Paper Contributions#

Source: examples/agents/paper_contributions.py

Introduction#

Agent studies need a compact writing-support boundary that records configuration without mistaking availability for execution. This offline example describes a tracer before any run evidence exists, so the resulting packet keeps the missing trace visible as a gap.

Technical Implementation#

  1. Configure a disabled Tracer without starting an agent run.

  2. Call collect_agent_paper_contributions(...) with a stable component identifier.

  3. Verify the public paper-contribution contract version and print the JSON-compatible packet.

        flowchart LR
    A["Configured tracer"] --> B["collect_agent_paper_contributions(...)"]
    B --> C["Configured Methods contribution"]
    B --> D["Execution evidence gap"]
    C --> E["Versioned component packet"]
    D --> E
    
 1from __future__ import annotations
 2
 3import json
 4
 5import design_research_agents as drag
 6
 7
 8def main() -> None:
 9    """Print deterministic writing support for one configured component."""
10    tracer = drag.Tracer(enabled=False, enable_jsonl=False, enable_console=False)
11    packet = drag.collect_agent_paper_contributions(
12        tracer,
13        component_id="example.tracer",
14    )
15    assert packet["schema_version"] == drag.PAPER_CONTRIBUTION_VERSION
16    print(json.dumps(packet, ensure_ascii=True, indent=2, sort_keys=True))
17
18
19if __name__ == "__main__":
20    main()

Expected Results#

Run Command

PYTHONPATH=src python examples/agents/paper_contributions.py

The printed packet contains a configured Methods contribution, package provenance, and an execution-not-provided gap. It contains no observed execution claim.

References#