Changelog
The work behind the method
A source-backed timeline of the releases and design decisions that make Deep Work Plan reliable for long-horizon agent work.
- v7 · Delegation with evidence Featured release
DWP v7: plans that delegate, with a record of everything
Deep Work Plan v7 keeps the v6 contract and journal, lets a plan hand bounded tasks to other agents, and adds four optional addons for autonomy.
Read the update - v6 · Stricter structure
DWP v6: the same method, a stricter contract
Deep Work Plan v6 keeps the v5 methodology and adds a stricter execution structure. Agent outcome non-inferiority has not been measured.
Read the update - v5 · Guaranteed by tests
DWP v5: the promises, now guaranteed
Deep Work Plan v5 turns eight standing promises into tested, versioned contracts — an honest verifier, safe orchestration, and a proven upgrade path.
Read the update - v4 · Lite-first plans
DWP v4: plans that fit the work
Deep Work Plan v4 shapes plan format around the work: Lite plans become complete, executable proposals for bounded tasks, removing the draft step.
Read the update - v3 · Token efficiency
DWP v3: token efficiency by construction
Deep Work Plan v3 rebuilds the harness around context cost: progressive disclosure, affected-only validation, and a mandatory Final Review.
Read the update - v2 · Official release
DWP v2: the repository becomes the harness
Deep Work Plan v2 makes the repository itself the agent's working harness: context, guardrails, validation, and state live natively in the repo.
Read the update - v1 · Foundation
DWP v1: from a plan loop to a portable harness
Deep Work Plan v1 expands from a single plan loop into a portable adoption kit with provenance, broader agent coverage, and conformance checks.
Read the update