30d roadmap · 7 milestones
Harness Engineering
Seven progressive builds turn instructions, repository legibility, environment setup, durable state, scope, feedback, verification, lifecycle, and evaluation into one complete coding-agent harness.
Outcome: Design, build, verify, and evaluate the environment around a coding agent so it can make reliable progress across bounded and long-running tasks.
How to work this roadmap
Treat each milestone as a claim that must be supported by evidence. Before reading, write what you think the mechanism does and where it will fail. After studying the linked concepts, run the drills without copying an answer, preserve the output, and revise the explanation. Move forward when you can connect the milestone goal to a working implementation, benchmark, architecture decision, or reviewable design artifact.
The 30d horizon is a sequencing aid, not a completion badge. Spend more time where your prediction and the observed behavior disagree. Keep a short decision log containing the mechanism selected, alternatives rejected, expected failure mode, measurement used, and remaining uncertainty. Review that log with FSRS prompts so the roadmap produces durable system judgment rather than a temporary tour of terminology.
At the end, explain Design, build, verify, and evaluate the environment around a coding agent so it can make reliable progress across bounded and long-running tasks. from first principles to a reader outside the domain. A strong explanation should survive follow-up questions about correctness, cost, latency, resource use, security, recovery, and operational visibility. If it cannot, return to the milestone that contains the missing mechanism and build a smaller falsifiable example.
Milestones
Milestone 1
Build 1 — Prompt-only versus rules-first
Turn implicit expectations into scoped repository instructions with executable checks.
Concepts
Milestone 2
Build 2 — Agent-readable workspace
Make repository structure, bootstrap, readiness, and capability boundaries discoverable from a fresh worktree.
Concepts
Milestone 3
Build 3 — Multi-session continuity
Persist goal, progress, evidence, blockers, and next action so a new session resumes safely.
Concepts
Milestone 4
Build 4 — Runtime feedback and scope control
Bound authority and change size while using runtime evidence to continue, complete, replan, or escalate.
Concepts
Milestone 5
Build 5 — Independent self-verification
Separate maker claims from a checker that reconstructs the contract and gathers fresh evidence.
Concepts
Milestone 6
Build 6 — Complete harness lifecycle
Connect initialization, bounded work, durable progress, checking, recovery, and terminal states in one harness.
Concepts
Build evidence
- Build a Complete Coding-Agent Harness — Build a repository-local harness that can take one bounded engineering task from initialization through implementation, verification, durable handoff, and explicit termination without relying on hidden machine state or chat memory.
Milestone 7
Build 7 — Automated maker-checker loop
Run a bounded autonomous loop, compare it with a baseline, and ablate one harness component before deciding whether it should ship.
Concepts
Build evidence
- Automate and Evaluate a Maker-Checker Loop — Automate a bounded loop in which a maker attempts one task, an independent checker evaluates explicit acceptance criteria, failures return actionable evidence, and the orchestrator stops on success, escalation, cancellation, or budget exhaustion. Compare it with a baseline harness on a frozen task suite.