Repository navigation
v4.0.1
4.0.1 is a critical fix. The run token and cost budgets shipped incomplete in 4.0.0: usage could not be reported once a run's members were terminal, and the budget measured the last turn's figure instead of the run's spend. They are removed (see Removed); a client that still sends a removed field gets unknown_parameter. There is no migration, so a rollback to 4.0.0 is safe.
Added
runs_readfrontierreports cancelled members and the in-flight items' titles.cancelledcounts the terminal members that carry the status label ofcancelunder the run's root (the one label policy: per-rootstatus_labels, then the server's, then the built-incancelled); they stay counted incounts.terminal. EachinFlightentry gains its item'stitle. Both are additive, on MCP and onPOST /api/v1/ops/runs_read.
Plugin
- Fixed: run-wave post-run reports name the pause decision and count open decisions.
report --event pausedtakes--paused <file>, a savedrun_pausedanswer (a write answer, a top-levelerroror a bare{code, detail}), and prints itspauseDecisionIdandpauseReason; the run view never carried them. A--pausedfile with norun_pausederror exits 2. The closed report reads the close snapshot'sopenDecisionsas the integer the server sends. A reviewer assign that hitsrun_at_capacitybacks off instead of routing to Step 5a, and the plugin prompts,ralph-loopand the run-wave, director and schema-workflow docs drop the removedpaused: {decisionId}write-success member and theamountanswer field. The work-summary runs section and the api-reference report row no longer describe run budgets. - Changed: the director guard covers host MCP tools, the web tools and the remaining body reads. Under the director marker, every
mcp__tool whose name lackstask-orchestrator(terminal, file-store and computer-use servers, or a Task Orchestrator server registered under a key without that word) is denied unless its bare name isstart_session,send_message,update_plan,wrap_up,mark_chapter,get_settingsorset_session_output_style, through a third guard registration (hostToolMatcher()).WebFetch,WebSearch,ReadMcpResourceToolandReadMcpResourceDirToolare denied;configis denied andruns_readis allowed only forget,list,frontier,decisionsanddecision;Skillruns only the director skill andtask-orchestrator:model-effort;gh pr viewandgh pr listtake only operands,-R, listing filters and--jsonwith status, metadata andbodyfields (no comments or reviews); aWorkflowcall carryingscriptPathorresumeFromRunIdis denied, so a tier-2 resume goes to the coordinator. Call and docs updated in the director and run-wave skills and the hooks README. - Removed: every usage-report path is removed with the server's heartbeat usage and run token and cost budgets.
run-wavepost-run no longer writesusage.jsonor sends a usage heartbeat (Step 0a and the Step 7 heartbeat are gone),run-ops --phase heartbeatand--usageare removed, andrun-ops --phase opensends onlyattemptsPerSeatandmaxActiveSeatsas the runbudget.planwarns once per key when a profile setsrunBudget.tokensorrunBudget.costUsd(ignored;maxActiveSeatsstill applies).ralph-loopdrops--run-budget-tokensand--run-budget-usd(now exit 64, like--actor) and no longer reports usage before thecompletetransition; the ralph, run-wave, director, init and manage-schemas docs no longer describe token or cost budgets (#677). - Fixed: the
orchestration-modeband shows only this session's runs, in every mode. It drew every open and paused run on the server, and counted every session's decisions, so a session with nothing of its own showed another session's run, even one from another project. It now draws only the run the session coordinates and the runs it or one of its subagents (a background coordinator, a Workflow, a seat) opened or acted on throughruns_write, and no longer reads the server-wide decisions list. Touching or reading an item claims no run, so another session's run over the same feature stays out. Only a write moves the focus: anitems_readnaming an item (a status check, a look at another session's work) no longer makes it the session's focus, and a focus a read set is dropped on the next load. Director mode follows the same rule: the band is the session's own graph, not the board. - Fixed: the
orchestration-modeband no longer reports a stalled coordinator for a Workflow-driven run. The alert follows the director skill's rule: nothing while a Workflow runs the run (hints.methodA,hints.phaselaunched), while any seat is live, or on a paused run. Before, any Workflow wave longer than the stall threshold showedCOORD STALLEDuntil the post-run step. - Changed: the
orchestration-modeband speaks plain words and shows each agent. The title row carries the bar on its right: one cell per member, green done, a red⧄cancelled, red stuck, amber waiting on a decision, grey finished and waiting to be advanced, hollow not started, and each member an agent works on in a colour of its own that it keeps from stage to stage, blinking. The status row puts what needs a look first (1 agent stuck,Coordinator not responding,1 decision for you,Run paused,All agents finished · advance the items), then what the agents do (3 agents implementing), with the members by phase in plain words on the right (3 planned · 2 in progress · 1 done); the run ids,ready/waitingandq → w → r → ✓are gone. Below it, one row per agent (its marker in the member's colour, blinking with its bar cell, then its seat and the item's title; at most three, then+N more), else the item the session last touched. The blink runs in aClienton the surface's own clock, one timer for the whole band, with no extra server reads; VS Code and the mobile app draw the band still. A short band drops rows from the bottom, never the title. The bar and the counts never shrink: a long title gives way (on the desktop app it wraps onto a second line) instead of squeezing the bar. - Fixed: the
orchestration-modeband's focus stops below a backlog bucket. With no run, the band showed the ancestor just below the project root of the item the session touched, so a task in a bucket such asBugsdrew the bucket's whole backlog, and counted the bucket itself as in progress. The walk now stops below a grouping item (typedcontainerorproject, or taggedcontainer), so the band shows the task's own feature; a bucket the session touched itself counts its work, not itself. The stored focus is now a cache of the touched item's walk, walked again once per load, so a focus stored before a reload or a resume follows the current rule. - Fixed: run-wave provenance records each seat's transcript meters instead of the run notification figure.
run-planner provenancenow reads every seat's input, cache write, cache read and output meters fromdispatch-metafor both methods and writes one measure on every line:tokensis the non-cached sum, withmeters=andseat-meters=tails and six-fieldextra-seatsrows carrying the four meters.see:<short>and the Workflow notification figure are gone,durationreadsunknown, a seat without a transcript readsunknownwith a stderr warning (never zero),--usageis ignored with a warning, anddispatch-metaalso matchesplanner-a<n>andreviewer-a<n>labels. Notes written before this change still parse. - Fixed: run-wave seats that own
implementation-notesorsession-trackingnow receive theclosure-notesrule. The run planner derived a stage's rules only from theskill:of the notes it owns, and no note namesclosure-notes, so a generic workflow seat (nodispatch.agent) never got the rule.deriveRulesSkillsnow adds it once whenever the server serves it (#676). - Fixed: the run-wave fix-cycle reseat prompt tells the seat to rewrite its notes in full. A fix-cycle seat (
-a<n>) kept the first seat's banned sections and appended a per-cycle narrative to its notes. TheFIX CYCLEline now ends with a rewrite-in-full instruction for each ofimplementation-notesandtest-manifestthe base stage owns: replace superseded text, drop the sections theclosure-notesrule bans, add no per-cycle section, and keep history to a finalAmendments:list. - Bumped plugin versions:
task-orchestratorto 4.0.1 (the usage-report removal and the run-wave and director fixes above) andorchestration-modeto 0.1.1 (the band fixes and redesign above).
Removed
- Run token and cost budgets and agent-reported usage are removed from the MCP and REST surface and from config (#677, #678). They shipped incomplete in 4.0.0: usage could not land once the members were terminal, and the budget measured the last turn's figure instead of the run's spend. Removed:
advanceheartbeatusage;runs_writeopen/updatebudget.tokensandbudget.costUsd;runs_writeansweramount; the run view'susage.reportedTokens/usage.reportedCostUsdand thebudget.tokens/budget.costUsdmembers; theusagemember of aruns_readseatsrow; thepaused: {decisionId}member of write results; thebudgetsection of thereportrun rollup; andtokensandcostMicroUsdin thebudgetpayload of therun.openedevent (it now carries the attempt and seat caps only). A request that still sends a removed field failsunknown_parameteron that field (/usage,/budget/tokens,/budget/costUsd,/amount), so a hardcoded REST client must drop it.runs_readgetkeepsusage.observedTokens, the server's own estimate, which enforces nothing. The attempt cap (attemptsPerSeat), the seat cap (maxActiveSeats) and theattempt-cappause and decision are unchanged. A config document that still setsruns.budgets.tokensorruns.budgets.cost_usdloads: it validates and pushes (any value), each key draws one warning at its pointer, and no ceiling is applied. There is no migration: therunsandrun_seatscolumns stay and are no longer written, so a rollback to 4.0.0 is safe. State 4.0.0 stored still works: a run paused by arun-budgetdecision resumes onextend(noamount, no budget change) and closesbudget_exhaustedonstop, and a stored rollup snapshot with abudgetsection decodes with it ignored.
Fixed
runs_writeassignaccepts seatrevieweron a seat-less item whose schema has a review-role note. A seat-less schema (noseats:after trait merge) allowed onlyimplementerandplanner, so assigning a reviewer wasnot_foundandreview-wavecould not run a token-bound review for such items (this repo'sfeature-implementation). The allowed set is now those two plusreviewerwhen the resolved schema has any review-role note; a seat-less schema without one, a schema-free item and a seat-aware schema that does not declarereviewerstill answernot_found. No lane is enforced on a seat-less schema: the first writer's seat owns a note.