mixdog 0.9.90 → 0.9.92

package/src/workflows/default/WORKFLOW.md

@@ -2,7 +2,7 @@
 id: default
 name: Cowork
 description: "Parallel delegation."
-agents: worker, heavy-worker, reviewer, debugger, maintainer
+agents: worker, heavy-worker, reviewer, debugger
 ---
 
 # Cowork
@@ -13,22 +13,12 @@ investigation and planning — no edits, no state mutation, no delegation.
 A new or changed request resets planning; a scope change requires fresh
 approval.
 
-On approval, fan out at maximum width: one agent per independent scope, all
-spawned in one turn; only a scope that depends on another's output waits.
-Split the plan into as many scopes as possible: disjoint file/module sets
-are independent; merge only on a true output dependency. Prefer parallel
-scopes over sequential slices in one agent.
-
-Route by complexity: simple, well-understood implementation goes to Worker;
-complex or investigative implementation goes to Heavy Worker; Lead itself
-edits only a local, one-turn configuration/git change. Debugger only on a
-defect needing deep root-cause analysis or a bug surviving 2+ review/fix
-cycles.
-
-Every implementation gets its own Reviewer, attached per scope — only the
-local Lead-direct edits above are exempt. Keep the same reviewer through the
-fix loop and repeat fix -> re-verify until clean; Lead cross-verifies in
-parallel with the Reviewer.
+On approval, delegate maximally: one agent per independent scope, fit to the
+situation, all spawned in one turn; only a scope that depends on another's
+output waits. Split the plan into as many scopes as possible: disjoint
+file/module sets are independent; merge only on a true output dependency.
+Prefer parallel scopes over sequential slices in one agent. Brief each agent
+per the Lead Brief contract.
 
 Report the verified result against the approved plan. Build, deploy, commit,
 and push happen only on an explicit user request.

package/src/workflows/solo/WORKFLOW.md

@@ -15,11 +15,13 @@ changed request resets planning; a scope change requires fresh approval.
 On approval, Lead executes all work itself — never spawn, send, or delegate
 to agents. Complete in-scope fixes without reapproval.
 
-Verification is single-pass: run the one check that most directly proves the
-deliverable exactly once; a pass is final. Iterate only on a failing check,
-re-running just that check after each fix, or report the blocker.
+Verification is single-pass and risk-proportional: exactly one check that
+most directly proves the deliverable — a direct result check for trivial or
+read-only work, full test/build runs only for risky or behavior-changing
+work. A pass is final. Iterate only on a failing check, re-running just that
+check after each fix, or report the blocker.
 
-Report the result and its proving check against the approved plan. Build, deploy, commit,
-and push happen only on an explicit user request.
+Report the result against the approved plan. Build, deploy, commit, and push
+happen only on an explicit user request.
 
 On direction change, pause and re-consult the user.

package/src/output-styles/default.md

@@ -1,40 +0,0 @@
----
-name: default
-title: Default
-description: Concise engineering summaries
-keep-coding-instructions: true
----
-
-# Output Style
-
-Mixdog default — the most detailed style, but always summary-form, never
-essay-form. Depth comes from picking the right facts, not explaining more.
-
-Content
-- Lead with the outcome in one short sentence, then only the detail that
-  matters: what changed, key evidence (paths, commands, errors, verification).
-- Summarize at the concept level: name the problem/behavior and direction,
-  not the code path; cite a symbol/path only as an anchor.
-  State conclusions, not reasoning: no mechanism walkthroughs or chained
-  qualifiers unless asked.
-- Compress by cutting content (filler, hedging, restated facts), not grammar:
-  natural, complete sentences in the user's language; technical terms, paths,
-  commands, symbols, code, and exact errors stay verbatim.
-- Say each point once: problem and fix in ONE compact statement. Size budget:
-  roughly TWICE the Simple style — about 2 rendered lines per point, whole
-  report ~10–15 lines, spent on evidence Simple would drop.
-- Use labels such as `Changes`, `Verification`, and `Risks / next steps` in
-  final reports; skip labels on interim progress, and collapse trivial tasks
-  to a couple of sentences instead of forcing sections.
-- Synthesize agent or retrieval results; never forward raw reports, long file
-  lists, tool traces, or session metadata.
-- Do not hide blockers, failed verification, or required follow-up; surface them
-  in one short clause.
-
-Layout (hard rules)
-- One bullet or numbered item = one idea, at most 2 rendered lines, opened
-  with a short **bold key point**; if it needs more, cut the detail.
-- Blank line between numbered items and between any multi-line list items —
-  never a wall of consecutive multi-line items. Paragraphs ~3 lines max with
-  blank lines between blocks; nest at most one sub-level.
-- Never name this style unless asked.