Cog Arch Aixi¶
cog_arch_aixi — AIXItl cognitive architecture.
Cluster: Experience & Autonomy | Type: component | MCP Tools: 38
Overview¶
Research-grade, unreviewed, uncalibrated AIXItl cognitive architecture implementation combining Context-Tree Weighting (CTW) environment modelling with Monte Carlo Tree Search (MCTS) action selection. Delegates 36 operations to an inner AIXIMCPBlock backed by SQLite, providing session-scoped world model updates, action recommendations, and read-only capability/status introspection.
When to use:
- Running research-grade CTW/MCTS sequential-decision demonstrations
- Inspecting session-scoped world model updates, planner snapshots, and conservative calibration metadata
- Experimenting with reward-based learning loops while surfacing unseeded randomness, fallback, and persistence-boundary limitations
Example:
from mvp.cog_arch_aixi import CogArchAIXIBlock, AIXIInput
block = CogArchAIXIBlock(name="aixi", db_path=":memory:")
block.infer(AIXIInput(op="ctw_create", session_id="s1", ctw_name="seq1", ctw_depth=8))
result = block.infer(AIXIInput(
op="mcts_search", session_id="s1",
num_simulations=50, num_actions=2,
))
# Inspect result.value.completion_state, warning_card, and evidence before relying on it.
Works well with: experience_loop, cog_arch_soar, autonomy_governor
Public API¶
CogArchAIXIBlock(AIBlock[AIXIInput, AIXIOutput, dict])¶
AIXItl cognitive architecture block — delegates to AIXIMCPBlock.
| Field | Type | Default |
|---|---|---|
name | str | 'cog_arch_aixi' |
state | dict \| None | None |
db_path | str | ':memory:' |
resource_bounds | ResourceBounds \| None | None |
usage | ResourceUsage | field(default_factory=ResourceUsage) |
Methods:
infer(data: AIXIInput) -> Result[AIXIOutput]¶
AIXIInput(BaseModel)¶
| Field | Type | Default |
|---|---|---|
op | AIXIOp | required |
session_id | str | '' |
ctw_name | str | '' |
ctw_depth | int | Field(default=16, ge=0, le=64) |
bits_json | str | '[]' |
symbol | int | Field(default=0, ge=0, le=1) |
action | int | 0 |
observation | int | 0 |
reward | float | 0.0 |
num_simulations | int | Field(default=100, ge=1, le=10000) |
num_actions | int | Field(default=2, ge=1, le=64) |
obs_space_size | int | Field(default=2, ge=1, le=1024) |
horizon | int | Field(default=10, ge=1, le=100) |
exploration_rate | float | Field(default=0.1, ge=0.0, le=1.0) |
window_size | int | Field(default=10, ge=1, le=10000) |
env_name | str | '' |
env_type | str | 'coin_flip' |
env_config_json | str | '{}' |
history_json | str | '[]' |
description | str | '' |
source_json | str | '' |
target_json | str | '' |
gate_type | str | '' |
query | str | '' |
top_k | int | Field(default=5, ge=1, le=100) |
limit | int | Field(default=50, ge=1, le=1000) |
AIXIOutput(BaseModel)¶
| Field | Type | Default |
|---|---|---|
op | str | required |
ok | bool | required |
action | int | -1 |
message | str | '' |
error | str | '' |
data_json | str | '' |
records | list[dict] | Field(default_factory=list) |
count | int | 0 |
metadata | dict | Field(default_factory=dict) |
session_id | str | '' |
degraded | bool | False |
degradation_reason | str | '' |
completion_state | str | 'qualified-draft' |
warning_card | dict | Field(default_factory=dict) |
evidence | dict | Field(default_factory=dict) |
MCP Tools¶
| Operation | Source |
|---|---|
ctw_create | cog_arch_aixi_mcp |
ctw_update | cog_arch_aixi_mcp |
ctw_predict | cog_arch_aixi_mcp |
ctw_revert | cog_arch_aixi_mcp |
ctw_info | cog_arch_aixi_mcp |
agent_act | cog_arch_aixi_mcp |
agent_observe | cog_arch_aixi_mcp |
agent_history | cog_arch_aixi_mcp |
agent_reset | cog_arch_aixi_mcp |
mcts_search | cog_arch_aixi_mcp |
mcts_select | cog_arch_aixi_mcp |
mcts_expand | cog_arch_aixi_mcp |
mcts_rollout | cog_arch_aixi_mcp |
mcts_best | cog_arch_aixi_mcp |
reward_total | cog_arch_aixi_mcp |
reward_window | cog_arch_aixi_mcp |
horizon_set | cog_arch_aixi_mcp |
env_register | cog_arch_aixi_mcp |
env_step | cog_arch_aixi_mcp |
env_state | cog_arch_aixi_mcp |
llm_suggest_action | cog_arch_aixi_mcp |
llm_interpret_obs | cog_arch_aixi_mcp |
find_analogy | cog_arch_aixi_mcp |
hat_gate | cog_arch_aixi_mcp |
aixi_info | cog_arch_aixi_mcp |
list_patterns | cog_arch_aixi_mcp |
describe_capabilities | cog_arch_aixi_mcp |
describe_ctw_model | cog_arch_aixi_mcp |
describe_mcts_policy | cog_arch_aixi_mcp |
describe_environment_types | cog_arch_aixi_mcp |
get_backend_status | cog_arch_aixi_mcp |
get_session_model_status | cog_arch_aixi_mcp |
get_planner_snapshot | cog_arch_aixi_mcp |
get_calibration_status | cog_arch_aixi_mcp |
get_randomness_policy | cog_arch_aixi_mcp |
probe_llm_advisor_backends | cog_arch_aixi_mcp |
help | cog_arch_aixi_mcp |
ops | cog_arch_aixi_mcp |