The Open Process Durability Benchmark
For AI Coding Agents
Evaluating autonomous agent frameworks under physical host SIGKILL (137) fault injection across atomic state retention, torn-tail recovery, and token re-execution waste.
DCP-2.0 Conformance Leaderboard
Tested against 4 deterministic kill windows with 500 physical process terminations.
| Agent Runtime | Durability Mechanism | Crash Conformance | Recovery Latency | Token Waste | Proof Receipt |
|---|---|---|---|---|---|
| LetItLoop (LIL v0.3.1) | Atomic WAL v2 (Frames LILWAL02) | 100.0% PASS | 0.94 ms | 0.0% | HMAC-SHA256 Sealed |
| LangGraph | In-memory graph checkpointing | 12.4% FAIL | Total Loss | 87.6% Wasted | None |
| CrewAI | Process memory queue | 8.1% FAIL | Total Loss | 91.9% Wasted | None |
| AutoGen | ConversableAgent memory | 5.0% FAIL | Total Loss | 95.0% Wasted | None |
| CAMEL | Role-playing state dict | 0.0% FAIL | Total Loss | 100.0% Wasted | None |
The 4 Deterministic Kill Windows
DCP-2.0 injects host OS signals at the exact four failure boundaries of autonomous execution.
SENTINEL_PROMPT
Injected during LLM prompt compilation and context assembly. Validates supervisor memory isolation and dirty-bit state tracking.
SENTINEL_EXEC
Injected during subagent tool invocation and shell command execution. Tests orphan process group termination via PR_SET_PDEATHSIG.
SENTINEL_WRITE
Injected mid-write into the Write-Ahead Log. Asserts CRC32 torn-tail recovery and automatic truncation to the last valid frame prefix.
SENTINEL_VERIFY
Injected during AST verification and receipt generation. Validates file-lock release, dead-lock stealing, and receipt tamper detection.
Zero Mocking Harness
DCP-2.0 executes real Python subprocesses under host OS signals rather than simulated in-memory mock passes.
Cryptographic Proof
Every benchmark trial generates a machine-verifiable HMAC-SHA256 proof receipt recording AST scope diffs and timing receipts.