers immediately before
compaction. Summarizing engines should include non-empty text in
their handoff prompt. Older engines may omit this parameter; the
host filters unsupported optional arguments by signature.
"""
# -- Optional: proactive tool-result prune -----------------------------
def prune_tool_results_only(
self,
messages: List[Dict[str, Any]],
current_tokens: int | None = None,
) -> tuple[List[Dict[str, Any]], int]:
"""Deterministically trim old tool-result payloads without an LLM call.
Runs on a low, cost-oriented trigger independent of ``should_compress``
so large-window engines can reclaim re-sent tool output long before full
compaction would fire. Returns ``(messages, n_pruned)``.
Default is a safe no-op: the list is returned unchanged with ``0``
pruned. Engines that don't implement a cheap prune — and any engine that
predates this hook — inherit this default, so the agent loop's
post-tool-call prune path never raises ``AttributeError`` on them. The
built-in ContextCompressor overrides this with the real implementation.
"""
return messages, 0
# -- Optional: per-turn context selection (distinct from compression) --
def select_context(
self,
request_messages: List[Dict[str, Any]],
*,
conversation_messages: List[Dict[str, Any]] = None,
incoming_message: Dict[str, Any] = None,
budget_tokens: int = 0,
) -> List[Dict[str, Any]]:
"""Optionally choose/replace the context for THIS request, pre-generation.
Called every turn after the request message list is assembled and
before it is dispatched to the provider — independent of
``should_compress()``. This lets an engine *select* which context
enters the prompt (retrieval, topic routing, role/branch switching)
rather than *shrink* context that is already there. The two verbs are
orthogonal:
- ``compress()`` : context is too long -> make it shorter.
- ``select_context()``: this turn belongs to a different context
-> use that one instead.
Without this hook, engines that need per-turn access to the message
list have to force ``should_compress()`` to return ``True`` so that
``compress()`` is invoked every turn purely as a callback — which
conflates selection with compression and degrades behaviour when the
engine's backend is unavailable. ``select_context()`` removes the need
for that workaround.
The returned list is request-only: it replaces the messages sent to
the provider for this single call and MUST NOT be treated as persisted
transcript state. The conversation history in the session DB is left
untouched, so nothing leaks across turns. Return ``None`` to leave the
request unchanged.
Unlike the ``pre_llm_call`` plugin hook (which appends to the user
message and intentionally never rewrites the list, to preserve the
cache prefix), ``select_context()`` may *replace* the message list.
Ordering / cache contract: the host runs this hook **before** prompt
cache-control and **before** every request sanitizer (orphaned-tool
cleanup, thinking-only/role normalization, whitespace/JSON
normalization). So (a) whatever the hook returns still passes through
the same validation as any request — a malformed replacement cannot
reach the provider — and (b) prompt-cache stability (an AGENTS.md
invariant) is preserved: the default no-op leaves the request
byte-identical, so cache behaviour is unchanged for the built-in
compressor and any non-implementing engine. An engine that *does*
replace the list changes its own cache prefix by definition; that is
the engine's concern, and cache-control breakpoints are re-derived on
the selected list. The hook is evaluated per provider request (so it
re-runs on retries within a turn), consistent with "select the context
for THIS request".
Args:
request_messages: The assembled request message list (system
prompt + history + any ephemeral prefill), in OpenAI format.
conversation_messages: The unmodified persisted conversation
history, for reference only (do not mutate).
incoming_message: The current turn's user message, if available.
budget_tokens: The active model's context length, or 0 if unknown.
Default returns ``None`` (no-op) — zero impact on the built-in
compressor or any existing engine.
"""
return None
def on_turn_complete(
self,
messages: List[Dict[str, Any]],
usage: Dict[str, Any] = None,
**kwargs: Any,
) -> None:
"""Observe a finished user turn (post-turn ingestion / observation).
Called from the standard turn-finalization path once the assistant/tool
loop completes, with the finalized in-memory transcript snapshot. This
is the complement to ``select_context()``: selection happens *before*
the request, while observation happens *after* the turn. It lets an
engine ingest, index, summarize, or update routing / topic / session
state from what actually happened — so the next ``select_context()``
can act on it.
Coverage: this fires from the normal finalization seam. Some abnormal
early-return paths in the loop (e.g. a content-policy block or a
provider terminal failure) persist and return without routing through
finalization, and therefore do not currently emit this hook. Treat it
as a best-effort post-turn observation for completed turns, not a
guaranteed callback for every possible early exit; unifying all
terminal paths behind one finalization seam is a separate follow-up.
Together the two hooks remove the need to abuse ``should_compress()`` /
``compress()`` as a generic per-turn callback just to observe history,
and they cover the case where a turn finishes and there may be no next
request from which to infer the previous turn.
``messages`` is a shallow copy and should be treated as read-only:
return values are ignored and this hook must not rely on transcript
mutation for persistence. ``kwargs`` may include ``turn_id``,
``task_id``, ``api_call_count``, ``interrupted``, ``failed``, and
``turn_exit_reason``.
``usage`` carries the completed turn's canonical token usage (the same
dict shape passed to ``update_from_response`` — ``prompt_tokens`` /
``completion_tokens`` / ``total_tokens`` plus the canonical
``input_tokens`` / ``output_tokens`` / ``cache_read_tokens`` /
``cache_write_tokens`` / ``reasoning_tokens`` buckets) so an engine can
weigh how large/expensive the selected context actually was when
deciding the next ``select_context()``. It is ``None`` on finalized
turns that never reached a provider response (e.g. interrupt); engines
must treat it as optional.
Default is a no-op.
"""
return None
# -- Optional: pre-flight check ----------------------------------------
def should_compress_preflight(self, messages: List[Dict[str, Any]]) -> bool:
"""Quick rough check before the API call (no real token count yet).
Default returns False (skip pre-flight). Override if your engine
can do a cheap estimate.
"""
return False
def should_defer_preflight_to_real_usage(self, rough_tokens: int) -> bool:
"""Return True when preflight should trust recent real usage instead.
Built-in compression uses this to avoid re-compacting from known-noisy
rough estimates after a compressed request has already fit. Third-party
engines can ignore it safely.
"""
return False
def get_automatic_compaction_status_message(
self,
*,
phase: str,
default_message: str,
**context: Any,
) -> str | None:
"""Return user-visible status for automatic host-triggered compaction.
Return ``None`` to suppress successful automatic lifecycle status for
this compaction event. ``phase`` identifies the host call site (for
example ``"preflight"`` or ``"compress"``). ``context`` contains
best-effort fields such as ``approx_tokens`` and ``threshold_tokens``.
This hook does not control warning/error messages or explicit manual
commands such as ``/compress``.
"""
if not self.emit_automatic_compaction_status:
return None
return default_message
# -- Optional: manual /compress preflight ------------------------------
def has_content_to_compress(self, messages: List[Dict[str, Any]]) -> bool:
"""Quick check: is there anything in ``messages`` that can be compacted?
Used by the gateway ``/compress`` command as a preflight guard —
returning False lets the gateway report "nothing to compress yet"
without making an LLM call.
Default returns True (always attempt). Engines with a cheap way
to introspect their own head/tail boundaries should override this
to return False when the transcript is still entirely protected.
"""
return True
# -- Optional: session lifecycle ---------------------------------------
def on_session_start(self, session_id: str, **kwargs) -> None:
"""Called when a new conversation session begins.
Use this to load persisted state (DAG, store) for the session.
kwargs may include hermes_home, platform, model, etc.
"""
def on_session_end(self, session_id: str, messages: List[Dict[str, Any]]) -> None:
"""Called at real session boundaries (CLI exit, /reset, gateway expiry).
Use this to flush state, close DB connections, etc.
NOT called per-turn — only when the session truly ends.
"""
def on_session_reset(self) -> None:
"""Called on /new or /reset. Reset per-session state.
Default resets compression_count and token tracking.
"""
self.last_prompt_tokens = 0
self.last_completion_tokens = 0
self.last_total_tokens = 0
self.compression_count = 0
# -- Optional: tools ---------------------------------------------------
def get_tool_schemas(self) -> List[Dict[str, Any]]:
"""Return tool schemas this engine provides to the agent.
Default returns empty list (no tools). LCM would return schemas
for lcm_grep, lcm_describe, lcm_expand here.
"""
return []
def handle_tool_call(self, name: str, args: Dict[str, Any], **kwargs) -> str:
"""Handle a tool call from the agent.
Only called for tool names returned by get_tool_schemas().
Must return a JSON string.
kwargs may include:
messages: the current in-memory message list (for live ingestion)
"""
import json
return json.dumps({"error": f"Unknown context engine tool: {name}"})
# -- Optional: status / display ----------------------------------------
def get_status(self) -> Dict[str, Any]:
"""Return status dict for display/logging.
Default returns the standard fields run_agent.py expects.
"""
# Clamp the -1 "compression just ran, awaiting real usage" sentinel
# (set by conversation_compression) to 0 so status readers don't see a
# raw -1 or a negative usage_percent on the transitional turn. Mirrors
# the CLI/gateway status-bar paths (cli.py, tui_gateway/server.py).
last_prompt = self.last_prompt_tokens if self.last_prompt_tokens > 0 else 0
return {
"last_prompt_tokens": last_prompt,
"threshold_tokens": self.threshold_tokens,
"context_length": self.context_length,
"usage_percent": (
min(100, last_prompt / self.context_length * 100)
if self.context_length else 0
),
"compression_count": self.compression_count,
}
# -- Optional: model switch support ------------------------------------
def update_model(
self,
model: str,
context_length: int,
base_url: str = "",
api_key: str = "",
provider: str = "",
api_mode: str = "",
) -> None:
"""Called when the user switches models or on fallback activation.
Default updates context_length and recalculates threshold_tokens
from threshold_percent. Override if your engine needs more
(e.g. recalculate DAG budgets, switch summary models).
"""
self.context_length = context_length
# Apply per-model threshold overrides if set (longest substring match).
# Falls back to _config_threshold_percent (the raw config value) when
# no override matches. Plugin engines that override update_model() can
# call resolve_model_threshold() for the same logic.
from agent.context_compressor import resolve_model_threshold
if not hasattr(self, "_config_threshold_percent"):
# Snapshot the pre-override percent ONCE so repeated model
# switches fall back to the engine's configured value, not the
# previous model's override.
self._config_threshold_percent = self.threshold_percent
self._base_threshold_percent = resolve_model_threshold(
model, getattr(self, "model_thresholds", {}),
self._config_threshold_percent,
)
self.threshold_percent = self._base_threshold_percent
self.threshold_tokens = int(context_length * self.threshold_percent)
|