This is a public, read-only report. Striff reads this repository's docs, turns each sentence that makes a claim about the code into a rule, and checks the rule against the code on the default branch. How this works
Added required FromAccountId and ToAccountId fields
2 rules from this sentence
CreateTransferRequest has a FromAccountId field
checks “Added required FromAccountId”
Holds
CreateTransferRequest has a ToAccountId field
checks “and ToAccountId fields”
Holds
Kept existing FromAccountNumber and ToAccountNumber fields for audit trail and response display
2 rules from this sentence
CreateTransferRequest has a FromAccountNumber field
checks “Kept existing FromAccountNumber”
Holds
CreateTransferRequest has a ToAccountNumber field
checks “and ToAccountNumber fields”
Holds
Transfer model already had FromAccountId/ToAccountId fields - no database schema change needed
2 rules from this sentence
Transfer has a FromAccountId field
Holds
Transfer has a ToAccountId field
Holds
batchableGroups() admits an item only through isBatchEligible() — L1, one required signer, and callerMaySign.
batchableGroups depends on isBatchEligible
Holds
Added a fan-out header when a step spawns two or more agents ("N agents in parallel — grouped under each, not interleaved") and an owner chip on any step-level tool call carrying a subagentId, so a fan-out's calls can never read as the root plan's own.
ToolCall has subagentId
Holds
Field Outcome, Value Completed — 515 lines net across 7 files TypeScript types (LastError, ApplicationResponse), shared ApplicationStages component (147 lines, eliminates 68% duplication), CustomerApplicationStatusPage (283 lines, polling + retry UX), AgentPipeline refactor (reduced 147→47 lines), routing integration, error handling
Judge rubric lives in code (_build_judge_instructions, _evaluator_description in src/ai-service/app/routes/api.py) — easy to version, review, and unit test (tests/test_llm_judge.py).
2 rules from this sentence
_build_judge_instructions is in src/ai-service/app/routes
Holds
_evaluator_description is in src/ai-service/app/routes
Holds
Implementation: run_foundry_evaluation in src/ai-service/app/routes/api.py
run_foundry_evaluation is in src/ai-service/app/routes