Playproof 0.10.0

What's Changed

• ci(publish): refuse a tag whose commit ci has not passed by @drewstone in #31
• feat(matrix): compare agent profiles across games, with the protocol, the clock and the sensor in the cell by @drewstone in #32
• feat(matrix): run a real coding agent, and give it time to think by @drewstone in #33
• fix(matrix): report every transport fact, do not substitute one for another by @drewstone in #34
• fix(matrix): score the channel the objective names, and rank it the right way by @drewstone in #35
• feat(matrix): author a policy, then evaluate the program by @drewstone in #36
• chore(deps): agent-runtime 0.175.0, add agent-eval, and prove the join compiles by @drewstone in #37
• chore(release): 0.10.0 by @drewstone in #38
• feat(examples): a complete optimization method for a game by @drewstone in #39
• fix(release): quarantine the RetroArch replay gate, and say what that costs by @drewstone in #41
• fix(stream): a child's pipe must not kill the run that owns it by @drewstone in #42

Full Changelog: v0.9.0...v0.10.0

Playproof 0.9.0

What's Changed

• feat(ale): a Breakout ladder that spans the range a player reaches by @drewstone in #29
• chore(release): 0.9.0 by @drewstone in #30

Full Changelog: v0.8.0...v0.9.0

Playproof 0.8.0

What's Changed

• feat(calibration): score what a run made happen, not what it lost by @drewstone in #27
• chore(release): 0.8.0 by @drewstone in #28

Full Changelog: v0.7.0...v0.8.0

Playproof 0.7.0

What's Changed

• feat(episode): stop an episode when the game reports it is over by @drewstone in #25
• chore(release): 0.7.0 by @drewstone in #26

Full Changelog: v0.6.0...v0.7.0

Playproof 0.6.0

What's Changed

• feat(calibration): separate earnable milestones from replay-identity checks by @drewstone in #22
• fix(calibration): a hash check identifies a state, not a trajectory by @drewstone in #23
• chore(release): 0.6.0 by @drewstone in #24

Full Changelog: v0.5.0...v0.6.0

Playproof 0.5.0

What's Changed

• feat(observation): give the agent the real screen, not an ASCII downsample by @drewstone in #20
• chore(release): 0.5.0 by @drewstone in #21

Full Changelog: v0.4.0...v0.5.0

Playproof 0.4.0

What's Changed

• ci(publish): create the GitHub release through the REST API by @drewstone in #17
• feat(calibration): reject a contract a trivial baseline satisfies by @drewstone in #18
• feat(adapters): drive any RetroArch core as a black-box host by @drewstone in #16
• chore(release): 0.4.0 by @drewstone in #19

Full Changelog: v0.3.0...v0.4.0

Playproof 0.3.0

What's Changed

• docs(adapters): RetroArch host candidate; evaluate xemu, WinDurango, Xbox RetroArch by @drewstone in #7
• fix(episode): position scriptedDriver by harness turn so fresh processes resume campaigns by @drewstone in #8
• feat(adapters): any Gymnasium environment as a game by @drewstone in #10
• feat(adapters): Atari through the Arcade Learning Environment by @drewstone in #9
• ci: run every job on the self-hosted Linux pool by @drewstone in #11
• chore(release): 0.3.0 by @drewstone in #12
• ci(publish): keep the job token for the tag identity fetch by @drewstone in #13
• ci(publish): run the release jobs on the ci-release pool by @drewstone in #14
• ci(publish): publish without provenance from the self-hosted pool by @drewstone in #15

Full Changelog: v0.2.0...v0.3.0

Playproof 0.2.0

What's Changed

• ci: make publish idempotent during registry lag and keep the ROM regression local by @drewstone in #3
• feat(campaign): segmented, steerable, resumable runs with analyst hooks by @drewstone in #5
• test(pyboy): run the real-emulator regression on the free Libbet ROM in CI by @drewstone in #4
• feat(adapters): libretro consoles through stable-retro by @drewstone in #6

New Contributors

• @drewstone made their first contribution in #3

Full Changelog: v0.1.0...v0.2.0

Playproof 0.1.0

Full Changelog: https://github.com/tangle-network/playproof/commits/v0.1.0