Self Debug¶

self_debug — mvp.self_debug

Cluster: Code Intelligence | Type: component | MCP Tools: 26

Overview¶

Self-debugging and introspection block that audits, benchmarks, evaluates, and reports on G6 component health. Provides four operations — audit (structure checks), benchmark (timing over configurable iterations), evaluate (correctness probes), and report (summary of all metrics) — with optional component-level filtering so a pipeline can self-inspect any subset of its stack.

When to use:

Continuously monitoring G6 component correctness and latency within a running pipeline
Generating a health report before a deployment gate to confirm all critical components pass self-checks
Benchmarking a specific component in isolation to detect performance regressions

Example:

from mvp.self_debug import SelfDebugBlock, SelfDebugInput

block = SelfDebugBlock(name="self_debug")
result = block.infer(SelfDebugInput(operation="report", component_filter="solver"))
# result.ok → True; result.value → SelfDebugOutput with passed, summary, metrics

Works well with: system_doctor, auto_engineer, solver

Public API¶

`SelfDebugInput(BaseModel)`¶

Input for the Tier 1 SelfDebugBlock.

Field	Type	Default
`operation`	`Literal['audit', 'benchmark', 'evaluate', 'report', 'info', 'list_patterns']`	'info'
`component_filter`	`str`	''
`iterations`	`int`	100

`SelfDebugOutput(BaseModel)`¶

Output from the Tier 1 SelfDebugBlock.

Field	Type	Default
`operation`	`str`	required
`passed`	`bool`	True
`summary`	`str`	''
`metrics`	`dict[str, float]`	Field(default_factory=dict)
`details`	`list[dict[str, Any]]`	Field(default_factory=list)
`count`	`int`	0
`degraded`	`bool`	False
`degradation_reason`	`str \\| None`	None
`completion_state`	`Literal['verified', 'qualified-draft', 'blocked-escalated']`	'qualified-draft'
`warning_card`	`dict[str, Any] \\| None`	None
`evidence`	`list[dict[str, Any]]`	Field(default_factory=list)
`request_id`	`str`	''
`task_id`	`str`	''
`run_id`	`str`	''

`SelfDebugBlock(AIBlock[SelfDebugInput, SelfDebugOutput, None])`¶

Tier 1 self-debug block. Delegates to SelfDebugMCPBlock.

Field	Type	Default
`name`	`str`	'self_debug'
`db_path`	`str`	field(default_factory=lambda: _DEFAULT_DB)
`resource_bounds`	`ResourceBounds \\| None`	None
`usage`	`ResourceUsage`	field(default_factory=ResourceUsage)

Methods:

`infer(data: SelfDebugInput) -> Result[SelfDebugOutput]`¶

`MCPSelfDebugInput(BaseModel)`¶

Input for the self-debug MCP block.

Field	Type	Default
`op`	`Literal['audit_wiring', 'audit_schemas', 'audit_imports', 'audit_stores', 'audit_registry', 'bench_registry', 'bench_invocation', 'bench_store', 'bench_routing', 'bench_pipeline', 'bench_memory', 'eval_routing_accuracy', 'eval_navigator_e2e', 'eval_grounding_quality', 'eval_coverage', 'report_summary', 'report_regressions', 'report_component', 'report_cluster', 'get_history', 'discover_components', 'inspect_component', 'configure_thresholds', 'run_suite', 'info', 'list_patterns']`	required
`component`	`str`	''
`cluster`	`str`	''
`iterations`	`int`	100
`run_id`	`str`	''
`metric`	`str`	''
`threshold_value`	`float`	0.0
`limit`	`int`	50
`suite`	`str`	''
`block_names`	`list[str]`	Field(default_factory=list)
`query`	`str`	''
`auto_heal`	`bool`	False

`MCPSelfDebugOutput(BaseModel)`¶

Output from the self-debug MCP block.

Field	Type	Default
`op`	`str`	required
`result`	`str`	''
`passed`	`bool`	True
`metrics`	`dict[str, float]`	Field(default_factory=dict)
`details`	`list[dict[str, Any]]`	Field(default_factory=list)
`message`	`str`	''
`count`	`int`	0
`metadata`	`dict[str, Any]`	Field(default_factory=dict)
`degraded`	`bool`	False
`degradation_reason`	`str \\| None`	None
`completion_state`	`Literal['verified', 'qualified-draft', 'blocked-escalated']`	'qualified-draft'
`warning_card`	`dict[str, Any] \\| None`	None
`evidence`	`list[dict[str, Any]]`	Field(default_factory=list)
`request_id`	`str`	''
`task_id`	`str`	''
`run_id`	`str`	''

`SelfDebugMCPBlock(AIBlock[MCPSelfDebugInput, MCPSelfDebugOutput, dict])`¶

25-op self-debug MCP block with SQLite persistence.

Field	Type	Default
`name`	`str`	'self_debug_mcp'
`state`	`dict \\| None`	None
`db_path`	`str`	field(default_factory=lambda: _DEFAULT_DB)
`resource_bounds`	`ResourceBounds \\| None`	None
`usage`	`ResourceUsage`	field(default_factory=ResourceUsage)

Methods:

`set_registry(registry: object) -> None`¶

Inject a registry for cross-component integration (e.g. auto_heal).

`infer(data: MCPSelfDebugInput) -> Result[MCPSelfDebugOutput]`¶

`SelfDebugStore`¶

Sync SQLite store with 3 tables: bench_results, thresholds, runs.

Constructor:

Parameter	Type	Default
`db_path`	`str`	':memory:'

Methods:

`create_run(suite: str, component_filter: str = '', git_sha: str = '') -> str`¶

`finish_run(run_id: str, total_pass: int, total_fail: int, summary: dict[str, Any] | None = None) -> None`¶

`get_run(run_id: str) -> dict[str, Any] | None`¶

`list_runs(suite: str = '', limit: int = 20) -> list[dict[str, Any]]`¶

`add_result(run_id: str, suite: str, op: str, metric: str, value: float, component: str = '', threshold: float | None = None, passed: bool = True) -> str`¶

`get_results(run_id: str = '', suite: str = '', op: str = '', component: str = '', metric: str = '', limit: int = 500) -> list[dict[str, Any]]`¶

`set_threshold(metric: str, value: float) -> None`¶

`get_threshold(metric: str) -> float | None`¶

`list_thresholds() -> dict[str, float]`¶

`add_healing_episode(episode_id: str, component: str, error_type: str, error_message: str = '', tier_used: int = 0, outcome: str = '', reward: float = 0.0, duration_seconds: float = 0.0, fix_applied: str = '', verified: bool = False) -> str`¶

`get_healing_episodes(component: str = '', outcome: str = '', limit: int = 100) -> list[dict[str, Any]]`¶

`count_all() -> dict[str, int]`¶

MCP Tools¶

Operation	Source
`audit_wiring`	self_debug_mcp
`audit_schemas`	self_debug_mcp
`audit_imports`	self_debug_mcp
`audit_stores`	self_debug_mcp
`audit_registry`	self_debug_mcp
`bench_registry`	self_debug_mcp
`bench_invocation`	self_debug_mcp
`bench_store`	self_debug_mcp
`bench_routing`	self_debug_mcp
`bench_pipeline`	self_debug_mcp
`bench_memory`	self_debug_mcp
`eval_routing_accuracy`	self_debug_mcp
`eval_navigator_e2e`	self_debug_mcp
`eval_grounding_quality`	self_debug_mcp
`eval_coverage`	self_debug_mcp
`report_summary`	self_debug_mcp
`report_regressions`	self_debug_mcp
`report_component`	self_debug_mcp
`report_cluster`	self_debug_mcp
`get_history`	self_debug_mcp
`discover_components`	self_debug_mcp
`inspect_component`	self_debug_mcp
`configure_thresholds`	self_debug_mcp
`run_suite`	self_debug_mcp
`info`	self_debug_mcp
`list_patterns`	self_debug_mcp

Self Debug¶

Overview¶

Public API¶

SelfDebugInput(BaseModel)¶

SelfDebugOutput(BaseModel)¶

SelfDebugBlock(AIBlock[SelfDebugInput, SelfDebugOutput, None])¶

infer(data: SelfDebugInput) -> Result[SelfDebugOutput]¶

MCPSelfDebugInput(BaseModel)¶

MCPSelfDebugOutput(BaseModel)¶

SelfDebugMCPBlock(AIBlock[MCPSelfDebugInput, MCPSelfDebugOutput, dict])¶

set_registry(registry: object) -> None¶

infer(data: MCPSelfDebugInput) -> Result[MCPSelfDebugOutput]¶

SelfDebugStore¶

create_run(suite: str, component_filter: str = '', git_sha: str = '') -> str¶

finish_run(run_id: str, total_pass: int, total_fail: int, summary: dict[str, Any] | None = None) -> None¶

get_run(run_id: str) -> dict[str, Any] | None¶

list_runs(suite: str = '', limit: int = 20) -> list[dict[str, Any]]¶

add_result(run_id: str, suite: str, op: str, metric: str, value: float, component: str = '', threshold: float | None = None, passed: bool = True) -> str¶

get_results(run_id: str = '', suite: str = '', op: str = '', component: str = '', metric: str = '', limit: int = 500) -> list[dict[str, Any]]¶

set_threshold(metric: str, value: float) -> None¶

get_threshold(metric: str) -> float | None¶

list_thresholds() -> dict[str, float]¶

add_healing_episode(episode_id: str, component: str, error_type: str, error_message: str = '', tier_used: int = 0, outcome: str = '', reward: float = 0.0, duration_seconds: float = 0.0, fix_applied: str = '', verified: bool = False) -> str¶

get_healing_episodes(component: str = '', outcome: str = '', limit: int = 100) -> list[dict[str, Any]]¶

count_all() -> dict[str, int]¶

MCP Tools¶

`SelfDebugInput(BaseModel)`¶

`SelfDebugOutput(BaseModel)`¶

`SelfDebugBlock(AIBlock[SelfDebugInput, SelfDebugOutput, None])`¶

`infer(data: SelfDebugInput) -> Result[SelfDebugOutput]`¶

`MCPSelfDebugInput(BaseModel)`¶

`MCPSelfDebugOutput(BaseModel)`¶

`SelfDebugMCPBlock(AIBlock[MCPSelfDebugInput, MCPSelfDebugOutput, dict])`¶

`set_registry(registry: object) -> None`¶

`infer(data: MCPSelfDebugInput) -> Result[MCPSelfDebugOutput]`¶

`SelfDebugStore`¶

`create_run(suite: str, component_filter: str = '', git_sha: str = '') -> str`¶

`finish_run(run_id: str, total_pass: int, total_fail: int, summary: dict[str, Any] | None = None) -> None`¶

`get_run(run_id: str) -> dict[str, Any] | None`¶

`list_runs(suite: str = '', limit: int = 20) -> list[dict[str, Any]]`¶

`add_result(run_id: str, suite: str, op: str, metric: str, value: float, component: str = '', threshold: float | None = None, passed: bool = True) -> str`¶

`get_results(run_id: str = '', suite: str = '', op: str = '', component: str = '', metric: str = '', limit: int = 500) -> list[dict[str, Any]]`¶

`set_threshold(metric: str, value: float) -> None`¶

`get_threshold(metric: str) -> float | None`¶

`list_thresholds() -> dict[str, float]`¶

`add_healing_episode(episode_id: str, component: str, error_type: str, error_message: str = '', tier_used: int = 0, outcome: str = '', reward: float = 0.0, duration_seconds: float = 0.0, fix_applied: str = '', verified: bool = False) -> str`¶

`get_healing_episodes(component: str = '', outcome: str = '', limit: int = 100) -> list[dict[str, Any]]`¶

`count_all() -> dict[str, int]`¶