Logging for LLM Agents

Let agents keep minimal context overhead by making logs summary-first markdowns for them to easily grab and ingest.

Summary-First

Every log starts with a concise markdown summary so agents can decide whether to read deeper without burning context.

Structured as short markdowns that agents grab and ingest in a single pass — no parsing, no filtering.

Each log is tied to a unique execution hash, so agents can trace from a summary back to the raw output and related runs.

app/refills/service.py

result = run_refill_check(payload)
raysurfer.log(result, "4b1f7f31a2c9d8e6")

agent_terminal.sh

curl -H "X-Key:$RAYSURFER_API_KEY" \
  api.raysurfer.com/logs/4b1f..a2c9.md
# summary markdown by default
# includes links to raw + related hashes

One line to log, one curl to retrieve. Summary-first by default.