Agent Goals · Checks for the outcome an agent is working toward

Stated, aligned, held to, and reported.

An agent holds a goal well when the goal states the outcome wanted with what it covers, a success test someone could check and a horizon; carries every requirement the principal asked for in the principal's own words; is compared with the principal's other goals, constraints and limits with every contradiction named and settled by the principal; stays the target of the work as actions pile up; is still current, with its premises holding and no later instruction from the principal replacing it; keeps its working scope to what the principal authorized; measures progress by its own success test with the records behind it; and reaches the principal in a report that states its standing, progress, blockers and changes against the goal record. Methods can change as the agent learns; the goal changes only when the principal's ask changes or the principal approves the change. Agent Goals checks each of those with eight tools: eight checks. Your agent sends the goal records, the principal's ask and its requirements, the other goals and comparisons, the action log, the premises and later instructions, the authorized and working scope, the progress claims and records or the goal report in the call, and each tool returns a signed report that names what is vague, misaligned, in conflict, drifted, superseded, exceeded, claimed or inflated, with the fix for each.

See the tools Read the contract

Tool access: 0.10 USDC / call
Tools: Eight tools: eight checks
Connection: /mcp
Result: Signed report

The eight tools.

Each tool reads one question about a goal and answers it from the record. Every tool reads the recorded material in the call: goals and asks, requirements and comparisons, actions, premises and instructions, scope, progress claims and goal reports go in, and a classification comes out. The report names the goal, requirement, action, premise, scope item or claim behind every finding, with the fix for each gap, so your agents and you see why it reads the way it does.

`wever_goal-statement-check`

Reads: Each goal as recorded: the outcome wanted, what it covers and leaves out, how to tell it is met, when it is due and when it is next reviewed.
The signed report states: Outcomes worded as an activity such as improve or work on, success tests that leave success to opinion, goals with no scope, success test or horizon, and reviews set after the horizon.
Price: 0.10 USDC per call
Verdicts: statement_vaguestatement_partialgoal_unstatedstatement_complete

`wever_goal-alignment-check`

Reads: The principal’s ask in their own words or the originating instruction, the working goal, and each requirement in the ask with how the goal carries it and who approved any change.
The signed report states: Requirements dropped from the goal or carried with another figure than the principal stated with no approval, requirements reworded or missing from the ask, and changes approved by no named principal.
Price: 0.10 USDC per call
Verdicts: goal_misalignedalignment_partialalignment_strainedgoal_aligned

`wever_goal-conflict-check`

Reads: One active goal with the principal’s other active goals, standing constraints and limits, and each comparison: whether they contradict, what contradicts what, how it was settled and by whom.
The signed report states: Contradictions left open, averaged away with a compromise, or traded against a constraint or limit by the agent alone, and conflicts with no detail or no recorded decider, each named.
Price: 0.10 USDC per call
Verdicts: goal_in_conflictconflicts_notedconflict_unexaminedconflicts_checked_clean

`wever_goal-drift-check`

Reads: A goal with its outcome and scope, and the agent’s recorded actions in time order, each with what it was for, the reason when it was for something else and who approved it.
The signed report states: Runs of consecutive work toward something other than the goal with no approval, where the earliest divergence sits, single off-goal actions and detours with a reason but no approval.
Price: 0.10 USDC per call
Verdicts: goal_drifteddrift_minordrift_unmeasureddrift_absent

`wever_goal-freshness-check`

Reads: A goal with when it was set, when its context expires and its review interval, the premises it rests on with when each was checked, and every instruction received since.
The signed report states: A later instruction from the principal that replaces the goal, a premise that no longer holds, an expired context, instructions never weighed against the goal, stale premise checks and overdue reviews.
Price: 0.10 USDC per call
Verdicts: goal_supersededfreshness_partialgoal_possibly_stalegoal_current

`wever_goal-scope-check`

Reads: The scope the principal authorized, what the principal ruled out, and each item in the goal’s working scope with when it joined and who approved it.
The signed report states: Additions with no approval or approved only by the agent, items the principal ruled out, additions approved by a team or TBD, and additions that joined before their approval.
Price: 0.10 USDC per call
Verdicts: scope_exceededscope_partialscope_unrecordedscope_as_authorized

`wever_goal-progress-check`

Reads: A goal with its success test, each progress claim with its measure, figure and cited records, and the records with the figures they hold.
The signed report states: Claims with no measure and no record, claims above the figures their records hold, claims measured by something other than the success test, and claims with no record cited.
Price: 0.10 USDC per call
Verdicts: progress_claimed_onlyprogress_partialprogress_unmeasuredprogress_evidenced

`wever_goal-report-check`

Reads: The goal report written for the principal: the goal it names, its standing, progress, blockers and changes as stated, whether each links to the goal record, and the record behind it.
The signed report states: A standing, progress, blockers or changes that flatter the goal record, parts with no record behind them, claims such as on track or no blockers while the goal is at risk, and parts left out.
Price: 0.10 USDC per call
Verdicts: goal_report_inflatedgoal_report_partialgoal_report_missinggoal_report_sound

Start with the two free tools.

On https://weverlabs.com/mcp-free, anonymous and with no wallet, two Agent Goals tools show what a sound goal record holds and what a well-kept goal looks like beside a flawed one. Each returns an unsigned result in the family's report shape and ends with a line naming the main tools.

`wever_goal-snapshot`

Reads: An agent name, a kind of agent such as a coding agent, a procurement agent or a multi-agent team, and optionally the names of its active goals.
Returns: The orientation checklist of a sound goal record for that kind of agent, part by part from the goal as stated to the principal report, with the items the kind of agent adds, the check that reads each part and the active goals named beside it.
Access: Free on /mcp-free

`wever_goal-record-sample`

Reads: Nothing, or one sample goal: a checkout redesign, a vendor selection or a market brief.
Returns: One worked goal record kept right beside a flawed one that states an activity with an opinion for a test, drops a requirement, averages a conflict away, drifts, carries on after the principal replaced it, works outside its scope and claims progress its records do not hold, each piece annotated with why it passes or fails, scored by seven of the family checks. The checks read your own goals.
Access: Free on /mcp-free

How agents and people use it.

Your agents

Pull the checklist with wever_goal-snapshot when a goal is set, so the outcome, success test, scope and review point are on the record before the first action.
Call the check that answers the question on https://weverlabs.com/mcp at the moment it matters: when a goal is written, when the principal's ask arrives, when a new goal joins the others, after each run of work, when a new instruction comes in, when the scope grows, when progress is claimed and before the report reaches the principal.
Act on the verdict and keep the signed report beside the goal it covers, so every change to the goal carries the principal's approval and the check behind it.

You

See the goal your agent is working toward in your own words, with every requirement you set still in it.
Spot work that has wandered from the goal, or a goal you replaced last week that the agent is still pursuing, while there is time to steer.
Read a goal report where the standing and progress match the record, and verify any check against Wever's published trust key whenever you need to show where a verdict came from.

A worked example

Seven actions, and one drift.

The request

{
  "name": "wever_goal-drift-check",
  "arguments": {
    "goal": {
      "id": "g1",
      "outcome": "Raise mobile checkout completion from 61 to 70 percent by January 31",
      "scope": [
        "payment step",
        "address form"
      ]
    },
    "actions": [
      {
        "id": "x1",
        "at": "2026-11-10T10:00:00Z",
        "action": "Cut the payment step to one screen",
        "toward": "payment step"
      },
      {
        "id": "x2",
        "at": "2026-11-11T10:00:00Z",
        "action": "Restyled the newsletter footer",
        "toward": "newsletter footer"
      },
      {
        "id": "x3",
        "at": "2026-11-12T10:00:00Z",
        "action": "Added address autocomplete",
        "toward": "address form"
      },
      {
        "id": "x4",
        "at": "2026-11-13T10:00:00Z",
        "action": "Built the holiday banner",
        "toward": "holiday banner",
        "reason": "the principal asked for it on Monday",
        "approved_by": "head of product"
      },
      {
        "id": "x5",
        "at": "2026-11-14T10:00:00Z",
        "action": "Drafted a loyalty banner",
        "toward": "loyalty banner"
      },
      {
        "id": "x6",
        "at": "2026-11-15T10:00:00Z",
        "action": "Moved the blog to the new fonts",
        "toward": "blog redesign",
        "reason": "it shares the checkout fonts"
      },
      {
        "id": "x7",
        "at": "2026-11-16T10:00:00Z",
        "action": "Started a gift guide",
        "toward": "gift guide"
      }
    ]
  }
}

What the report returns

Verdict: goal_drifted.

Finding, drift run	1 run of 3 or more actions turns away from the goal with no approval: x5 to x7 (3 actions).
Finding, off goal action	1 action serves neither the outcome nor a scope item, with no reason recorded: x2.
Fix, drift run	Stop the run of off-goal work, bring it back to the goal or take the new direction to the principal for approval.
Fix, off goal action	Record what each action is for, and tie it to the goal or take it to the principal.

The work has drifted from the goal, first at x2 (2026-11-11T10:00:00.000Z) after 1 on-goal action.

The goal is mobile checkout completion, and the first action shortens the payment step. The second restyles a newsletter footer that serves neither the outcome nor a scope item: that is where the work first leaves the goal. The holiday banner is a detour the head of product approved, so it stands. Then three actions in a row go to a loyalty banner, the blog fonts and a gift guide, with one reason and no approval between them, and that run is the drift. Bringing the next action back to the payment step or the address form, or taking the new direction to the principal for approval, brings the record back to drift_absent.

Call it from your agent.

Connect to https://weverlabs.com/mcp and call any of the eight tools. Each call costs 0.10 USDC on Base. Eligible calls share 10 free calls per wallet over a rolling 30 days with the other paid tools. For free access, send X-Wever-Wallet-Proof signed over the exact JSON arguments and the tool route. The payment challenge supplies the x402 terms for a paid call.

Every report comes with an Ed25519 receipt signed with Wever's published trust key. The Agent Goals contract lists the schemas, limits, verdicts and signing details. For the promises and deadlines an agent makes in service of a goal, Agent Commitments; for the plan that divides the work among agents, Agent Coordination; for evaluation scores drifting over time, Agent Evaluation; for the confidence an agent states in what it knows, Agent Uncertainty; for an agent updating its methods from experience, Agent Learning; for authorization to act, the mandates on Trading Mandate; for the dates inside one document, the deadline check on Wever Checks.

Browse all tools