Agent Receipts engine
Deterministic packet compiler for AI agent runs.
Takes raw subagent output (evidence JSONL, verifier findings, raw artifacts) and compiles a schema-validated, hash-provenanced next_pass_packet.json that the orchestrating model reads instead of raw subagent prose.
Part of the agent-receipts project — see the repo for the full runtime that drives this crate.
What this crate does
- Reads a run directory containing
manifest.json,worker-results/evidence.jsonl,verifier-results/findings.jsonl, andraw/artifacts. - Validates every
source_ref— hashesfile:references against disk, checks line-range spans, enforces provenance rules for substantive evidence kinds. - Emits schema
2.0.0typed trust: integrity, command outcome, applicability, and claim status are independent engine-owned fields. - Promotes a claim only through a current
.receipts/checks.tomlbinding over exact subject bytes, dependency locks, environment, check version, and target claim. Agent confidence is retained only asreported_confidencediagnostics. - Auto-detects
Contradictionentries when different agents assert divergent summaries on the same direct source span, graduating severity by evidence kind. - Emits
next_pass_packet.json,snapshot.json,decision_log.jsonl— byte-deterministic for byte-identical inputs.
Install
Install the standalone Rust CLI from crates.io:
Or install the full npm dispatcher plus bundled engine source from the public
repository with npm install -g github:inchwormz/agent-receipts.
Run
Expected run directory shape:
manifest.jsontask.mdraw/worker-results/evidence.jsonlverifier-results/findings.jsonl
Outputs land in state/ inside the run directory.
Minimal evidence shape
Determinism guarantee
The compile_determinism integration test runs the compiler twice on a byte-identical fixture and asserts byte-identical next_pass_packet.json + snapshot.json. A regression here means a non-deterministic code path slipped in.
License
MIT