v0.7.1

Changes

• CI: fire tests + security on fix/** pushes @killerboyyy777 (#391)
• CI: Security scanning (CodeQL, dependency review, secrets) (#138) @killerboyyy777 (#354)
• Add AGENTS.md contributor notes @killerboyyy777 (#347)

🖥️ Server

• Milestone batch: remaining bughunt findings #374-378 + dashboard @killerboyyy777 (#393)
• Cross-platform start.py (7 subcommands, retire 7 bats) + torch done-key port @killerboyyy777 (#390)
• Spawn cap: one extra hostile instance per template (20->40) @killerboyyy777 (#384)
• [Hygiene] remove unused variable total in server.py @killerboyyy777 (#383)
• Bughunt batch 2: training integrity + #380 residuals @killerboyyy777 (#381)
• Bughunt batch 1: tonic brief, re-farm flag, bounty validation, save bool, TD dedup, prune guard @killerboyyy777 (#380)
• Standing buy orders: two-sided market books (#158) @killerboyyy777 (#372)
• [Scout] Reassign party leader when leader accepts new party invite @killerboyyy777 (#369)
• Convergence tuning: dungeon crown, trade score (#352), cooldown cap (#351) @killerboyyy777 (#368)
• [Scout] Fix duplicate equipped items shielding unequipped copies from death drops @killerboyyy777 (#363)
• Salvage #361: clear equip slots only when consumed item fully leaves inventory @killerboyyy777 (#362)
• Role-mix tuning: discovery nerf, death XP loss, risk-zone drops, gather scores (#334) @killerboyyy777 (#353)
• Bot roles: quester, crafter, party_leader + adaptive hint + dungeon cooldown (#335) @killerboyyy777 (#350)
• [Hygiene] remove unused os.path import in server.py @killerboyyy777 (#349)
• CI node harness + lint + requirements-train (dash smoke, browser proof) @killerboyyy777 (#344)
• Bughunt fix batch C: farm reconnect, server ping-timeout, immediate best snapshot @killerboyyy777 (#341)

🤖 ML env

• Milestone batch: remaining bughunt findings #374-378 + dashboard @killerboyyy777 (#393)
• Cross-platform start.py (7 subcommands, retire 7 bats) + torch done-key port @killerboyyy777 (#390)
• Bughunt batch 2: training integrity + #380 residuals @killerboyyy777 (#381)
• Bughunt batch 1: tonic brief, re-farm flag, bounty validation, save bool, TD dedup, prune guard @killerboyyy777 (#380)
• Standing buy orders: two-sided market books (#158) @killerboyyy777 (#372)
• Retire maker role: books survive without makers (V5) @killerboyyy777 (#371)
• [Hygiene] remove unused import random in ml/conductor/mixer.py @killerboyyy777 (#364)
• Deadlock fixes: stuck circuit-breaker, delver ready-gating, maker broke-hold (#357) @killerboyyy777 (#358)
• Role-count launcher: --roles SPEC with explicit per-role counts @killerboyyy777 (#356)
• Role-mix tuning: discovery nerf, death XP loss, risk-zone drops, gather scores (#334) @killerboyyy777 (#353)
• Bot roles: quester, crafter, party_leader + adaptive hint + dungeon cooldown (#335) @killerboyyy777 (#350)
• CI node harness + lint + requirements-train (dash smoke, browser proof) @killerboyyy777 (#344)
• Bughunt fix batch C: farm reconnect, server ping-timeout, immediate best snapshot @killerboyyy777 (#341)

🔥 Torch agents

• Milestone batch: remaining bughunt findings #374-378 + dashboard @killerboyyy777 (#393)
• Cross-platform start.py (7 subcommands, retire 7 bats) + torch done-key port @killerboyyy777 (#390)
• Bughunt batch 2: training integrity + #380 residuals @killerboyyy777 (#381)
• Bughunt batch 1: tonic brief, re-farm flag, bounty validation, save bool, TD dedup, prune guard @killerboyyy777 (#380)
• Quest shaping fix: per-mat intrinsics + TD turn-in reward (#340) @killerboyyy777 (#346)
• CI node harness + lint + requirements-train (dash smoke, browser proof) @killerboyyy777 (#344)
• Bughunt fix batch C: farm reconnect, server ping-timeout, immediate best snapshot @killerboyyy777 (#341)

🎛️ Conductor

• Milestone batch: remaining bughunt findings #374-378 + dashboard @killerboyyy777 (#393)
• [Hygiene] remove unused import random in ml/conductor/mixer.py @killerboyyy777 (#364)

📊 Dashboard

• Milestone batch: remaining bughunt findings #374-378 + dashboard @killerboyyy777 (#393)
• Standing buy orders: two-sided market books (#158) @killerboyyy777 (#372)
• Bind GM-tab treasury readout to live snapshot ticks + GM ack @killerboyyy777 (#366)
• Dashboard spawn-border fix: outdoor home gets accent dashed border @killerboyyy777 (#338)

🧪 Tests

• Milestone batch: remaining bughunt findings #374-378 + dashboard @killerboyyy777 (#393)
• Cross-platform start.py (7 subcommands, retire 7 bats) + torch done-key port @killerboyyy777 (#390)
• Bughunt batch 2: training integrity + #380 residuals @killerboyyy777 (#381)
• Bughunt batch 1: tonic brief, re-farm flag, bounty validation, save bool, TD dedup, prune guard @killerboyyy777 (#380)
• Standing buy orders: two-sided market books (#158) @killerboyyy777 (#372)
• Retire maker role: books survive without makers (V5) @killerboyyy777 (#371)
• [Scout] Reassign party leader when leader accepts new party invite @killerboyyy777 (#369)
• [Hygiene] Remove unused imports in test_dashboard.py @killerboyyy777 (#370)
• Bind GM-tab treasury readout to live snapshot ticks + GM ack @killerboyyy777 (#366)
• [Scout] Fix duplicate equipped items shielding unequipped copies from death drops @killerboyyy777 (#363)
• Salvage #361: clear equip slots only when consumed item fully leaves inventory @killerboyyy777 (#362)
• [Hygiene] Remove unused types import in test_server_unit.py @killerboyyy777 (#359)
• Deadlock fixes: stuck circuit-breaker, delver ready-gating, maker broke-hold (#357) @killerboyyy777 (#358)
• Role-count launcher: --roles SPEC with explicit per-role counts @killerboyyy777 (#356)
• Live-suite flake hardening: dungeon-enter + party-race asserts (#342) @killerboyyy777 (#355)
• Role-mix tuning: discovery nerf, death XP loss, risk-zone drops, gather scores (#334) @killerboyyy777 (#353)
• Bot roles: quester, crafter, party_leader + adaptive hint + dungeon cooldown (#335) @killerboyyy777 (#350)
• Add behavioral unit tests for snapshot additions @killerboyyy777 (#345)
• CI node harness + lint + requirements-train (dash smoke, browser proof) @killerboyyy777 (#344)
• Dashboard spawn-border fix: outdoor home gets accent dashed border @killerboyyy777 (#338)

📚 Docs

• Milestone batch: remaining bughunt findings #374-378 + dashboard @killerboyyy777 (#393)
• Cross-platform start.py (7 subcommands, retire 7 bats) + torch done-key port @killerboyyy777 (#390)
• Spawn cap: one extra hostile instance per template (20->40) @killerboyyy777 (#384)
• Bughunt batch 2: training integrity + #380 residuals @killerboyyy777 (#381)
• Bughunt batch 1: tonic brief, re-farm flag, bounty validation, save bool, TD dedup, prune guard @killerboyyy777 (#380)
• Standing buy orders: two-sided market books (#158) @killerboyyy777 (#372)
• Retire maker role: books survive without makers (V5) @killerboyyy777 (#371)
• Convergence tuning: dungeon crown, trade score (#352), cooldown cap (#351) @killerboyyy777 (#368)
• Salvage #361: clear equip slots only when consumed item fully leaves inventory @killerboyyy777 (#362)
• Deadlock fixes: stuck circuit-breaker, delver ready-gating, maker broke-hold (#357) @killerboyyy777 (#358)
• Role-mix tuning: discovery nerf, death XP loss, risk-zone drops, gather scores (#334) @killerboyyy777 (#353)

What's Changed

• Dashboard spawn-border fix: outdoor home gets accent dashed border by @killerboyyy777 in #338
• Bump release-drafter/release-drafter from 6 to 7 by @dependabot[bot] in #339
• Bughunt fix batch C: farm reconnect, server ping-timeout, immediate best snapshot by @killerboyyy777 in #341
• CI node harness + lint + requirements-train (dash smoke, browser proof) by @killerboyyy777 in #344
• Add behavioral unit tests for snapshot additions by @killerboyyy777 in #345
• Add AGENTS.md contributor notes by @killerboyyy777 in #347
• Quest shaping fix: per-mat intrinsics + TD turn-in reward (#340) by @killerboyyy777 in #346
• [Hygiene] remove unused os.path import in server.py by @killerboyyy777 in #349
• Bot roles: quester, crafter, party_leader + adaptive hint + dungeon cooldown (#335) by @killerboyyy777 in #350
• Role-mix tuning: discovery nerf, death XP loss, risk-zone drops, gather scores (#334) by @killerboyyy777 in #353
• CI: Security scanning (CodeQL, dependency review, secrets) (#138) by @killerboyyy777 in #354
• Live-suite flake hardening: dungeon-enter + party-race asserts (#342) by @killerboyyy777 in #355
• Role-count launcher: --roles SPEC with explicit per-role counts by @killerboyyy777 in #356
• Deadlock fixes: stuck circuit-breaker, delver ready-gating, maker broke-hold (#357) by @killerboyyy777 in #358
• [Hygiene] Remove unused types import in test_server_unit.py by @killerboyyy777 in #359
• Salvage #361: clear equip slots only when consumed item fully leaves inventory by @killerboyyy777 in #362
• [Hygiene] remove unused import random in ml/conductor/mixer.py by @killerboyyy777 in #364
• [Scout] Fix duplicate equipped items shielding unequipped copies from death drops by @k...

v0.7.0

Changes

• /oc hardening: actor gate on trial, OWNER/MEMBER gate on review @killerboyyy777 (#332)
• Bughunt batch B: tonic herb truth + emoji removal (dashboard) @killerboyyy777 (#328)
• Docs release-audit: labeler comment + CONTRIBUTING reset/resume @killerboyyy777 (#326)
• Add opencode review-only workflow @killerboyyy777 (#295)
• Trial trigger tighten: startsWith /oc-space @killerboyyy777 (#294)
• Trial trigger rename: /oc-trial -> /oc @killerboyyy777 (#283)
• Fix trial workflow: GITHUB_TOKEN via env @killerboyyy777 (#281)
• Tweak pre-commit comment: fast test scripts @killerboyyy777 (#197)
• CI: run test_dashboard.py offline + pre-commit @killerboyyy777 (#163)
• CI: run test_env_reset + test_eval_stats @killerboyyy777 (#159)
• Release versioning from labels + docs @killerboyyy777 (#150)

🖥️ Server

• Dashboard revamp 4/4: promote dashboard2 to canonical dashboard (#207) @killerboyyy777 (#331)
• Bughunt fix batch A: gold persistence, party message, split rounding, exact recon, release gate @killerboyyy777 (#329)
• Dashboard revamp 3/4: Agents, Quests, Crafting, Config-read tabs (#209) @killerboyyy777 (#321)
• Dashboard revamp 1/4: parallel /v2 (#205) @killerboyyy777 (#276)
• Bughunt batch: env/conductor/torch/dashboard correctness fixes @killerboyyy777 (#257)
• Revert unreviewed dashboard code from #279 @killerboyyy777 (#282)
• Add opencode explain-only trial workflow @killerboyyy777 (#279)
• Docs audit fixes (no behavior change) @killerboyyy777 (#268)
• Only living NPCs retaliate (fixes #265) @killerboyyy777 (#269)
• Starting purse: 10g for brand-new characters (#258) @killerboyyy777 (#259)
• New unit suites: training loops, conductor run, GM treasury (+ gm_reward validate-before-spend) @killerboyyy777 (#252)
• Combat/party/dungeon balance + race pass (#195) @killerboyyy777 (#249)
• Quest-giver immunity to player attacks (#193) @killerboyyy777 (#219)
• Soak readiness: --config flag, commissioner slot, TTL overlay, snapshot report @killerboyyy777 (#201)
• Implement #191 env desync fixes + #192 server follow-ups @killerboyyy777 (#200)
• Drop inaccurate parenthetical in collusion comment @killerboyyy777 (#199)
• Review follow-ups: one-line handler errors, commission econ fixes, world validator @killerboyyy777 (#198)
• P0 exploits closed: wash-trade reject, XP cap + kill consumption, input hardening @killerboyyy777 (#196)
• Remove misleading/restating comments @killerboyyy777 (#182)
• #173 server bind flags; #166 scores backup rotation + recovery @killerboyyy777 (#179)
• #128 dashboard Phase 2: overview polish, canvas map v2, room drawer @killerboyyy777 (#161)
• #69 eval stats + run records; #72 protocol version handshake @killerboyyy777 (#151)

🤖 ML env

• Bughunt fix batch A: gold persistence, party message, split rounding, exact recon, release gate @killerboyyy777 (#329)
• Reset CLI plus reconnect handling (#290) @killerboyyy777 (#303)
• Supervisor per-step learn hook (throttled) (#292) @killerboyyy777 (#301)
• Stable role_I character names persisted in registry (#291) @killerboyyy777 (#299)
• Goal-weighted PBT fitness from per-step reward vectors (#288) @killerboyyy777 (#296)
• Lineage checkpoints plus registry resume default (#293) @killerboyyy777 (#300)
• Bughunt batch: env/conductor/torch/dashboard correctness fixes @killerboyyy777 (#257)
• Collision-proof churn spawn ids plus retry loop (#277) @killerboyyy777 (#278)
• #162 Phase 1: reward vector + inheritable goals (infrastructure, no divergence yet) @killerboyyy777 (#271)
• Docs audit fixes (no behavior change) @killerboyyy777 (#268)
• Churn spawn-to-target with per-tick cap (#214) @killerboyyy777 (#261)
• Combat/party/dungeon balance + race pass (#195) @killerboyyy777 (#249)
• Env expressiveness (#194) + remedy/tonic representation (#183) @killerboyyy777 (#216)
• Soak --arrivals default 2.0 -> 5.0/min (part 1 of #214) @killerboyyy777 (#215)
• Wire log_episode: supervisor logs every completed episode (#210) @killerboyyy777 (#212)
• Docs-gap fixes: CHANGELOG #201 entry, commissioner in choices/help/registry test @killerboyyy777 (#202)
• Soak readiness: --config flag, commissioner slot, TTL overlay, snapshot report @killerboyyy777 (#201)
• Implement #191 env desync fixes + #192 server follow-ups @killerboyyy777 (#200)
• Remove misleading/restating comments @killerboyyy777 (#182)
• Review nits: CHANGELOG coverage, drop unreachable dims note, prune scaling note @killerboyyy777 (#181)
• Review follow-ups: gitignore backups, tick prune, shared versioning, wave baseline, SCORES restore @killerboyyy777 (#180)
• Soak lifecycle fixes: episode lifetimes, differential waves, corpse prune, spawn hardening, linear versioning @killerboyyy777 (#175)
• Fix drift: torch_farm quest constants; metrics rotation caps @killerboyyy777 (#165)
• Hygiene: ruff/black clean in plugin files, CHANGELOG infra batch @killerboyyy777 (#164)
• #62/#152 agent plugins: registry, built-ins, weighted conductor slots @killerboyyy777 (#155)
• #69 eval stats + run records; #72 protocol version handshake @killerboyyy777 (#151)

🔥 Torch agents

• Bughunt fix batch A: gold persistence, party message, split rounding, exact recon, release gate @killerboyyy777 (#329)
• Bughunt batch: env/conductor/torch/dashboard correctness fixes @killerboyyy777 (#257)
• Docs audit fixes (no behavior change) @killerboyyy777 (#268)
• Env expressiveness (#194) + remedy/tonic representation (#183) @killerboyyy777 (#216)
• Remove misleading/restating comments @killerboyyy777 (#182)
• Review follow-ups: gitignore backups, tick prune, shared versioning, wave baseline, SCORES restore @killerboyyy777 (#180)
• Remove redundant torch launchers, link torch_farm in start.bat @killerboyyy777 (#178)
• Fix drift: torch_farm quest constants; metrics rotation caps @killerboyyy777 (#165)
• #69 eval stats + run records; #72 protocol version handshake @killerboyyy777 (#151)

🎛️ Conductor

• Bughunt fix batch A: gold persistence, party message, split rounding, exact recon, release gate @killerboyyy777 (#329)
• Reset CLI plus reconnect handling (#290) @killerboyyy777 (#303)
• Supervisor per-step learn hook (throttled) (#292) @killerboyyy777 (#301)
• Stable role_I character names persisted in registry (#291) @killerboyyy777 (#299)
• Goal-weighted PBT fitness from per-step reward vectors (#288) @killerboyyy777 (#296)
• Lineage checkpoints plus registry resume default (#293) @killerboyyy777 (#300)
• Bughunt batch: env/conductor/torch/dashboard correctness fixes @killerboyyy777 (#257)
• Collision-proof churn spawn ids plus retry loop (#277) @killerboyyy777 (#278)
• #162 Phase 1: reward vector + inheritable goals (infrastructure, no divergence yet) @killerboyyy777 (#271)
• Churn spawn-to-target with per-tick cap (#214) @killerboyyy777 (#261)
• Soak --arrivals default 2.0 -> 5.0/min (part 1 of #214) @killerboyyy777 (#215)
• Wire log_episode: supervisor logs every completed episode (#210) @killerboyyy777 (#212)
• Docs-gap fixes: CHANGELOG #201 entry, commissioner in choices/help/registry test @killerboyyy777 (#202)
• Soak readiness: --config flag, commissioner slot, TTL overlay, snapshot report @killerboyyy777 (#201)
• Remove misleading/restating comments @killerboyyy777 (#182)
• Review nits: CHANGELOG coverage, drop unreachable dims note, prune scaling note @killerboyyy777 (#181)
• Review follow-ups: gitignore backups, tick prune, shared versioning, wave baseline, SCORES restore @killerboyyy777 (#180)
• Soak lifecycle fixes: episode lifetimes, differential waves, corpse prune, spawn hardening, linear versioning @killerboyyy777 (#175)
• Fix drift: torch_farm quest constants; metrics rotation caps @killerboyyy777 (#165)
• Hygiene: ruff/black clean in plugin files, CHANGELOG infra batch @killerboyyy777 (#164)
• #62/#152 agent plugins: registry, built-ins, weighted conductor slots @killerboyyy777 (#155)

📊 Dashboard

• Dashboard revamp 4/4: promote dashboard2 to canonical dashboard (#207) @killerboyyy777 (#331)
• Dashboard hover tooltips for metrics and GM config (#76) @killerboyyy777 (#286)
• Bughunt batch: env/conductor/torch/dashboard correctness fixes @killerboyyy777 (#257)
• Fix dashboard renderAll dropping the snapshot @killerboyyy777 (#211)
• #128 dashboard Phase 2: overview polish, canvas map v2, room drawer @killerboyyy777 (#161)
• #127 dashboard foundation: Overview tab, placeholder tabs, per-tab rendering @killerboyyy777 (#160)

🧪 Tests

• Dashboard revamp 4/4: promote dashboard2 to canonical dashboard (#207) @killerboyyy777 (#331)
• Dashboard revamp 3/4: Agents, Quests, Crafting, Config-read tabs (#209) @killerboyyy777 (#321)
• Reset CLI plus reconnect handling (#290) @killerboyyy777 (#303)
• Flake guard: heal both envs before shared dungeon entry (#274) @killerboyyy777 (#302)
• Supervisor per-step learn hook (throttled) (#292) @killerboyyy777 (#301)
• Stable role_I character names persisted in registry (#291) @killerboyyy777 (#299)
• Spawn-goal distribution + offspring proof (#289) @killerboyyy777 (#297)
• Goal-weighted PBT fitness from per-step reward vectors (#288) @killerboyyy777 (#296)
• Lineage checkpoints plus registry resume default (#293) @killerboyyy777 (#300)
• Dashboard revamp 2/4: Market, Dungeons, GM port (#206) @killerboyyy777 (#298)
• Account pre-reward heals in live treasury assert (#285) @killerboyyy777 (#287)
• Dashboard hover tooltips for metrics and GM config (#76) @killerboyyy777 (#286)
• Fix presence obs flags keyed backwards, always zero (#263) @killerboyyy777 (#284)
• Dashboard revamp 1/4: parallel /v2 (#205) @killerboyyy777 (#276)
• Bughunt batch: env/conductor/torch/dashboard correctness fixes @killerboyyy777 (#257)
• Harden test_live dungeon entries against graveyard idle-death flake (#274) @killerboyyy777 (#275)
• Collision-proof churn spawn ids plus retry loop (#277) @killerboyyy777 (#278)
• Revert unreviewed dashboard code from #279 @killerboyyy777 (#282)
• Add opencode explain-only trial workflow @killerboyyy777 (#279)
• CI: run tests on agent/** pushes @kille...

v0.6.0

What's Changed

• #89 dependabot automerge + protection script fix by @killerboyyy777 in #133
• Docs match live ruleset by @killerboyyy777 in #134
• Bump actions/checkout from 4 to 7 by @dependabot[bot] in #124
• Bump actions/upload-artifact from 4 to 7 by @dependabot[bot] in #126
• Bump softprops/action-gh-release from 2 to 3 by @dependabot[bot] in #125
• P0 #135 death metrics + docs: slim README, wiki content by @killerboyyy777 in #145
• Canonical Apache-2.0 LICENSE (fix detection) + README license section by @killerboyyy777 in #146
• Docs audit: protection fix, changelog coverage by @killerboyyy777 in #147
• CONTRIBUTING: docs ship with the change by @killerboyyy777 in #148
• Release prep: version changelog as 0.6 by @killerboyyy777 in #149

New Contributors

• @killerboyyy777 made their first contribution in #133
• @dependabot[bot] made their first contribution in #124

Full Changelog: https://github.com/killerboyyy777/Machine-Learning-mmo/commits/v0.6.0