AgentMeasure

# AgentMeasure **Find repeated failures and retries in your Codex and Claude Code sessions, with local evidence.** Healthcheck reads existing **Codex rollout logs** and **Claude Code session logs** and produces a terminal summary and a local HTML report. It checks duplicate records, retry chains, and consecutive tool failures (HC-01..03), plus audit checks for operation-resolution coverage, cache accounting, and token stability (HC-04..06, `--audit`). Missing evidence is **UNPROVABLE**, never silently zero. ```bash # Python 3.9+. NOTE: PyPI still serves 0.4.0, which crashes on the first # command (NameError on demo/check/compare). The 0.4.1 fix is on GitHub but # not yet on the index — install from git until it lands. pipx install "git+https://github.com/roy-tong/AgentMeasure#subdirectory=healthcheck" agentmeasure demo # synthetic example; no personal logs needed agentmeasure check # your local sessions, last 7 days agentmeasure check --runtime claude # force the Claude Code adapter ``` Analysis runs locally with no runtime network calls. Shipped in v0.4.0/v0.4.1: Claude Code adapter v1, embedded conformance pack (`agentmeasure conformance`, also a [GitHub Action](action.yml)), OTel / Prometheus exports, run trends (`agentmeasure trend`), and settlement statements for outcome-based billing — `agentmeasure settle --format md|html` generates an [AMS-1](standard/SETTLEMENT.md) one-pager (two lines, both directions, cannot-settle removals, third-party reproduction block) plus the dispute-bundle JSON. For per-resolution vendors there is a full buyer-side pipeline: `agentmeasure prepare` maps a native Intercom/Zendesk export onto the canonical columns (judgement columns stay empty for human review, every value carries provenance); `recount` applies the vendor's own published rules line by line, with an optional `--contract` overlay that lists outcome-standard findings as a separate lane, never netted into the claim; `dispute` builds the negotiable pack; `recovery` records what the vendor actually conceded, counted separately from the claim. `verify` composes all lanes into one Verified Ledger (json + md), `crosscheck` puts your billing ledger beside the export, `dashboard` renders the whole story as an offline single-file HTML page, `rules-diff` gates a material vendor-rule change (exit 1), `narrative` drafts the settlement letter in two steps where the model's draft is refused if any number does not come from the pack, and `delivery` logs engagements to report the appendix-E reuse metrics. [**Quick start and supported formats**](healthcheck/README.md) · [**AMS-1: open settlement-statement standard**](standard/SETTLEMENT.md) · [**Try it and share feedback safely**](campaigns/healthcheck-first-run.md) · [**What our contributions changed**](campaigns/measurement-casebook.md) · [**The Token-Accounting Bug Report — ~110 tools audited, 45+ verified bugs, 23 fixes merged or accepted upstream (openlit, langfuse, codeburn and more)**](campaigns/audit-report-2026-09.md) · [**Maintain a usage tool? Audit it in 10 minutes**](campaigns/tool-authors.md) · [中文](README.zh-CN.md) Found a useful result? Keep a snapshot and compare your next run. Feedback is optional: you can use the tool without opening an issue or uploading logs. ## Measurement infrastructure **The open yardstick for agent usage and AI outcomes.** **Test whether your agent metrics mean what their labels claim.** Conformance checks for AI-agent telemetry — a retry is one logical operation, not two requests; a reasoning-token subset must not be added into totals; a cache hit is not a new measurement. Every check reports **PASS / FAIL / UNPROVABLE**, and UNPROVABLE is a first-class result: when the evidence to decide is absent, it is disclosed, never zeroed. External evidence keeps arriving: one dedup fix survived a maintainer's 136k-event corpus re-review ([codeburn #1264](https://github.com/getagentseal/codeburn/pull/1264)), and the first externally surfaced real-world case — a public 436k → 54,154-token miscount where a replayed cache prefix was read as fresh per-spawn overhead — is now a [conformance evidence case](conformance/evidence/rulestack-cache-prefix-overhead/README.md) with cited sources. ```yaml # .github/workflows/conformance.yml — turn measurement assumptions into CI checks - uses: roy-tong/AgentMeasure@2cf476d6f7d0fc45401db5a822e1f12de009ac74 with: fixture: fixtures/telemetry.jsonl # your FMT-002 event fixture ``` [**Run conformance locally**](conformance/pack/README.md) · [**See real failures**](#external-provider-trials) · [Read the spec](standard/CORE.md) · [中文](README.zh-CN.md) [](https://github.com/roy-tong/AgentMeasure/actions/workflows/conformance.yml) [](standard/CORE.md) [](https://github.com/roy-tong/AgentMeasure/releases) [](LICENSE) [](https://github.com/roy-tong/AgentMeasure/discussions) > **AgentMeasure separates execution facts from logical operations, evidence > from inference, and economic settlement from value** — an open measurement > layer for the Agent Capability Economy. One number for your dashboards > (logical operations), the evidence for what each number counts, and explicit > disclosure for what it cannot prove. > **When machine execution produces charges, execution facts and billing facts > live in different systems.** AgentMeasure is the independent verification > layer between them: what actually happened, what the declared policy says, > whether the two agree — and what cannot be proven. It measures and checks > conformance to declared rules; it does not adjudicate (no "should have been > billed" verdicts, no liability). Payment receipts prove money moved; > **[AMS-1 settlement statements](standard/SETTLEMENT.md)** prove what the > money was for. Verdicts come only from named rules with tests behind them — > models, where used at all, sort and summarize but never judge — so the same > inputs re-run to the same statement on any machine. **Today:** measure agent-facing capability usage — attempts, operations, retry inflation, success rates with numerators you can audit. **Next:** make capabilities comparable and meterable — including the outcome units now being billed (what counts as one resolution, one completed task). **Long term:** provide the measurement foundation for Capability as a Service (CaaS). **Reach → Choice → Use → Utility → Value** > AgentMeasure is **not** a payment protocol, marketplace, or universal ranking system. > It standardizes the facts and measurement semantics those systems can build on. ## Why now AI services have started charging by the outcome: [Zendesk at $1.50–2.00 per automated resolution](https://www.zendesk.com/newsroom/articles/zendesk-outcome-based-pricing/) since Aug 2024, [Intercom Fin at $0.99 per resolution](https://www.intercom.com/pricing/fin) with a money-back guarantee, [Salesforce Agentforce at $2 per conversation](https://www.salesforce.com/agentforce/pricing/), plus Sierra, Ada, and Decagon. In two years, per-outcome pricing went from differentiator to industry default. In May 2026 Zendesk went a step further and began billing only "Verified Resolutions" — vendors are now competing to define what one outcome is. The buyer still has no checker. The rules are written by the seller and the bill is computed by the seller; Intercom's own documentation says "We cannot guarantee the limit will be 100% accurate." We checked 20 vendors' public docs — none publishes a dispute or refund process for miscounts. So we are building the buyer's side of the bill: re-run it on your own machine and get an [AMS-1 settlement statement](standard/SETTLEMENT.md) — what was charged, what the vendor's own rules would charge, and what the evidence cannot prove. Per resolution, per task, or per usage: one settlement semantics for all three billing units. When money rides on a measured unit, someone has to define the unit. That is what we build. → [Read the note](https://roy-tong.github.io/AgentMeasure/blog/outcome-yardstick.html) ## Pricing The standard and the local tooling are open source and free, permanently. What we sell is reconciliation work on top of them — every fee below maps to a countable unit, the way we think all AI billing should work. | Tier | Price | What you get | |---|---|---| | Open source | **$0, forever** | CLI checker, conformance pack, CI action, self-serve AMS-1 statements — [run it yourself](healthcheck/README.md) | | First-Look reconciliation | **$990 one-time**, per vendor export | You send the export (Intercom Fin / Zendesk / Decagon); we return the AMS-1 one-pager — confirmed vs assumed resolutions, the dollar delta, a named rule behind every figure — plus flagged dispute lines, the dispute bundle, and a 30-minute walkthrough. Processed locally. | | Monthly reconciliation | **from $490/month** | The same statement every month, rule updates when a vendor changes how it counts, dispute-line watch, quarterly review. Tiers: ≤10k resolutions/mo **$490** · 10k–50k **$990** · larger or multi-vendor stacks quoted as enterprise. | No accuracy SLAs — every deliverable ships a coverage block instead (what entered the computation, what stayed UNPROVABLE). No recovery commissions — we produce the evidence; the negotiation stays yours. Design partners: the first five are free, in exchange for a public testimonial. → [Pricing page](https://roy-tong.github.io/AgentMeasure/#pricing) · [tongroy18@gmail.com](mailto:tongroy18@gmail.com) ## Principles (written down before anyone asked) - **Free, forever.** Spec, engine, SDK, conformance, local dashboard — the free list is part of governance, not revocable later. Adoption is the product. - **Local-first, opt-in.** Raw data never leaves your machine by default; only aggregated, anonymized evidence returns, and only if you opt in. - **The referee doesn't open a store.** No paid rankings, no recommendation slots, no custody of funds. Neutrality is the product. [**Website**](https://roy-tong.github.io/AgentMeasure/) · [Send a trace → get a measurement check](mailto:tongroy18@gmail.com?subject=AgentMeasure%20measurement%20check%20-%20%5Byour%20capability%5D&body=Hi%2C%0A%0AI%27d%20like%20a%20zero-install%20measurement%20check.%0A%0A1.%20Data%3A%2020-100%20anonymized%20trace/log%20rows%2C%20or%20a%20link%20to%20a%20public%20export%20%28production%20or%20synthetic%20-%20please%20say%20which%29%3A%0A%0A2.%20Source%20and%20time%20window%3A%0A%0A3.%20The%20decision%20this%20should%20inform%3A%0A%0ANote%3A%20raw%20data%20stays%20local%20by%20default.%20If%20I%20share%20a%20sanitized%20sample%2C%20I%27ll%20state%20what%20may%20be%20done%20with%20it.) · [Free 7-day audit — apply](https://github.com/roy-tong/AgentMeasure/issues/new?template=5-provider-trial.yml) · [30 Projects / 30 Days campaign](campaigns/30-projects-30-days.md) · [Whitepaper](whitepaper/measuring-software-used-by-ai-agents.md) · [Core Specification](standard/CORE.md)  **Start with the story:** [When the Software Consumer Becomes an Agent](https://roy-tong.github.io/en/notes/when-the-software-consumer-becomes-an-agent/) (EN) · [当软件的消费者变成 Agent](https://roy-tong.github.io/notes/when-the-software-consumer-becomes-an-agent/) (ZH) ## Try it in 2 minutes ```bash ./examples/demo-e2e.sh ``` Mock MCP server → canonical observations → local metrics, all on your machine, no cloud. The demo is **reproducible**: it runs in an isolated workspace (never touches `~/.agentmeasure`) — same fixture + same policy = same result (42 calls → 84 canonical observations, caller claims claude:14 · codex:14 · unknown:14). Then read why we audit the ecosystem's usage claims: [Benchmark Run #001](reports/benchmark-run-001.md) — six real claims profiled with the Evidence Profile (multi-axis, no composite scores), and [Pipeline Validation #001](reports/pipeline-validation-001.md) — our own fixture, kept out of the ranking as a reference baseline. --- ## Run a preregistered experiment — AgentMeasure Lab ```bash python3 lab/am lab selftest # planted uplift recovered + honest null python3 lab/am lab init # workspace + example experiment python3 lab/am lab preregister am-lab/experiments/example-manifest.json python3 lab/am lab run am-lab/experiments/example-manifest.prereg.json ``` The open experiment engine ([lab/](lab/README.md)): task set × harness matrix × factor variants → Reach → Choice → Success → Consumption funnel → effect sizes with confidence intervals, guardrails, honest nulls, and an offline HTML report that opens with a bilingual decision-maker one-pager. Preregistration is enforced (hypothesis / primary metric / guardrails / analysis plan hashed before the run — with a scale / power / budget preview), seeds replay deterministically, and a budget circuit breaker stops safely with the data it has. Selection uplift that loses consumption is **rejected at the decision exit** (`unverified_growth` — do not ship), and candidates that make no more money at higher cost are flagged as dominated. The shipped demo runs on a **synthetic harness** (planted ground truth at realistic, literature-scale amplitudes, disclosed in every report) — it validates the engine, not real-agent claims. Real harness adapters (Claude Code / Codex) are runner plugins against the same interface; that is the highest-value contribution right now. Docs: [lab/README.md](lab/README.md) · formats: [lab/schemas/](lab/schemas/) (experiment manifest / funnel events / report). --- ## Why the split matters, in one line ```text 1 user intent 2 provider attempts 1 final success ``` That is **1 operation**, **2 attempts**, operation success **100%**, attempt success **50%**, attempts per operation **2**. It is *not* "2 operations with 50% success". Attempts are execution facts. Operations are logical intents. Mixing the two is how agent reliability reports get distorted. --- ## External provider trials Start smaller than a trial: **send a trace, get a measurement check** — 20–100 anonymized trace/log rows (or a public export), mapped locally, with a short report of what safely counts as attempts vs operations, where retries may inflate usage, and what the telemetry cannot prove. No SDK, no integration; raw data stays local unless you explicitly authorize sharing a sanitized sample. A sample check is not a full audit and claims no causal effect. If the check pays for itself, AgentMeasure is looking for **3 MCP/API providers** for short external measurement runs. The goal is not to prove the model works. It is to find where it breaks on real provider traffic. A trial can stay fully local, requires no agent-side install, and can start with one capability. **Best feedback:** a concrete trace, counterexample, or field that cannot be measured safely. → The formal offer, schedule, and FAQ live in **[product/AUDIT.md](product/AUDIT.md)** · apply in **[Issue #2 — Looking for 3 MCP/API providers for external measurement runs](https://github.com/roy-tong/AgentMeasure/issues/2)**. --- ## Why capabilities need a new measurement layer The software consumer is changing from humans to agents, and the economic unit is shifting from software seats toward callable capabilities. ```text Skill / MCP / CLI / SDK ↓ Describe / expose / distribute a capability Capability ↓ Data / Compute / Action / Permission / Transaction ↓ Creates scarce economic value ``` **Interfaces may become cheap to create; capabilities remain scarce to deliver.** The first generation of capability distribution is already here — open Skills, open MCP adapters, open CLIs. The scarce layer underneath them is what the next economy is built on: proprietary data, compute, execution, permissions, and real-world fulfillment. ## From software to capability economy ```text Human Software Economy User → UI → SaaS → Seat / Month ↓ Agent Capability Economy Agent → Capability → Execution → Outcome ↓ Usage / Value / Transaction ``` If a capability is to become an economic unit that agents can discover, compare, and eventually purchase automatically, it must first be **identifiable, measurable, and comparable under a shared semantics**. That is what AgentMeasure provides. Traditional usage metrics cannot support this economy — the old chain breaks at every link, and the last link is new: ```text Install ≠ Available Available ≠ Presented Presented ≠ Selected Selected ≠ Used Used ≠ Useful Useful ≠ Incremental Value Measured Usage ≠ Billable Usage ``` The last inequality is why the Operation/Attempt model matters commercially: 3 attempts of one operation are not 3 billable operations — unless the metering policy says so. > The extended thesis — economic units, scarcity, measurement before monetization, > metering semantics, and the evidence that commerce is arriving before measurement — > lives in [docs/CAPABILITY-ECONOMY.md](docs/CAPABILITY-ECONOMY.md). ## The measurement view: Reach → Choice → Use → Utility → Value AgentMeasure defines **metric families**, not a universal KPI. A search capability, a booking API, and a compute job have different value structures. | Layer | Question | Representative metrics | | --- | --- | --- | | **Reach** | Did the capability enter the agent's choice range? | Eligible Opportunities · Presentations · Presentation Rate · Distribution Coverage | | **Choice** | When the agent had the chance, did it choose it? | Observed Selection Rate · Observed Head-to-Head Choice Share | | **Use** | After selection, was it actually used? | Operations · Attempts · Completion Rate · Success Rate | | **Utility** | Did it produce a usable result or confirmed effect? | Result Consumption · Effect Confirmation | | **Value** | Did it improve the task outcome? | Incremental Task Success（Draft 0.5） | The five layers are the **measurement view**. The same facts map onto an economic view: | CaaS Domain | AgentMeasure | | --- | --- | | Demand | Reach + Choice | | Delivery | Use + Utility | | Outcome | Value | | Economics | Metering / Attribution（future extension） | Claim discipline throughout: *observed choice is not preference*. A selection can be made by the model, a router, a workflow, the user, a policy, or the platform; **Observed Selection Rate** reports what was observed, and **Observed Head-to-Head Choice Share** is an *observed head-to-head choice share under comparable candidate conditions* — comparable means the same candidate set, category, choice mode, and decision axes (Decision Authority / Selection Constraint) are declared. ## How AgentMeasure works in production ```text Agent Runtime Capability Provider Claude / Codex │ │ MCP / API ▼ ┌──────────────────────┐ │ Customer Capability │ │ │ │ AgentMeasure SDK │ │ Business Handler │ └──────────┬───────────┘ │ observations │ ▼ Collector │ ▼ AgentMeasure Cloud ``` Four things a real developer needs to know: > Your software does **not** need to be open source. > > MCP is **not** required — it is the first reference surface. > > Third-party agents do **not** need AgentMeasure installed for provider-side usage measurement. > > AgentMeasure is **not** on the critical request path. Product architecture (Provider SDK → local buffer → hosted ingestion → dashboard): [product/ARCHITECTURE.md](product/ARCHITECTURE.md). ## What AgentMeasure measures today - **Decision Opportunity / Candidate Set / Presentation / Selection** — the four objects of choice; the Observed Selection Rate denominator is Presented, not Available - **Software Entity → Capability → Interaction Surface** — what exists, what it can do, and the observable interface; observation happens on surfaces, attribution resolves to entities through the machine-readable registry - **Operation / Attempt** — one logical use vs. one execution; **retries are multiple attempts of one operation**, not a validity class (they are kept as reliability signals, not counted as distinct logical uses) - **Qualified Usage** — production usage after excluding test, benchmark, synthetic, replay, duplicate and other invalid traffic according to policy - **Result Consumption** — *defined, reference partial*: the result was used by the task - **Effect Confirmation** — *domain model defined, metric planned for Draft 0.5*: the intended world-state change was confirmed - **Measurement Label** — a nutrition label for every public number (coverage / sampling / policy / method) The full model lives in the [Core Specification](standard/CORE.md), [Metrics](standard/METRICS.md), [Entity](standard/ENTITY.md), and [Quality](standard/QUALITY.md). The README is not a spec summary. ## From measurement to CaaS AgentMeasure is **progressively standardizing** the measurement chain from discovery and choice through execution, utility and value: core usage semantics are defined today; utility/value and commercial metering remain active drafts. Metering and commercial attribution are future extensions; payment rails can be provided by existing payment infrastructure. > **AgentMeasure standardizes economic facts, not money movement.** > > Extended thesis: [docs/CAPABILITY-ECONOMY.md](docs/CAPABILITY-ECONOMY.md) · > Economic semantics: [extensions/COMMERCIAL.md](extensions/COMMERCIAL.md)（Experimental） ## Harness profiles — what each runtime can (and cannot) observe Portable semantics need a public record of observation blind spots. Per-harness profiles map each runtime's native objects to AgentMeasure semantics: | Harness | Profile | Highlights | | --- | --- | --- | | Codex | [profiles/codex.md](profiles/codex.md) | hook 观察，无 trace/精确时间戳；App Server 事件流为优先生效观察面 | | Claude Code | [profiles/claude-code.md](profiles/claude-code.md) | 内置成败判定；第一个 Consumption 可实证平台 | | DeepSeek Harness | [profiles/deepseek-harness.md](profiles/deepseek-harness.md) | append-only session log；subagent lineage/depth 是 Delegation 的首个真实数据源 | | Pydantic AI | [profiles/pydantic-ai.md](profiles/pydantic-ai.md) | Logfire spans → attempt 语义 | | OpenTelemetry GenAI | [profiles/opentelemetry-genai.md](profiles/opentelemetry-genai.md) | Route B mapping | As harnesses compose software at runtime, one behavior gets described by different objects and units in different runtimes. [Experiment D](experiments/EXPERIMENT-D-cross-harness-compatibility.md) turns that into evidence; [Proposal: Delegation](proposals/2026-08-21-delegation-graph.md) defines the agent-to-agent boundary the object model was missing. ## What AgentMeasure is / is not | AgentMeasure is | AgentMeasure is not | | --- | --- | | Measurement standard | Payment protocol | | Usage analytics foundation | Marketplace | | Metering semantics | Wallet | | Comparable quality signals | Universal reputation score | | Attribution framework | Single global source of truth | ## Who it is for | Audience | Why | | --- | --- | | **Teams billed per outcome** (AI support: Intercom Fin, Zendesk, Salesforce Agentforce) | re-run the bill locally; AMS-1 settlement statements for renewals and disputes | | **Teams billed per task / compute** (Copilot premium requests, Cursor, Devin ACU) | one semantics for "what counts as one task" when retries and loops keep billing | | **Teams billed per usage** (APIs and gateways, multiple vendors) | align incommensurable meters into one auditable statement | | **Capability Provider** | measure and eventually meter agent usage of your capabilities | | **Agent Runtime** | expose decision / usage signals consistently | | **Registry / Marketplace** | compare capabilities using standardized signals | | **Data / Measurement Provider** | produce comparable agent-usage analytics | | **Commerce / Payment Infrastructure** | consume standardized billable events in future profiles | | **Researchers / Standard Contributors** | evolve the methodology | ## Try the standard ```bash git clone https://github.com/roy-tong/AgentMeasure && cd AgentMeasure python3 conformance/runners/run_metrics.py # metric vectors (M2.2 / M2.5 / M4.1) python3 verify_vectors.py # verification / correlation / operation vectors python3 registry/validate_entities.py # validate the machine-readable registry ``` ## Product MVP — first real measurement (in development) The first product path is **Remote MCP / API Capability Measurement**: an [AgentMeasure Provider SDK](sdk/) (`@agentmeasure/mcp`) that emits observations from the provider side (no agent-side install), feeding a local collector. Local analytics run without any cloud: ```bash npm install https://github.com/roy-tong/AgentMeasure/releases/download/v0.1.1/agentmeasure-mcp-0.1.1.tgz # (npm registry publish pending scope/token — tarball is the current install path) # wrap your MCP server's tool handlers: server.tool = (name, schema, mw.wrapTool(name, handler)) node examples/mcp-integration.js # synthetic traffic → local JSONL python3 product/local-analytics.py ~/.agentmeasure/events/agentmeasure-events.jsonl ``` Status: SDK v0.1.1 — External-Ready (canonical output, non-blocking spool with loss accounting, per-request caller, MCP v1/v2, 21 tests, deterministic fixture) + local analytics implemented; hosted ingestion and dashboard next. First real external Provider = Product Gate A ([ROADMAP.md](ROADMAP.md), [MVP.md](product/MVP.md)). Scope and acceptance: [product/MVP.md](product/MVP.md) · SDK contract: [product/PROVIDER-SDK.md](product/PROVIDER-SDK.md) · Deployment: [product/DEPLOYMENT.md](product/DEPLOYMENT.md) ## Repository map ```text AgentMeasure/ ├── standard/ # the normative standard (CORE / METRICS / QUALITY / DATA / ...) ├── extensions/ # experimental, non-normative profiles (COMMERCIAL.md) ├── product/ # product architecture (SDK / hosted analytics, in development) ├── whitepaper/ # methodology papers (EN/CN) ├── conformance/ # language-neutral test vectors + runners ├── reference/ # reference implementation (collector + adapters) │ ├── collector/ # normalization, correlation, aggregation, evidence │ └── adapters/ # codex / claude / dsh / mcp observation adapters ├── schemas/ # machine-readable schemas (entity registry) ├── registry/ # machine-readable registries (entities / project identity) ├── experiments/ # empirical experiment designs ├── reports/ # public reports (Discrepancy Report) ├── proposals/ # standard change proposals (AUP) └── archive/ # retired early documents ``` **The standard is the artifact; the code is a reference implementation.** Using the standard does not mean uploading data to any central server. ## Current status & roadmap **Draft 0.4.5** — 唯一 Canonical Observation（schemas/observation.schema.json，6 类 payload）；Choice/Execution 从同一 Envelope 派生；M3.1 只计已解析 operation（无回退）；Attempt 级 qualification 派生； metrics.yaml 单一事实源；四维正交（Evidence/Caller/Use Profile/Billing）。 | Capability | Standard | Reference | Real Runtime | | --- | --- | --- | --- | | Observed Selection Rate | Defined | Implemented | Limited | | Observed Head-to-Head Choice Share | Defined | Implemented | Experimental | | Operations / Attempts | Defined | Implemented | Yes | | Operation Resolution Coverage | Defined | Implemented | No | | Result Consumption | Defined | Implemented | Claude partial | | Incrementality | Defined (formula) | Planned | No | | Qualified Usage (Strict) | Defined | Implemented | Yes | The roadmap runs on two tracks — the standard (0.4 objects & quality → 0.5 utility & economic semantics → 1.0) and the product (Remote Capability Analytics → Provider SDK + hosted analytics → metering). See [ROADMAP.md](ROADMAP.md). ## Contribute - **Join the community**: [Discussions](https://github.com/roy-tong/AgentMeasure/discussions) — categories: Metric Semantics · Runtime Observation · Experiments · Capability Economy · Implementers; ground rules in [docs/DISCUSSIONS.md](docs/DISCUSSIONS.md) - **Where to start**: the open debate on [Strict Qualified Usage as the default](https://github.com/roy-tong/AgentMeasure/discussions/1), or [onboard the first external Provider](https://github.com/roy-tong/AgentMeasure/issues/2) - **Propose standard changes**: `proposals/` (AUP: Draft → Discussion → Accepted → Experimental → Stable) - **Report measurement discrepancies**: `reports/` (Discrepancy Report template) - **Fix the reference implementation**: PRs must pass all `conformance/` vectors --- *AgentMeasure does not define who owns the truth. It defines what evidence, under what rules, can support what conclusions.*