deepline-plays

Deepline Plays

Quick Start

npm install -g deepline
# Fallback for secure sandboxes: mkdir -p "$HOME/.local" && npm config set prefix "$HOME/.local" && export PATH="$HOME/.local/bin:$PATH" && npm install -g deepline --registry https://code.deepline.com/api/v2/npm/
deepline auth register --wait auto
deepline auth wait --timeout 120 # completes Cowork/browser approval; no-op if already connected

CLI resolution

Run deepline when it is available. If the shell reports that command is missing, use <workspace-root>/.deepline/runtime/bin/deepline (or the npm-created .cmd shim on Windows). If neither exists, follow https://code.deepline.com/INSTALL.md to set up Deepline.

Debug every Play run: use deepline plays run <file.play.ts> --input '<json>' --debug while iterating, or deepline runs logs <run-id> --debug for an existing run. completed means the runtime finished, not that an external destination accepted the result; the debug logs retain caught non-2xx ctx.fetch responses with their key, method, destination origin, and HTTP status.

Before the first Deepline fanout in a task, run deepline preflight --json as one standalone command and wait for it to finish. Never submit preflight beside another Deepline command. After it succeeds, prefix every Deepline command that may run concurrently with DEEPLINE_SKIP_SELF_UPDATE=1; serial commands may stay bare.

Ordinary TypeScript, no DSL. A SearchProgram is one function that calls a tool, a fetch, a child Play, a connector, or a local artifact and returns a typed attempt. runSearchExperiment owns the pilot, ranked waterfall, holdout, gap-only retries, and cost/coverage report.

Deliverable

• Marginal, never amortized. Total ÷ successes reported 1.51 credits/email for a route whose real marginal cost was 0.21.
• Pass the printed block through. Do not recompute credits in prose.
• A catalog ceiling stops the run; it is not spend. Label it. A 120-credit ceiling truncated recovery at ~12 credits actual, and two apparent logic regressions were budget artifacts.

Read one job page

Read the row that matches this job, and only that row. Each page is complete for its job: source geometry, route ladder, pilot sizing, stop conditions.

Two lookups, consulted on a trigger rather than read up front: shared/authoring.md for Play syntax outside the scaffold, and references/debugging.md for a failed, empty, or misshapen run.

For every public definePlay/ctx.* type, binding, durable-cache rule, and runtime error, read references/sdk-reference.md. It is generated from the authoring contract and SDK source; do not duplicate that surface in this skill.

If your configuration forbids subagents, say so before starting serial work. Resolving that conflict silently cost one run ~30 minutes.

Topology

Write unit + decision + required facts + scale before touching tools. Requested fields stay required; demoting one to promote a run is not a pass. A null needs an absence receipt: materially different routes tried, typed outcomes retained.

One shape. Known rows: one experiment over the supplied rows. Open-world discovery: rows are query/page/geography/registry partitions, never remembered companies. Company → person: two sequential stages, not consensus; only companyExperiment.finalResults become contact rows. End-to-end: compare only when every program produces the same complete final row from the same seam.

Catalog

tools describe is the authoring contract and can disagree with runtime: a declared getter has been absent, and a tool documenting one scalar has returned a full list. Bind a named declared playExpression and sentinel-probe one row before scaling. toolResponse.raw is for an exact source excerpt, debugging, or an undeclared field after that probe — never a cast into an invented Company[].

Cover source classes before provider names — index, SERP, primary document, registry, event feed, first-party data, aggregator, validator. Two vendors reaching the same terminal corpus are one evidence lineage.

Record each route's pricing basis: per call, per returned result, or unknown. A confirmed-uncharged miss justifies a broader challenge wave, not a narrower one.

Build and run

Read its printed next list: it carries the four seams, tools: [...], coherenceChecks, and the company→person handoff at the point you edit them. --input-csv also writes a stratified fixture.csv. Iterate route code against that; use the full cohort only for a scored run.

Keep the top-level definePlay description short and concrete. The UI shows it below the Play identifier. Catalog categories are derived from the registered tools used by the Play; do not author category metadata on the Play itself.

run-and-export does the structural check, Play check, completed Play, run-bound export of both the results dataset and the route scorecard, then the COST RECEIPT. Its {ok: true, runId, output} is the completion receipt: before it the work is a probe, and a CSV written from remembered values hides which route won.

Receipt labels:

• CUT CANDIDATE — spent credits, completed nothing. Cut it. One route at 3.95 credits/call, 200× a search, ran ten rounds for zero results because the scorecard reported no cost at all.
• NEVER REACHED — never invoked, so its zero results are not a ceiling and not a source miss. maxFallbacks bounds the dependency-closed waterfall and defaults to 2; raise it (up to 4, scaled to pool size) or drop the route.
• cached calls — reruns of the same inputs reuse tool receipts. Quote the marginal rate, not this run's total.

Quality gates precede economics; among valid results prefer fewer observed credits, then fewer calls. Never expose provider spend.

Reusing a route across jobs is an eval, not a score: freeze the contract, verifier, cases and ceiling, and stratify the case set (normal, sparse, likely-miss, collision-prone) rather than picking easy rows after seeing results. A concept is an information geometry, never a vendor.

Product friction feedback

Send one proactive deepline feedback send report per issue cluster when a Play is disproportionately slow, appears stuck through repeated status checks, or an ordinary result requires avoidable product steps. Continue the user's task after reporting; feedback is not the deliverable.

Use judgment rather than a fixed threshold. Do not report normal provider work just because it is not instant. Report when the duration is surprising for the input size or prior comparable runs, progress stays uninformative, or the path needs avoidable discovery, conversion, ID extraction, retry, export, or manual workaround steps.

Include the goal and Play/run ID. For latency, include input size, observed duration, status transitions or polling count, and expected progress. For excess steps, include the actual path, avoidable detours, and expected direct path. Do not pass --requested for a proactive report. Omit --error-outcome when no error occurred; if an error did occur, use terminal only when it stopped the task and continued when work continued.

Subagents

One or two, only when several source geometries are plausible: same contract, one source lane each, returning a strategy card and ordinary TypeScript. The parent binds, runs, and judges. Verification fans out the same way — four defects found in four sequential rounds of eyeballing output fit in one pass over row batches.