[eersnington]

Closes #370. Tracing is built in, so pi() has no telemetry option.

agent/prompt                                 the caller's trace
└── invoke_agent pi

pi.durable.submit                            the run's trace, linked to agent/prompt
├── pi.durable.phase  pi.generation prepare
├── pi.durable.phase  pi.generation request
│   └── pi.durable.phase  pi.tool call
│       └── app.lookup                       an app span
└── pi.durable.phase  pi.generation tools

• packages/pi/src/tracing.ts implements Pi's TelemetryContext over @opentelemetry/api. Harness.open gets it as telemetry.
• Each run is its own trace, linked to its caller. Root spans use ROOT_CONTEXT, so a background phase cannot nest under an unrelated action.
• A run keeps its trace ID across retries and actor sleeps, because Pi stores traceparent with its tasks. With no SDK, nothing is stored.
• actorSqlite runs Pi's storage through c.run.runOutsideWorkflowSpan. The app's own c.db calls still export rivet.sqlite.* while an action is open.
• Every OpenTelemetry call is guarded, so a broken tracer cannot break a run.
• Docs: exporting Pi's spans needs the NodeSDK step. The environment variable alone exports only RivetKit's spans.
• Security: a failed span records error.type and drops the message, because Pi's diagnostics can contain caller content.

On upgrade

Apps that already run NodeSDK start exporting Pi's spans, about 10 per prompt. Pi's storage no longer exports rivet.sqlite.* spans. There is no switch for Pi's spans alone.

---

This is part 1 of 2 in a stack:

•  2  feat(pi): trace model requests #386
•  1  feat(pi): trace Pi Durable with OpenTelemetry #385 👈

[railway-app]

This PR was not deployed automatically as @eersnington does not have access to the Railway project.

In order to get automatic PR deploys, please add @eersnington to your workspace on Railway.

[github-actions]

📟 Pip-Boy · a718407

📦 Packages · fresh npm install resolve, no lockfile · this PR vs base

No notable install size, dependency, or entrypoint changes across 10 packages.

All 10 packages · linux-x64-gnu

package (linux-x64-gnu)	@latest	base	this PR	Δ vs base	packages
@rivet-dev/pi	372.8 MB	381.9 MB	381.9 MB	+17.7 KB (+0.0%)	220 (0)
@rivet-dev/sandbox-adapter	113.8 KB	114.5 KB	114.5 KB	0	1 (0)
@sandbox-agent/cli-shared	27.2 KB	27.2 KB	27.2 KB	0	1 (0)
@sandbox-agent/persist-indexeddb	12.4 KB	12.4 KB	12.4 KB	0	1 (0)
@sandbox-agent/persist-postgres	12.3 KB	12.3 KB	12.3 KB	0	1 (0)
@sandbox-agent/persist-rivet	12.2 KB	12.2 KB	12.2 KB	0	1 (0)
@sandbox-agent/persist-sqlite	12.3 KB	12.3 KB	12.3 KB	0	1 (0)
@sandbox-agent/react	41.6 MB	41.9 MB	41.9 MB	0	15 (0)
acp-http-client	6.7 MB	6.7 MB	6.7 MB	0	3 (0)
sandbox-agent	31.0 MB	31.0 MB	31.0 MB	0	8 (0)

🧹 Rust dependencies

Skipped: this PR doesn't change any Rust code or dependencies.

🦀 Binaries

Skipped: this PR doesn't change any Rust code.

⏱ Timing

job	wall time	slowest step
changes	21s	checkout · 16s
zizmor	16s	checkout · 11s
footprint	3m30s	Run pnpm --dir sandbox-agent install --frozen-lockfile · 1m39s
⏭ native · ${{ matrix.platform }}	—	skipped
⏭ hygiene	—	skipped
⏭ self-test	—	skipped
report	33s	setup-node · 11s

🔭 Full CI trace for this commit

One OpenTelemetry trace of CI and Pip-Boy for a718407: every job and step, with the sizes above attached. Run after CI finishes.

pnpm install --frozen-lockfile --filter 'pipboy...'
pnpm --filter pipboy pipboy trace --sha a71840742a917db24549a68cd79a24a8ca0062c4

Add --otlp <url> to send it to any OTLP/HTTP collector.

Agent instruction: From a rivet-dev/agents checkout, run the commands above, then read the trace file the last command prints. List the slowest jobs and steps and explain where CI time went.