CLI reference: Experiments

6 commands. Every command runs on your own computer; nothing is sent to a social network unless a publishing command is run by the owner with its live switches.

reachweaver experiment-preregister

Offline: validate a growth plan and write a NEW hashed preregistration; never publishes

reachweaver experiment-preregister [-h] --out OUT plan

Option	Description
plan	Growth plan JSON (examples/growth-plan.example.json)
--out OUT	NEW 0600 record file; an existing file is never overwritten

reachweaver experiment-verify

Offline: detect post-hoc edits by recomputing a plan’s sha256 hash

reachweaver experiment-verify [-h] --hash EXPECTED_HASH plan

Option	Description
plan	Original growth plan JSON, or the preregistration record
--hash EXPECTED_HASH	Registered plan_hash, sha256:<hex>

reachweaver experiment-evaluate

Matched-age, non-causal readout of preregistered post pairs

reachweaver experiment-evaluate [-h] --file FILE

Option	Description
--file FILE

reachweaver experiment-readout

Read-only: verify a preregistration, map arms to posts, label the matched-age readout; non-causal

reachweaver experiment-readout [-h] --prereg PREREG --posts POSTS
                                      [--hash EXPECTED_HASH] [--attribution ATTR.json]
                                      [--attribution-source {analytics,ga4}] [--json]

Option	Description
--prereg PREREG	Record written by experiment-preregister
--posts POSTS	reachweaver.growth-post-map JSON: arm id -> post id per pair
--hash EXPECTED_HASH	Optional externally kept plan_hash that must also match
--attribution ATTR.json	Saved ‘attribution-report –json’ output; required for a docs_visits or repo_stars_utm primary metric (per-post attributed sessions/conversions; unknown stays unknown, never 0)
--attribution-source ATTRIBUTION_SOURCE	Which section of –attribution to read: analytics (the –analytics join, default) or ga4 (the –ga4-csv join; key events count as conversions). Never mixed One of: analytics, ga4.
--json

reachweaver experiment-next

Offline: deterministic next dimension from prior classified results

reachweaver experiment-next [-h] [--results RESULTS] [--readout FILE]
                                   [--readout-dir DIR] [--json]

Option	Description
--results RESULTS	JSON list of {experiment_id, dimension, label, direction}
--readout FILE	Saved ‘experiment-readout –json’ file; its label and direction are used unchanged; repeat
--readout-dir DIR	Folder of saved ‘experiment-readout –json’ files (*.json; other JSON files are listed as ignored). Give –results, –readout or –readout-dir
--json

reachweaver experiment-simulate

Offline SIMULATED label distribution for a seeded synthetic experiment

reachweaver experiment-simulate [-h] --seed SEED --pairs PAIRS --lift LIFT --noise
                                       NOISE [--trials TRIALS] [--json]

Option	Description
--seed SEED
--pairs PAIRS	Pairs per synthetic experiment (2..100)
--lift LIFT	True relative lift of the treatment, e.g. 0 or 0.2
--noise NOISE	Per-post log-normal noise sd, e.g. 0.3
--trials TRIALS	Synthetic experiments to run (1..20000)
--json