Back to Plate

{{TITLE}}

.agents/skills/autogoal/assets/templates/major-task.md

53.0.82.7 KB
Original Source

{{TITLE}}

Objective: TODO: Write the short create_goal objective, under 240 characters. Put the full major-task contract in the sections below.

Goal plan: {{PLAN_PATH}}

Template: {{TEMPLATE_PATH}}

Major source:

  • type: pending
  • id / link: pending
  • title: pending
  • decision to make: pending
  • decision criteria: pending

Completion threshold:

  • TODO: Define the decision, proposal, benchmark, architecture, or migration done state.

Verification surface:

  • TODO: Name repo audit, benchmark, external-source audit, review pass, prototype, command, or plan artifact.

Constraints:

  • Start from repo evidence before external claims.
  • Separate measured evidence, source evidence, inference, and recommendation.
  • Do not execute implementation unless this goal explicitly includes it.

Boundaries:

  • Source of truth: TODO.
  • Allowed edit scope: TODO.
  • External sources: TODO.
  • Browser surface: TODO.
  • Non-goals: TODO.

Blocked condition:

  • TODO: Name missing source, benchmark, access, decision, external evidence, or user judgment that stops autonomous work.

Completion rule:

  • Do not call update_goal(status: complete) until evidence is recorded and node .agents/skills/autogoal/scripts/check-complete.mjs {{PLAN_PATH}} passes.

Start Gates:

GateAppliesEvidence
Active goal checked or createdpendingpending
Source of truth read before analysispendingpending
Decision criteria statedpendingpending
Existing repo patterns / prior decisions checkedpendingpending
External research decision recordedpendingpending

Work Checklist:

  • Current state is mapped before proposing a new architecture, migration, benchmark, or plan.
  • Existing repo patterns and prior decisions are recorded before external research.
  • Options, recommendation, tradeoffs, blast radius, and rejection reasons are recorded.
  • Facts, inference, and recommendation are separated.
  • Review or pressure passes are completed, or marked N/A with reason.

Completion Gates:

GateAppliesRequired actionEvidence
Decision criteria satisfiedpendingMap evidence to each criterionpending
Source audit completependingRecord repo evidence and external evidencepending
Review / pressure passpendingRecord review lens or N/Apending
AutoreviewpendingReview final artifact against objective, criteria, constraints, and newest user requestpending
Goal plan completeyesRun node .agents/skills/autogoal/scripts/check-complete.mjs {{PLAN_PATH}}pending

Findings:

  • None yet.

Decisions and tradeoffs:

  • None yet.

Timeline:

  • {{CREATED_AT}}: plan created.