traffic-webmcp.netlify.app

Last change

None yetUnchanged since first read.

Declared read-only

3 of 8

llms.txt

None

1. apply_configCan change things

   Commit a signal timing to the junction the operator is watching. This is the only tool that changes what is on screen. Applying a timing RESTARTS the live run: changing timing mid-cycle would leave the controller in an inconsistent phase and contaminate the measured figures with the previous timing. Expect roughly four minutes of simulated time before live metrics settle. Refuses, with the speci

   First seen 12 Oct 2026. Unchanged since.

2. describe_modelReads only

   Return the full specification of this intersection: tunable parameters with units and ranges, metric definitions, the hard constraints that govern what may be deployed, and the available demand scenarios. Call this first — it is sufficient to understand the problem without any prior traffic-engineering knowledge.

   First seen 12 Oct 2026. Unchanged since.

3. pin_resultCan change things

   Add a timing to the operator's comparison tray with a short rationale. Use this to surface the two or three candidates worth a human decision — for example the fastest admissible timing alongside one that trades a little mean wait for a much shorter worst case. Pinning does not commit anything. The operator applies a pinned timing themselves, or asks you to. Pin the trade-offs, not every candidat

   First seen 12 Oct 2026. Unchanged since.

4. play_sequenceCan change things

   Hand the operator a scrubbable, narrated walkthrough of your reasoning. Each step puts a timing — or a pair of timings — on the cycle diagram with a one-line caption. The operator can pause, scrub and replay it. Two kinds of step: - focus="structure" (default) draws the cycle diagram: where the phases sit, what a protected phase costs in cycle time. - focus="metrics" compares OUTCOMES side by

   First seen 12 Oct 2026. Unchanged since.

5. read_stateReads only

   Return what is currently on screen: the committed signal timing, the active demand scenario, metrics measured so far on the live run, the controller's current phase, per-arm queues, and which parameters the operator has locked. Locked parameters cannot be changed by apply_config — read this before proposing a timing.

   First seen 12 Oct 2026. Unchanged since.

6. run_batchReads only

   Evaluate up to 50 candidate signal timings headlessly and return the resulting metrics for each. Runs off the main thread; the live junction keeps running while this executes. Deterministic: identical (config, seed) pairs always produce identical metrics. Single-seed figures carry roughly ±2 s of noise on mean_wait_s, so average across at least 3 seeds before concluding that one timing beats anot

   First seen 12 Oct 2026. Unchanged since.

7. set_scenarioCan change things

   Switch the demand profile the junction is running under, and what the operator sees. Like apply_config, this RESTARTS the live run. The best timing differs substantially between scenarios, and so does which left-turn phasing is admissible — a timing that is legal under one demand profile may be prohibited under another. To evaluate against a scenario without switching the operator's view, pass sc

   First seen 12 Oct 2026. Unchanged since.

8. unpin_resultCan change things

   Remove a timing from the operator's comparison tray, or clear the tray entirely. Use this when a pin turns out to be wrong — superseded by a better candidate, or measured under assumptions that have since changed. Without it you would have to ask the operator to clean up after you.

   First seen 12 Oct 2026. Unchanged since.

Run this site? This page shows what your site told our crawler, ToolChronicleBot. To be left out, disallow it in robots.txt or write to hello@toolchronicle.com.