Progress Dashboard

Progress Dashboard

This is the shared human + AI interface for seeing what Codex, Claude and Codex sub-agents are doing right now.

Keep this page open when parallel documentation work is running. It links to the live board, event feed and compact JSON status files without forcing readers through the whole sidebar.

What this page is

• Shared source-of-truth for lane state, active owners and recently validated work.
• Human + AI handoff surface for current ownership and next action.
• Quick index for where to read next when a lane claims changes.

Where this dashboard lives

• Wiki page: docs/wiki/Progress-Dashboard.md
• Runtime state files: agent-status.json, agent-events.jsonl, agent-knowledge.jsonl
• Coordination board: Coordination-Board.md
• Source of truth pointers: agent-context.json

How to use this page

1. Start with At A Glance to find active lanes and owners.
2. Read Latest Batch for the newest validated status update only.
3. Use Current Lanes to separate active docs work, watchlists and future code-owner work.
4. Use Agent worklog, Discovery swarm and agent-events.jsonl for historic scout detail.

Latest Batch

Lane	Status	Output
mini-scout-wave-depth-leftovers-2026-06-04	Published / validated / pushed	Steff asked to run more Spark scouts. Three gpt-5.3-codex-spark starts hit the active quota before evidence, so Codex closed the dead slots and launched six read-only gpt-5.4-mini scouts instead. All six returned and were closed. Codex promoted source-checked deltas into AI/order, HC failover, gear templates, PerformanceAuditAnalyzer, server ops, tooling readiness and authority-migration owner pages. Ampere's smd_sahrani_a2 mission.sqm correction was rejected by source check because the tracked stub still lacks mission.sqm, description.ext and initJIPCompatible.sqf. Validation passed and content pushed as docs d02f3709 / wiki 7be237f. No gameplay source changed.
mini-scout-wave-community-config-ai-ui-tooling	Published / validated / pushed	Steff asked for more Spark scouts. Three gpt-5.3-codex-spark starts hit quota/overflow before evidence, so Codex closed those slots and launched six read-only gpt-5.4-mini scouts. One feature-status scout overflowed without evidence; five returned. Community/dev provenance confirmed the existing Community & Dev hub and contributor/upstream-remotes evidence. Config/data-model scouting added positional-array and post-load mutation guidance to Assets/config/localization/parameters. UI scouting sharpened Player UI workflow for buy-gear views/tabs, command-menu map-click flow, help-menu text ownership and respawn discovery. AI/runtime scouting added the spawned-unit follow-up, player-AI watchdog/recovery and AI leader respawn branch split to AI runtime/HC loop map. Tooling scouting tightened PerformanceAuditAnalyzer output wording. Validation passed; content commits are docs b5a40dd4 / wiki 5699c82, and final pushed heads after reconciliation are docs b2a33c70 / wiki 741e21a. No gameplay source changed.
fallback-mini-scout-wave-feature-lifecycle-economy-pv-ui-tooling	Harvested / validated / pushed	Steff asked for more Spark scouts, but six gpt-5.3-codex-spark starts hit quota until 23:53 before returning evidence. Codex closed those slots and launched six read-only gpt-5.4-mini fallback scouts. Bohr mostly confirmed already-canonical Feature Status/dead-chain rows, including RU para-ammo, AntiStack, WASP wheel repair, unitCaching and resistance patrol. Laplace confirmed UI findings are already mostly routed, but _Sidebar lacked a click-through UI cluster, so Codex added one. Descartes found a real stale PV inventory sentence in SQF code atlas and Wiki source consistency; Codex corrected those. Descartes' version.sqf correction was rejected by source check because initJIPCompatible.sqf does include version.sqf and separately runs Common/Init/Init_Version.sqf. Huygens' economy scout produced a compact Gameplay systems constants table, Hubble sharpened Networking/PV transport-vs-authority wording, and Kuhn's tooling scout led to a dedicated PerformanceAuditAnalyzer page plus tool/testing links. Validation passed; content commits are docs b878f7e4 / wiki bad8646, and final pushed heads after the breadcrumb are docs b7f03baf / wiki 14a3a2d. No gameplay source changed.
static-reference-triage-lesson	Published / validated / pushed	Codex source-checked the missing-reference scout's remaining lead and clarified that the Source inventory static-reference table mixes live paths with commented/dead archaeology. description.ext:37 unitCaching, Init_Towns.sqf:168,174 respatrol.fsm, and WASP/Init_Client.sqf:12 KeyDown.sqf are commented/absent references, while car_wheel_new.sqf is inactive because its action caller and WASP_procInitComm bootstrap are commented. Added Development lessons and agent-development-lessons.jsonl so future sweeps do executable-line checks before promoting missing references to release gates. Validation passed and content pushed as docs a391a0f1 / wiki 9e08393. No gameplay source changed.
spark-scout-wave-version-lifecycle-tooling-feature-pv-caps	Published / validated / pushed; scouts closed	Steff asked for another bunch of Spark scouts. Codex launched six gpt-5.3-codex-spark read-only scouts for generated version.sqf release gating, mission lifecycle/init graph, tooling/integrations, Feature Status depth, PV/PVF authority and player-role/AI-cap formulas; the UI scout failed from output overflow and the player AI-cap scout failed from long-thread compaction. Harvested reports: Peirce confirmed the ignored/generated version.sqf release hazard, so Codex added versionSqfGeneratedInput to agent-release-readiness.json and promoted hard root-gate wording into Tools/build, Source fix propagation, Mission config/version and Source inventory. Averroes confirmed lifecycle docs are mostly strong; Codex added architecture/entrypoint/SQF-atlas callouts for HC timing, JIP stall risk, mixed Common ownership and init hygiene while keeping the source-verified town object count at 40 after recheck. Nash confirmed the PV/PVF index has no undisclosed major channels and that AFK, supply polling, MASH marker, server-FPS compatibility and player-object-list edges are already routed in canonical networking/feature/backlog pages. Bacon's Feature Status pass added missing-file clarity for old map blinking and legacy AT/bomb hooks, plus a supply-mission dead-twin compile caution. Hilbert's tooling report was harvested into machine-readable toolingAuditGates for LoadoutManager packaging/copy risk, PerformanceAuditAnalyzer headless/large-log limits, DiscordBot/Extension schema/config risk and BattlEye bundle incompleteness. Fresh commander/economy mostly confirmed canonical docs; its claim that commander reassignment was patched was rejected by local source recheck (Server_AssignNewCommander.sqf still sets _side = _this). Fresh server-ops/integrations and missing-reference scouts confirmed already-canonical ops/dead-feature coverage; the missing-reference scout's useful supplyMissionActive dead-twin caution was already promoted. Fresh construction/factories, UI and respawn/MASH overflowed before evidence and were closed. No gameplay source changed.
dashboard-open-item-cleanup	Published / validated	Codex compacted this dashboard back into a live status surface, moved old batch detail behind worklog/swarm links, closed the Coverage Ledger sidebar request, and reclassified stale open/ready rows as published, watchlist or future code-owner work. No gameplay source changed.
spark-scout-wave-release-targets-hc-ai-caps-economy	Published / validated / pushed	Steff asked for more Spark scouts. Codex launched six GPT-5.3-Codex-Spark read-only scouts for player AI caps, support authority, UI/dialog hazards, generated/modded release tiers, town/economy/supply edge cases and HC/locality/FPS, then refilled one freed slot with tooling/integrations and one with mission lifecycle/init. Returned reports were duplicate-aware: Player AI caps confirmed the existing Discord-ready table, HC scout confirmed existing failover/dual-FPS/static-defense gaps, town/economy scout confirmed already-canonical camp flag, patrol latch, side-supply, income-cap and supply-authority rows, and generated/modded scout sharpened release target counts. The content delta is focused in agent-release-readiness.json and Tooling release readiness audit: release targets are now split into source Chernarus, maintained Takistan, branch-only Zargabad candidate and per-folder modded blocked/stub tiers with file counts, conflict-marker counts and bootstrap blockers. UI scout also added the missing Client\Images\wf_*.paa old-upgrade-dialog asset references to the UI atlas/checklist. Validation passed and content pushed as docs 058427d3 / wiki 4b72809. No gameplay source changed.
ai-commander-branch-head-refresh-and-spark-scout-wave	Published / validated	Current branch evidence is recorded: origin/feat/ai-commander is c20ce153, source-Chernarus-only with 9 files, +416/-5 and no maintained Vanilla changes. The post-4dba060e cleanup series only rewrites five Server/AI/Commander/AI_Commander*.sqf scripts. origin/codex/quad-ai-commander is d4e0fa38 concept/report evidence only. Returned scout notes remain follow-up leads until source-checked into canonical pages.
fresh-background-spark-scout-wave-buy-menu-harvest	Published / validated / pushed	Ten fresh background Spark threads were launched after normal subagent startup failed. Five completed reports mostly confirmed existing canonical docs, four were interrupted and one UI thread errored. The only promoted delta was the buy-menu driver-default profileNamespace key split, routed through Factory/purchase, Development lessons, agent-development-lessons.jsonl and agent-feature-status.jsonl.

Historic batch rows were intentionally removed from this page. Use Agent worklog, Discovery swarm and agent-events.jsonl for the long audit trail.

At A Glance

Actor	Status	Current lane	What to expect next
Codex	Active	documentation-finisher-loop	Depth leftovers mini scout wave is published; continue the docs finisher loop from the next highest-value source-backed lane.
Codex-2	Ready	None	Pick a bounded source-backed lane from PVF dispatcher lookup, side-supply clamp first, commander reassignment call-shape repair or remaining supply authority hardening.
Claude	Autonomous-ready	collaboration-follow-autonomous-ready	Coverage Ledger navigation is wired. Claude can self-select the next bounded source-backed review from the ledger or hardening backlog.
Sub-agents	Returned / closed	mini-scout-wave-depth-leftovers-2026-06-04	Six gpt-5.4-mini scouts returned. Useful findings are source-checked into owner pages; duplicate and false leads remain non-canonical.
Shared docs	Live	GitHub wiki + docs mirror	Wiki and mirror should stay in parity after scoped validation; use event logs and git history for commit IDs.

One-Link Check

Need	Link
Human progress page	Progress dashboard
Sub-agent swarm board	Discovery swarm
Detailed coordination page	Coordination board
Machine progress file	agent-status.json
Active lanes and ownership	agent-collaboration.json
Agent-readable knowledge records	agent-knowledge.jsonl
Agent-readable hardening backlog	agent-hardening-backlog.jsonl
Latest event stream	agent-events.jsonl
Dated narrative notes	Agent worklog
External report intake	External research reports
External report manifest	external-research-report-manifest.json

Current Lanes

Lane	Owner	Status	Next action
documentation-finisher-loop	Codex	Active / ongoing	Keep the wiki current from source evidence, update machine files on visible status changes, and keep this dashboard compact.
autonomous-claude-research	Claude	Autonomous-ready	Self-select the next bounded source-backed review from Codebase coverage ledger when Claude is active.
feature-status-reconciliation	Codex / future agent	Watchlist	Fold newly confirmed findings into Feature status, owner pages and machine records. No untriaged finding is blocking this dashboard.
implementation-hardening-from-backlog	Future code owner	Owner decision / code lane	Pick implementation work from agent-hardening-backlog.jsonl only when Steff asks for gameplay patches or a code owner claims the work.
testing-debugging-release-workflow	Codex / future tester	Published release gate	Use Testing workflow to distinguish source review, propagated source fixes and actual Arma smoke evidence.
source-propagated-smoke-pending	Future tester / release owner	Smoke pending	Client skill init, hosted FPS, supply scan and paratrooper markers are tracked in owner pages and Feature Status; they are not dashboard cleanup blockers.
wasp-marker-wait-cleanup	Future code owner	Source needs code	WASP marker wait cleanup remains a tiny patch-ready opportunity, not an active docs lane.

Recently Closed / Reclassified Open Items

• Coverage Ledger navigation request: Home, _Footer.md, llms.txt and agent-context.json already linked it; _Sidebar.md now links it too, and the Coordination Board mailbox is marked done.
• OA object-lifecycle command addendum: the old Codex-2 active claim is closed into Arma 2 OA command version reference, covering createVehicle, createUnit, deleteVehicle, hideObject / enableSimulation and the A3-only *Global variants.
• Old ready-for-review rows for PVF dispatch, victory/endgame and factory/purchase are no longer dashboard blockers. Their findings are published and integrated; next work is code-owner implementation or validation.
• Historic scout waves are closed, harvested or non-evidence until relaunched. Their detail belongs in Discovery swarm, Agent worklog and agent-events.jsonl, not in this live dashboard.
• Standing future work remains visible as watchlist/code-owner lanes instead of generic open dashboard spam.

Recent Published Work

Batch	Output	Details
AI commander branch-head refresh	AI commander audit, Current source snapshot, Quad AI Commander	Refreshed branch heads and kept branch/design evidence separate from shipped source.
HC upstream history	HC upstream history and lessons	Captured older headless-client branch/comment-message lessons for future HC work.
PR8 and drone lesson match	PR8 and Drone upstream lesson match	Mapped upstream lessons to PR #8 and drone-branch review gates.
Quad AI/support/lifecycle/UI/tooling scout harvest	Quad AI Commander, Support specials, Tools/build	Promoted source-backed non-duplicate deltas and left concept-only evidence clearly labeled.
Config/factory/UI/AI/cleanup/upstream scout harvest	Mission config graph, Factory/purchase, Client UI systems	Added version/runtime contract, latent AIBuyUnit wording, UI cleanup debt and loop-topology notes.
Town/respawn/upstream/performance scout harvest	Gameplay systems, Respawn/death, Performance sweep	Added respawn penalty, resource cap, capture reward and loop-topology refinements.
Commander vote/reassignment playbook	Commander vote/reassignment	Routed commander UI/PV/authority pages through one patch-ready playbook.
Economy supply commander audit	Economy authority first cut, Supply authority cleanup	Completed source-backed economy/commander/supply boundary map and handed implementation choices to code owners.

Update Ritual

Moment	Codex / Claude action	Human-visible result
Starting work	Update agent-collaboration.json and append a claim event.	This dashboard and the coordination board show who owns what.
Still working after a long pass	Append a heartbeat event.	You can see that an agent is alive and where it is reading.
Finding a source-backed issue	Append a finding event and add a short worklog note.	The finding is visible before the full page is integrated.
Finishing a lane	Append a complete event, update affected pages and note validation status.	The lane moves from active to ready/integrated.
Handing off	Append a handoff event with the exact next action.	The next agent can pick it up without chat memory.

Status Legend

Status	Meaning
active	Someone is working this lane now.
watchlist	Keep synchronized when new evidence appears; no standalone action is currently blocked.
owner decision / code lane	Documentation is ready; a gameplay/release owner must choose or implement.
published / validated	Published into the wiki/docs set and reflected in navigation/context.
blocked	Needs user input, missing access or an external state change.

For Agents

1. Load agent-status.json, agent-collaboration.json, agent-knowledge.jsonl, agent-events.jsonl and Agent collaboration protocol before claiming work.
2. Treat agent-events.jsonl as append-only.
3. Keep claims bounded by source scope.
4. Never overwrite another agent's page or navigation changes from a stale branch.
5. Publish human notes to Agent worklog and machine notes to the JSON files.

Continue Reading

Previous: Agent context | Next: Coordination board

Main map: Home | Fast path: Quickstart | Agent file: agent-context.json