Repository navigation
config.json support for defaultEffort and agentEffortOverrides #773
Description
Problem
The copilot --effort flag exists at session level, but it's unclear whether it propagates to task tool sub-agent spawns (#729 shows the platform injects low regardless). It's unclear if there is a way to figure effort per-agent or set a persistent default through Squad's config.
(It's also not clear to me if this is a limitation of Copilot CLI and how it spawns sub-agents.)
Proposal
Extend .squad/config.json with effort overrides, following the same pattern as defaultModel / agentModelOverrides:
{
"version": 1,
"defaultEffort": "xhigh",
"agentEffortOverrides": {
"scribe": "medium",
"ralph": "medium"
}
}defaultEffort— applies to all agents unless overridden. Persistent across sessions.agentEffortOverrides— per-agent effort level. Useful for downgrading mechanical agents (Scribe, Ralph) while keeping code-writing agents at xhigh.- Valid values:
"low","medium","high","xhigh"(matchingcopilot --effortchoices)
The coordinator would resolve effort using the same layered approach as model selection:
agentEffortOverrides.{name}(per-agent persistent)defaultEffort(global persistent)- Session directive ("use xhigh effort")
- Charter preference (
Effort: xhighin## Modelsection) - Platform default
Why this matters
Different agents need different effort levels. Code-writing agents (implementations, refactors, statistical modeling) benefit from xhigh. Mechanical agents (Scribe logging, changelog generation) waste tokens at xhigh and work fine at medium. The current all-or-nothing approach forces users to choose between quality and cost.
Related
- task tool sets reasoning_effort: low which is incompatible with gpt-5.2-codex #729 —
tasktool setsreasoning_effort: low, causing a 400 error forgpt-5.2-codex(only supportsmedium) - Copilot CLI supports
--effortat session level (copilot --effort xhigh), but propagation to sub-agent spawns is unverified - Squad SDK already has
reasoningEffortonSquadSessionConfigand per-modelsupportedReasoningEfforts— the plumbing exists at the SDK layer
Activity
github-actions commentedon Apr 8, 2026
🏗️ Squad Triage — Flight (Lead)
Issue: #773 — config.json support for defaultEffort and agentEffortOverrides
Assigned to: Flight (Lead)
Reason: No specific domain match — assigned to Lead for further analysis
@copilot evaluation: 🔴 Not suitable — issue involves work outside the coding agent's capability profile.
Team roster:
squad:flightsquad:proceduressquad:eecomsquad:fidosquad:paosquad:capcomsquad:controlsquad:surgeonsquad:boostersquad:gncsquad:networksquad:retrosquad:incosquad:guidosquad:telemetrysquad:voxsquad:dskysquad:simssquad:handbooksquad:ralphsquad:copilotfeat(config): add defaultEffort and agentEffortOverrides to config.js…
docs: update GITHUB_ISSUES_FIXED.md with confirmed-fixed issues
feat(config): add defaultEffort and agentEffortOverrides to config.js…
docs: update GITHUB_ISSUES_FIXED.md with confirmed-fixed issues
bradygaster commentedon May 18, 2026
Closing — depends on #729 which is a platform limitation. Squad has no mechanism to pass reasoning_effort through the task tool. Will reopen if/when the platform exposes this parameter.