Delivery-capacity landing governor — by TextForAgents LLC. GitHub logged 7.38 billion commits in one month and is rebuilding Git storage; GitLab is rebuilding source control itself — because agent fleets generate change volume no human-era pipeline was sized for. The platforms fix their own plumbing. Your existing repo still gets no budget, no meter, and no triage. This governor is that missing layer: every merge, checkout, CI minute, and submission your agents consume is metered per accountable operator across all four surfaces — under a project-wide capacity cap checked before the per-operator budget, so spinning up fresh agent identities mints no new capacity (sybil-resistant by construction, with a hash-chained attempt log to prove it). Every agent-authored change arrives with a pre-triage verdict — structural receipt (build/tests, declared-vs-actual files, risk note) plus the operator's acceptance history — that re-orders the human review queue: in the lab's modeled simulation, verdict-first landed 396 good changes per reviewer budget vs FIFO's 85. And an optional Git proxy serves scoped sparse checkouts for existing repos through the stock git protocol: 81% fewer bytes transferred in the lab, metered per checkout, never flat. Proven in the lab: 17/17 tests, top operator measured at 38.8% of a fleet's total capacity, metering at 0.054 ms per event. Stated plainly: reviewer-time figures are MODELED (reviewer minutes modeled, not human-measured), demand is INFERRED from the GitHub/GitLab platform rebuilds with no operator purchase measured, and cache byte savings at lab scale were small (~4%) and are not oversold.
Meter + Verdicts $9/mo per operator · Per-verdict API $0.01/verdict (x402) · Proxy $0.01/proxied checkout (metered — no flat proxy tier). First 100 verdicts free, one time per operator (account-identity enforced; a second account for the same operator does not re-grant). There is no perpetual free tier.
Decisions: admit / warn (at >=80% of the per-operator budget) / deny (per-operator budget exceeded) / deny_project_cap (project-wide cap exceeded, checked first). Denied events consume no budget. A verdict (score 0-100, disposition LAND / REVIEW / REJECT) is a machine ordering of reviewer attention, not a correctness guarantee; the lab's modeled false-reject cost was ~5.7% of good changes wrongly auto-rejected (MODELED minutes, not human-measured), and REJECT remains appealable to a human reviewer under the operator's own policy. Proxy scope paths rejecting '.', '..', absolute, empty-segment and '.git/' paths are refused before any charge. The proxy serves checkouts through the stock git protocol only — there is no server-side repository query API. Shared object cache is capped per project with eviction, never unbounded. Request bodies are capped at 2 MB.
POST https://textforagents.com/products/delivery-capacity-landing-governor/register {"name": "your-operator-name", "payer_wallet": "0x..."} → account_id + claim_token. Then POST https://textforagents.com/products/delivery-capacity-landing-governor/trial/claim {"claim_token": "..."} → your API token (shown once). Create a project: POST https://textforagents.com/products/delivery-capacity-landing-governor/projects {"name": "my-project", "origin_url": "file:///path/to/repo or https://..."} → project id + per-project forge webhook secret (shown once). Meter (Meter + Verdicts subscription required): POST https://textforagents.com/products/delivery-capacity-landing-governor/meter {"project_id": "...", "operator": "agent-1", "surface": "merge|checkout|ci_minutes|submission", "units": 1}. Verdict: POST https://textforagents.com/products/delivery-capacity-landing-governor/verdicts {"project_id": "...", "operator": "agent-1", "receipt": {"build": true, "tests_passed": 20, "tests_total": 20, "files": ["a.py"], "declared_files": ["a.py"], "intent": "...", "risk_note": "..."}}. Ranked queue: GET https://textforagents.com/products/delivery-capacity-landing-governor/projects/<id>/queue. Attempt log: GET https://textforagents.com/products/delivery-capacity-landing-governor/projects/<id>/log (hash chain verified server-side). Proxy checkout: POST https://textforagents.com/products/delivery-capacity-landing-governor/proxy/checkout {"project_id": "...", "operator": "agent-1", "scope": ["mod03"]} ($0.01 x402). Forge ingestion: POST https://textforagents.com/products/delivery-capacity-landing-governor/projects/<id>/ingest with header X-Forge-Signature: sha256=<HMAC-SHA256 of the raw JSON body under the project's webhook secret> and body {"event_type": "merge|checkout|ci|submission", "operator": "agent-1", "units": 1}. Subscribe: POST https://textforagents.com/products/delivery-capacity-landing-governor/subscribe {} ($9/mo x402 — Meter + Verdicts for one operator account). MCP: POST https://textforagents.com/products/delivery-capacity-landing-governor/mcp (tools: governor_record, governor_log, governor_stats, verdict_submit, verdict_queue, proxy_checkout).
Docs: https://textforagents.com/products/delivery-capacity-landing-governor/docs · OpenAPI: https://textforagents.com/products/delivery-capacity-landing-governor/openapi.json · llms: https://textforagents.com/products/delivery-capacity-landing-governor/llms.txt · x402 discovery: https://textforagents.com/products/delivery-capacity-landing-governor/.well-known/x402 · Terms: https://textforagents.com/products/delivery-capacity-landing-governor/terms · Privacy: https://textforagents.com/products/delivery-capacity-landing-governor/privacy · Support: support@textforagents.com
Reviewer-time figures are MODELED, demand is INFERRED (see above).