Why: Q3 WP4 requires A2A 1.0 interop from zero: publish this team's
capabilities as an Agent Card and consume remote agents as squad.Executors
with the A2A task state machine mapped onto ExecResult.
Evidence: adapters/a2a conformance with a local fake A2A server (both
directions): card build from role/blueprint registries, well-known JSON
serving, FetchCard fail-closed validation, message/send -> tasks/get
polling through submitted/working -> completed with TextPart/FilePart
artifacts mapped to patches/targets, failed/canceled/rejected/input-required
mapped to Success=false fail-closed (no cross-process hold), direct Message
result, JSON-RPC protocol and ctx-cancel error propagation. Module green
with -race; squad root suite unaffected (SELF). CI authoritative (GATE)
for the root module; the adapter module follows the adapters/intent and
adapters/native separate-go.mod precedent.
Impact: new module agent-squad/adapters/a2a (card.go, executor.go); no
changes to the squad root module.
Refs: Q3 plan WP4 (2026-10-07-agent-stack-q2q3-plans.md); task card
.task/feature-agent-stack-q3.md
Why: the Q3 compiler maps core.WorkflowDef prompt steps onto squad graph
steps; the rendered prompt text needs a first-class carrier in the step
task brief (until now only leader delegations had a briefing channel).
Evidence: TestStepBriefRendersIntoTaskFile asserts the section presence,
body, and position (after targets, before the context table); full squad
suite green (SELF), CI authoritative (GATE).
Impact: additive StepSpec field brief + writeTaskFile section; no behavior
change when Brief is empty.
Refs: Q3 plan WP1 (2026-10-07-agent-stack-q2q3-plans.md); Dev task card
.task/feature-agent-stack-q3.md
Why: Q2 acceptance 1 requires a scripted-backend conformance for every one of
the 12 builtin topologies; task_board was the last unlit driver.
Evidence: wp4_conformance_test.go: diamond-dependency board respects the
dependency gate in claim order; orphan lease (claimed + expired) re-arms on
driver re-entry and re-claims with Attempts=2 via Resume; non-critical task
failure cascades dependents to skipped and finishes degraded; fail-closed
when tasks remain unclosable. TestTopologyConformance_allTwelve runs all 12
names (10 runnable to completion, handoff/mailbox rejected as primitives).
Full suite green with -race.
Impact: task_board lit in the topology registry (all 12 topologies now
covered); RunState gains Board (persisted in runstate.json); RunOptions gains
TaskBoardLeaseSeconds; leader topology now maps to mode before plan
confirmation so Topology=leader runs no longer hit the confirmation gate.
Refs: Q2 plan WP4 (2026-10-07-agent-stack-q2q3-plans.md); task card
.task/feature-agent-stack-q2.md
Merge pull request 'Add leader progress ledger, replan cap, and mailbox with actor attribution (Q2 WP2)' (#43) from feature/q2-leader-mailbox-wp2 into main
Why: Q2 acceptance requires leader runs to detect stuck delegation loops and
cap replans instead of spinning to the turn limit, and a mailbox primitive
whose messages can never resolve a pending permission approval (acceptance 3).
Evidence: squad conformance tests (leader_mailbox_conformance_test.go):
ledger stuck->replan, replan cap finalize, mailbox delivery after hold,
mailbox-cannot-approve-permission (structural + actor second defense),
post guards; full suite green with -race.
Impact: LeaderState gains Progress/StuckTurns/Replans/MaxReplans/ReplanNote
(persisted in leader.json); RunState gains Mailbox (persisted in
runstate.json); RespondDecision gains Actor - non-human actors are rejected
on choice gates; step briefs and leader task briefs include pending mail.
Refs: Q2 plan WP2 (2026-10-07-agent-stack-q2q3-plans.md); task card
.task/feature-agent-stack-q2.md
Why: the completion doc Q2 requires twelve builtin topologies each with
a scripted-backend conformance; today the engine has only the
standard/leader mode string with no topology abstraction, no loop
re-arm (cycles are fail-closed by design), no in-graph routing, and no
run deadline.
Evidence: GOWORK=off go vet + go test -count=1 green across squad and
all three adapter modules; ten conformance/validation tests:
sequential, parallel, loop-until-field (exact iteration count +
persisted LoopRuns), loop-budget-exhausted (loop_exhausted event, run
still completes), router-selects-branch (loser skipped, run NOT
degraded), router-no-match fail-closed, evaluator-optimizer (draft
re-armed via loop body until verdict=pass), long-running deadline
(pending steps skipped, degraded closeout), unknown/primitive/
unregistered-driver fail-closed, builtin registry completeness (12).
Impact: additive — Topology/Deadline on RunOptions, Topology on
Blueprint, Loop/LoopRuns/Branches/Handoff on StepSpec; empty topology
keeps byte-identical standard behavior; MaxIterations is the total
iteration cap (first run included); router-skipped steps are exempt
from degraded derivation; runParallel now preserves the first specific
RunFailed error instead of re-wrapping via failStep; deadline metadata
uses RFC3339Nano (second-precision RFC3339 truncated sub-second
deadlines into instant expiry).
Refs: Agent-System-Documents#44 (Q2 plan WP1)
Why: the exec bridge now threads Last*/cache (Agent-Native#39); squad
ProcessEvent carries them to hosts (the product's squad runner reads
them for usage accounting).
Evidence: GOWORK=off go vet green across squad and adapters/native;
nested pin at the merged native.
Impact: additive event fields.
Refs: Agent-Native#39, N3 plan WP6
Why: native main carries the cache-on-message_end follow-up; the Dev
combination check requires the nested pin to follow.
Evidence: tree-only pin move.
Impact: none.
Refs: Agent-Native#37
Why: native main carries the WP6 usage split and default retry; the
Dev combination check requires the nested pin to follow.
Evidence: tree-only pin move.
Impact: none.
Refs: Agent-Native#35
Why: native main carries the memory port and memory_write (N3 WP5);
the Dev combination check requires the nested pin to follow.
Evidence: tree-only pin move.
Impact: none.
Refs: Agent-Native#34
Why: native main carries the checkpoint closure (N3 WP3); the Dev
combination check requires the nested pin to follow.
Evidence: tree-only pin move.
Impact: none.
Refs: Agent-Native#33
Why: native main carries the GenAI span convention work (N3 WP4); the
Dev combination check requires the nested pin to follow.
Evidence: tree-only pin move.
Impact: none.
Refs: Agent-Native#32
Why: native main carries JournalOnly sessions and the seed-persist fix
(N3 WP2); the Dev combination check requires the nested pin to follow.
Evidence: tree-only pin move.
Impact: none.
Refs: Agent-Native#31
Why: native main carries repairOrphanTail and the N6 fix (N3 WP1); the
Dev combination check requires the nested pin to follow.
Evidence: tree-only pin move.
Impact: none.
Refs: Agent-Native#30
Why: native main (7425b32b) validates skills against the open spec
(N2 WP4); the Dev combination check requires the nested pin to follow.
Evidence: tree-only pin move.
Impact: none.
Refs: Agent-Native#28, N2 WP4 cascade
Why: native main (77e7f978) aligns the MCP client with 2026-07-28
(N2 WP3); the Dev combination check requires the nested pin to follow.
Evidence: tree-only pin move.
Impact: none.
Refs: Agent-Native#27, N2 WP3 cascade
Why: native main (64b8bd4c) adds the tool-result spill and cache
contract tests (N2 WP1); the Dev combination check requires the nested
pin to follow.
Evidence: tree-only pin move.
Impact: none.
Refs: Agent-Native#25, N2 WP1 cascade
Why: native main (ead690be) keeps workspace-less requests out of the
process CWD; the Dev CI build step requires a clean tree after tests.
Evidence: go build ./... green; tree-only pin move.
Impact: none beyond the pin.
Refs: Agent-Native#24, Agent-System-Dev#242
Why: the Dev combination check requires every nested agent-intent pin
to equal 343b1e6 (I2 increments 1-2 on intent main); the nested native
pin follows to 3a1ad5a (which itself carries intent 343b1e6).
Evidence: tree-only pin moves; squad code does not consume the new
ports.
Impact: none.
Refs: Agent-Native#23, I2 increment 3 cascade
Why: Q1 WP5 - platform's SQLite becomes an implementation of the
squad Store port so run state and orchestration state live in one
database, and platform's in-process execution goes through the
Executor port instead of bypassing it (X6).
Evidence: all five modules go vet clean and go test ok, including
engine-on-SQLite (state in DB, no run-level JSON files, second
engine recovers from the same DB), SQLiteStore round trip, and the
full squad compatibility net.
Impact: runState/hitlGate exported so the port is implementable;
Options.Store wires into engineDeps; leader persistence and
checkpoints route through the port; PrepareWorkspace no longer
writes run-level JSON (st.save owns it); platform RunInProcess
takes squad.Executor (callers updated); serve mounts the engine on
platform's RunStore - single orchestration.db for both.
Refs: Documents proposal 2026-10-07-agent-stack-n1q1i1-plans.md Q1 WP5
Why: Q1 WP4/WP6/WP7 - run persistence goes through a Store port
(FileStore carries the current layout), superstep boundaries write
tiered snapshots with a graph fingerprint for recovery continuity,
and the four hardcoded stop conditions move behind one Termination
registry with And/Or composition.
Evidence: go vet clean; full suite ok including new checkpoint,
continuity-rejection, termination-primitive, grant-cap and the
Q23-updated leader closeout tests; cmd module green with
SweepExpired wired into the sweeper loop.
Impact: runState saves via Store (FileStore unchanged by default);
advance writes sync/async/on_exit checkpoints and Resume rejects
graph-shape changes from the crash window; leader turn cap now
degrades through finalize instead of a bare failure (Q23); manual
repair grants are bounded (Q26); wakeup cap and step timeout read
from the same registry; the dead SweepExpired is now called (Q6).
Refs: Documents proposal 2026-10-07-agent-stack-n1q1i1-plans.md Q1 WP4/WP6/WP7
Why: Q1 WP3 - each step execution carries a stable attempt id
(runID/stepID/N) for audit and idempotency, and auto-mode
interrupted recovery must not re-execute a step whose products all
landed before the crash (side effects exactly once).
Evidence: go vet clean; full squad suite ok, including the new
kill-at-superstep-boundary test (completed steps never re-execute,
running-with-products judged complete by attempt) and
missing-product retry tests.
Impact: StepSpec gains AttemptID set when entering running and used
as the execution id; recoverInterrupted (HitlAuto) checks the
step's targets first - all present means completed with artifact
merge, otherwise the legacy pending retry. Transient retries share
the same AttemptID.
Refs: Documents proposal 2026-10-07-agent-stack-n1q1i1-plans.md Q1 WP3
Why: Q1 WP1/WP2 - replace the linear step scan with per-round graph
compilation and layer advancement so Needs drives scheduling and
write-conflicting steps never run concurrently.
Evidence: go vet clean in all five modules; go test ok everywhere,
including the full compatibility net (engine, leader, recovery, r2
regression, intent, evaluator suites) and new graph tests.
Impact: new squad/graph.go (compile/layer/conflict); StepSpec gains
Writes; advance advances by superstep (parallel run of
conflict-free members, serial for write-conflicts, single-member
groups keep dynamic adjustment). Legacy semantics preserved:
adjacent-Parallel mapping, degraded-continue on non-critical
failure (failed counts as satisfied), empty status = pending,
cycle/unknown/dup needs fail-closed.
Refs: Documents proposal 2026-10-07-agent-stack-n1q1i1-plans.md Q1 WP1/WP2
Why: Agent-Native PR #22 landed after Squad #21; squad's nested
native must carry the same head or Dev check-pins fails.
Evidence: 78b6c647 is the #22 merge on agent-native main.
Impact: Nested pin only.
Why: Agent-Intent PR #8 landed the multi-turn eval format; nested
copies must carry it or Dev check-pins fails.
Evidence: intent main head is the #8 merge.
Impact: Nested pins only.
Why: Agent-Native PR #21 landed the nested intent bump; squad's
nested native must carry the same head or Dev check-pins fails.
Evidence: 6ea69293 is the #21 merge on agent-native main; squad
modules run next.
Impact: Nested pin only.
Why: Both intent hosts should fold compatibility-equivalent input
(backend intentgate already does); the nested pin carries the I1
command model the gate depends on.
Evidence: adapters/intent suite ok; all five squad modules green.
Impact: NFKC on in NewFromOverlay; nested agent-intent at 2554b50.
Why: Dev moved its agent-native pin to c9c2f23 (PRs #19/#20);
check-pins requires the same combination in squad's nested copy.
Evidence: c9c2f23 is the #20 merge on agent-native main; squad
suites run next.
Impact: Nested pin only.
Why: Squad members must reach the host's HITL verdict when a kernel
hook answers ask - the new native GateOptions.PermissionAsk field
carries it; nil preserves the legacy ask=reject behavior.
Evidence: go vet clean in all five modules; go test ok everywhere.
Impact: Additive Config field plus its buildOptions mapping.
Why: Library PR #18 landed AgentSpec, guardrails, subagent events and
the GateOptions.PermissionAsk field; squad's register mapping needs
that field, and check-pins requires the same combination.
Evidence: native main head is the #18 merge; squad suites run next.
Impact: Nested pin only.
Why: Dev moved its agent-native pin to c2befaf (PR #17); squad's
nested pin must carry the same combination or check-pins fails.
Evidence: c2befaf is the merge of PR #17 on agent-native main; squad
test suite runs next.
Impact: Nested pin only.
Why: Dev now builds provider requests through PrepareMessages. Squad's nested copy has to sit on that same commit or the pin check fails.
Evidence: The gitlink moves from e4020d2 to 3a110a7. No squad source changed.
Impact: Squad compiles against the message shaper. Empty MessageShape still leaves transcripts unchanged.
Refs: docs/records/2026-10-04-sillytavern-review.md
Why: squad members use the same native kernel, so the script switch and the user-input callback have to reach that backend.
Evidence: GOWORK=off go test . in adapters/native.
Impact: the new fields default to off and nil, so existing member runs stay the same. The agent-native pin moves to the orchestration commit.
Refs: feature/code-orchestration
Co-authored-by: Cursor <cursoragent@cursor.com>
Why: squad members write files and call tools through the same kernel, so rewind and span hooks have to reach that backend.
Evidence: GOWORK=off go test . in adapters/native.
Impact: nil checkpoints and a nil trace func leave member runs unchanged. The agent-native pin moves to the durability commit.
Refs: feature/durable-rewind
Co-authored-by: Cursor <cursoragent@cursor.com>
Why: The hook-bus commit was replayed onto origin/main, so the squad pin has to follow that commit.
Evidence: deps/agent-native is 0a52d9f, which contains origin/main.
Impact: Squad members load the same kernel commit as the top-level pin.
Refs: feature/permission-hooks
Why: Squad members used the same kernel as direct chat but dropped stop, reminder, and before-tool hooks.
Evidence: GOWORK=off go test . in adapters/native passed after pinning agent-native to 99d623a.
Impact: Zero GateOptions still builds the previous backend. A set BeforeTool now runs inside the member loop.
Refs: feature/permission-hooks
Co-authored-by: Cursor <cursoragent@cursor.com>
Why: Squad runs on the same native loop, so it should see tool-result clearing, folded reminders, and the compaction retry.
Evidence: deps/agent-native is 6fd1d44.
Impact: No squad behavior change beyond the native loop it already calls.
Refs: feature/context-engineering
Co-authored-by: Cursor <cursoragent@cursor.com>
Why: Squad members share the same model stream, so the retry and overflow fixes have to be the commit this repo builds.
Evidence: agent-native 9f5751c. No squad behavior change.
Impact: Only the agent-native gitlink moves.
Refs: feature/llm-robustness
Co-authored-by: Cursor <cursoragent@cursor.com>
Why: the squad checkout has to record the native head that treats a cancelled stream as aborted.
Evidence: go test -count=50 -race ./runtime/llm/ passed on d84a67c.
Impact: gitlink only.
Refs: D1
Why: the squad checkout has to record the same intent and native commits Dev will pin.
Evidence: gitlinks only. Schema and adapter tests already passed on those heads.
Impact: no squad source change.
Refs: D1
Why: the command has to load guard overlays without naming agent-intent types. Only adapters/intent may import that library.
Evidence: GOWORK=off go test -count=1 in adapters/intent and cmd/agent-squad passed.
Impact: CLI calls NewFromOverlay. Guard domains stay guard.
Refs: D1
Co-authored-by: Cursor <cursoragent@cursor.com>
Why: the squad core has to compile without agent-native or agent-intent. Ports stay in the core; adapters and the CLI own the concrete clients.
Evidence: GOWORK=off go test -count=1 ./squad/ ./store/ ./textextract/ ./schema/ passed, and the intent, sqlite, native, and cmd modules passed their own go test.
Impact: core go.mod keeps error-code-system only. Nested gitlinks stay. Callers register the native backend through agent-squad/adapters/native.
Refs: D1
Co-authored-by: Cursor <cursoragent@cursor.com>
Why: squad vet failed against native 11d1a76 because that module did not yet sum golang.org/x/text.
Evidence: GOWORK=off go vet ./squad/ ./squadintent/ ./platform/ ./cli/ passed on native 855fe41.
Impact: the native gitlink and this module's go.sum move. Squad behavior is unchanged.
Refs: I4
Co-authored-by: Cursor <cursoragent@cursor.com>
Why: squad must build against the native and intent commits that carry this batch.
Evidence: agent-intent 05e74d8 and agent-native 11d1a76 are the feature/agent-stack-r2 heads whose race tests passed.
Impact: nested gitlinks move; squad behavior otherwise stays on the preceding commit.
Refs: Q6, I9
Co-authored-by: Cursor <cursoragent@cursor.com>
Why: parallel groups dropped a hold, HITL timeouts were unused, the planner failed closed on any model error, and event sequence numbers restarted at 1.
Evidence: GOWORK=off go test -count=1 -race ./squad/ ./squadintent/ ./store/ ./platform/ passed, including TestR2ExpiredDecisionUsesCancel, TestR2IntentStepSkipsRoleRegistry, TestR2PlannerFallsBackAndCapsSteps, TestR2TaskFileUsesLatestBlackboard, TestR2FanoutCancelDoesNotPanic, TestR2EventSeqContinuesAfterRestart, TestR2JournalModeWAL, and TestR2GateReloadSeesNewIntent.
Impact: a failed or held parallel sibling cancels the rest; expired requests synthesize cancel or fail; the intent gate follows Registry.Reload.
Refs: Q6, Q10, Q11, Q12, Q13, Q17, Q18, Q19, I9
Co-authored-by: Cursor <cursoragent@cursor.com>
Why: The runner cannot reach github.com, so cloning actions/checkout timed out before any test ran.
Evidence: The Go 1.27.1 tarball is available from mirrors.aliyun.com. Checkout uses the job token and GITHUB_SHA.
Impact: Submodule checkout still requires SUBMODULE_TOKEN. A missing secret fails the job with that message.
Refs: feature/agent-stack-r0-r1
Co-authored-by: Cursor <cursoragent@cursor.com>
Why: Squad must execute the native and intent commits that this batch is asking to merge.
Evidence: deps/agent-native is 91c155618487e21edeb4f1d3cc27e985aae2b0d2 and deps/agent-intent is 7f7879b06b053f0a62746309b79e0e1b9a00f42d.
Impact: Both gitlinks match the open pull request heads. Nested guideline pins are unchanged.
Refs: feature/agent-stack-r0-r1
Co-authored-by: Cursor <cursoragent@cursor.com>
Why: SSH submodule checkout failed on the runner, and the docker job required a path this repo does not have.
Evidence: vet-and-test uses recursive HTTPS checkout, setup-go 1.27, GOPROXY, and GOWORK=off vet, test, and race. The docker job only checks go.mod.
Impact: SUBMODULE_TOKEN must be set by the owner before checkout can succeed.
Refs: feature/agent-stack-r0-r1
Co-authored-by: Cursor <cursoragent@cursor.com>
Why: A parallel group only checked the first step, free text cancelled confirmation and permission holds, and a failed reply deleted the pending request before the decision ran.
Evidence: GOWORK=off go vet ./...; go test -count=1 ./squad/ ./cli/; go test -race -count=1 ./squad/. New tests cover the second parallel member, free-text holds, restored pending, AtomicWrite, and listenGuard.
Impact: Any member that needs approval or bash permission keeps the whole group suspended. Free text on confirmation, permission, and quality review is stored as steering and the same request stays pending. A dispatch error restores pending.json.
Refs: feature/agent-stack-r0-r1
Co-authored-by: Cursor <cursoragent@cursor.com>
Why: A step confirmation cancelled the whole run, approvals and repair budget vanished on restart, and the HTTP server accepted anonymous and oversized requests.
Evidence: GOWORK=off go vet ./... && GOWORK=off go test -count=1 ./... && GOWORK=off go test -race ./squad/... passed. scripts/submodule-consistency.sh ok.
Impact: Approve and skip apply to one step; runstate.json and unix flock survive restart; wakeup leaves permission holds pending; serve defaults to loopback and requires a token off-loopback.
Refs: Q1 Q2 Q3 Q4 Q5 Q8 Q9 Q14 Q15 Q16
Co-authored-by: Cursor <cursoragent@cursor.com>
Why: squad must compile against the same kernel commit as the direct path, while leaving the new loop switches unset.
Evidence: scripts/submodule-consistency.sh printed ok, and go test ./... passed.
Impact: only the agent-native gitlink changes. Squad runtime behavior is unchanged.
Refs: feature/agent-harness-p1
Co-authored-by: Cursor <cursoragent@cursor.com>
Why: Squad runs were one file workspace and one JSON file the leader had to emit, so agents, issues, and retries had no durable queue.
Evidence: GOWORK=off go test -count=1 -timeout 180s ./...
Impact: /v1/runs stays. New routes mount only when the orchestration database opens. SSE for a run id now skips other runs. External CLIs are not a sandbox.
Refs: feature/orchestration-platform S1-S7
Co-authored-by: Cursor <cursoragent@cursor.com>
Why: Tests and product budgets need to drive each kernel step without a live model, and stop it when the tool-call cap is hit.
Evidence: go test ./squad/ -count=1 passed after fetching agent-native cfc840c.
Impact: AgentNativeConfig.Stream and MaxToolCalls reach native.Options. The nested kernel pin moves to cfc840c.
Refs: deliverable-quality-system B2
Co-authored-by: Cursor <cursoragent@cursor.com>
Why: Simple tasks should finish as one assistant step, while complex tasks stay in leader mode, and every kernel step needs the host's tools and events.
Evidence: go test ./squad/ -count=1 passed, including direct route, leader entry, skill override, and delegation score fields.
Impact: Hosts can inject HostTools and OnEvent; a Router can choose direct or plan without changing product roles.
Refs: deliverable-quality-system B2
Co-authored-by: Cursor <cursoragent@cursor.com>
weichen
