tamirdresher · Aug 13, 2026 · May 25, 2026 · Jun 7, 2026 · Jun 9, 2026 · Jun 9, 2026
@@ -362,3 +362,8 @@ Triaged 14 untriaged issues (3 docs, 6 community features, 3 bugs, 2 questions).
 **By:** Booster (CI/CD)
 **What:** Templates under `packages/squad-cli/templates/workflows/` and `packages/squad-sdk/templates/workflows/` are now linted by actionlint via explicit file path arguments in the new `squad-workflow-lint.yml` CI job. SC2086 findings fixed across all `>> $GITHUB_OUTPUT` / `>> $GITHUB_STEP_SUMMARY` redirects in `.squad-templates/workflows/`, `templates/workflows/`, `.github/workflows/squad-heartbeat.yml`, `squad-repo-health.yml`, and `squad-ci.yml`. Actionlint pinned to tag `v1.7.12`; shellcheck 0.10.0 installed explicitly.
 **Why:** Downstream repos running actionlint in their CI saw SC2086 errors in files generated by `squad upgrade` because Squad's own CI did not lint templates. The fix must live in Squad's templates — `squad upgrade` overwrites any downstream patches on every run. PR #1557.
+## Deferred
+
+- #357, #336, #335, #334, #333, #332, #316 (A2A) — stays shelved per existing decision
+- #581 (ADO PRD) — P2, blocked until #341 (SDK-first parity) ships
+

@@ -0,0 +1,4 @@
+### 2026-03-26: CI deletion guard and source tree canary
+**By:** Booster (CI/CD)
+**What:** Added two safety checks to squad-ci.yml: (1) source tree canary verifying critical files exist, (2) large deletion guard failing PRs that delete >50 files without 'large-deletion-approved' label. Branch protection on dev requested (may need manual setup).
+**Why:** Incident #631 — @copilot deleted 361 files on dev with no CI gate catching it.
@@ -0,0 +1,35 @@
+# Decision: Release Process Skill Update — v0.9.4 Learnings
+
+**Author:** Booster (CI/CD Engineer)
+**Date:** 2026-04-25
+**Status:** Implemented
+**Requested by:** Brady
+
+## Summary
+
+Updated both release-process skill files with critical learnings from the v0.9.4 release session. The v0.9.4 release was delayed by three distinct issues, each fixed by a separate PR.
+
+## Changes Made
+
+### Files Updated
+1. `.squad/skills/release-process/SKILL.md` (team-level skill)
+2. `.copilot/skills/release-process/SKILL.md` (copilot-level skill)
+3. `.squad/agents/booster/history.md` (learnings log)
+
+### New Knowledge Added
+
+| Issue | Root Cause | Fix PR | Skill Section |
+|-------|-----------|--------|---------------|
+| Root package.json version drift | squad-release.yml reads from root, not sub-packages | #1043 | Known Gotchas + v0.9.4 Incident Learnings |
+| CHANGELOG missing `## [$VERSION]` | Workflow validates version entry exists | #1042 | Known Gotchas + Release Checklist |
+| Lockfile integrity check rejects workspace packages | Check didn't filter for registry-only packages | #1044 | Known Gotchas + Common Failure Modes |
+| GITHUB_TOKEN can't trigger downstream workflows | GitHub security feature prevents event propagation | N/A (design) | GITHUB_TOKEN section + Manual Publish |
+| Prebuild bump breaks workspace linking | bump-build.mjs mutates versions breaking exact match | N/A (known) | Local Development section |
+
+### Cross-References
+- Added bidirectional cross-references between team-level and copilot-level skill files
+- Added PR references (#1042, #1043, #1044) as source evidence throughout
+
+## Rationale
+
+These are high-impact, recurring failure modes. Documenting them in the skill files ensures every agent (human or AI) working on releases has the knowledge to avoid repeating the v0.9.4 delays. The GITHUB_TOKEN limitation in particular is non-obvious and would catch any future release.
@@ -0,0 +1,100 @@
+---
+author: Data
+date: 2026-08-13
+status: decided
+references:
+  - "PR #1699"
+  - "PR #1698 (v0.12.0 dev -> main promotion, unblocked by this)"
+  - "PR #1415 (tools: ['*'] fix, verified already on dev)"
+---
+
+# Decision: restore shared ancestry between dev and main via merge, dev-wins conflict policy
+
+## Context
+
+`dev` and `main` had unrelated git histories. `dev`'s root commit (`4c5772c5`, a
+dependabot bump dated 2026-07-13) shares no ancestor with `main`'s root
+(`f4830e48`, 1722 commits). `dev` itself has 196 commits. `git merge-base
+upstream/dev upstream/main` returned nothing and `git merge-tree` refused
+outright. This made the v0.12.0 promotion PR (#1698, `dev` -> `main`) flatly
+unmergeable, with no way to review a normal diff. The v0.11.0 promotion
+(2026-06-29) predates the July 13 reset, which is why it worked and why
+#1698 was the first to hit this wall.
+
+## Decision
+
+Merge `upstream/main` into `dev` with `--allow-unrelated-histories`, resolving
+every one of the 275 conflicts in `dev`'s favor, on a new branch
+`squad/restore-main-ancestry`, landed via PR #1699 into `dev`.
+
+Rejected alternative: rebasing `dev` onto `main`. `dev` carries a
+`non_fast_forward` ruleset that blocks the force-push a rebase requires, and a
+rebase would rewrite all 196 commits on `dev`, invalidating every PR currently
+open against it. A merge is additive only and needs no force-push.
+
+## Why dev-wins conflict resolution is safe
+
+All 275 conflicts were `add/add` (identical path on both branches, no common
+base to 3-way merge from). Resolved every one with `git checkout --ours` (dev's
+content). This is safe because the only `main`-only code fix that mattered,
+#1415 (`tools: ['*']` in `.github/agents/squad.agent.md`), was verified already
+present on `dev` before starting the merge, so dev-wins loses zero code.
+Verified the resolution was a true no-op against dev's pre-merge tree: diffing
+all 275 resolved files plus the 48 removed changesets plus the 6 auto-merged
+files against dev's HEAD showed only 18 real changes total, exactly the
+files deliberately kept (see below). Confirmed unchanged post-merge:
+`package.json` / `packages/squad-cli/package.json` / `packages/squad-sdk/package.json`
+(all `0.12.0`), `CHANGELOG.md` (`## [0.12.0] - 2026-08-12` intact),
+`test/gh-aw-quality.test.ts` (#1697 fix intact), `workflows/squad.md` and
+`workflows/squad-implement-worker.md` (#1682 feature intact).
+
+## What was restored (kept from main, lost in the July 13 reset)
+
+7 docs pages:
+- docs/src/content/blog/015-wave-2-the-repl-moment.md
+- docs/src/content/blog/032-v010-stabilisation-insider.md
+- docs/src/content/blog/033-swe-bench-lite-results.md
+- docs/src/content/docs/features/remote-control.md
+- docs/src/content/docs/get-started/choosing-your-path.md
+- docs/src/content/docs/guide/personal-squad.md
+- docs/src/content/docs/guide/shell.md
+
+5 decision records:
+- .squad/decisions/inbox/booster-ci-deletion-guard.md
+- .squad/decisions/inbox/booster-release-skill-v094.md
+- .squad/decisions/inbox/flight-versioning-policy.md
+- .squad/decisions/inbox/procedures-fix-coordinator-inline-dispatch-gate.md
+- .squad/decisions/inbox/retro-copilot-git-safety.md
+
+Plus 6 files that auto-merged cleanly via a genuine 3-way merge (not a
+conflict), appending older history from main onto dev's existing content with
+no loss on either side: `.squad/agents/{eecom,fido,flight,pao,procedures}/history.md`
+and `.squad/decisions.md`.
+
+## What was dropped
+
+48 `.changeset/*.md` files that arrived from `main`. These are spent: their
+content is already consumed into `CHANGELOG.md`'s `[0.12.0]` entry, and
+re-adding them risked tripping the Changeset Drift check. `.changeset/` after
+this merge contains exactly what `dev` had before: `README.md`,
+`config.json`, `max-reasoning-effort.md`.
+
+## Validation
+
+Static/diff verification was thorough (see above). Dynamic validation
+(`npm run build`, `npx vitest run`) could **not** be executed in this sandbox:
+the corporate npm proxy does not mirror `eslint-plugin-n@18.3.0` (its cache
+tops out at 18.2.2) and there is no direct route to the public npm registry
+from this environment. Confirmed unrelated to the merge:
+`package-lock.json`'s staged content is byte-identical to dev's pre-merge
+HEAD, and this exact dependency/version was already in dev's lockfile before
+any of this work. CI (which has full registry access) is the source of truth
+for build/test validation on PR #1699.
+
+## Outcome
+
+`git merge-base <this-branch> upstream/main` now succeeds, confirming `dev`
+has a real common ancestor with `main` again. This unblocks PR #1698 and
+prevents the same failure on future promotions, provided the new release-process
+skill gate (verify `git merge-base dev main` before starting release-prep) is
+followed going forward.
@@ -0,0 +1,32 @@
+# Decision: Versioning Policy — No Prerelease Versions on dev/main
+
+**By:** Flight (Lead)
+**Date:** 2026-03-29
+**Requested by:** Dina
+**Status:** DECIDED
+**Confidence:** Medium (confirmed by PR #640 incident, PR #116 prerelease leak, CI gate implementation)
+
+## Decision
+
+1. **All packages use strict semver** (`MAJOR.MINOR.PATCH`). No prerelease suffixes on `dev` or `main`.
+2. **Prerelease versions are ephemeral.** `bump-build.mjs` creates `-build.N` for local testing only — never committed.
+3. **SDK and CLI versions must stay in sync.** Divergence silently breaks npm workspace resolution.
+4. **Surgeon owns version bumps.** Other agents must not modify `version` fields in `package.json` unless fixing a prerelease leak.
+5. **CI enforcement via `prerelease-version-guard`** blocks PRs with prerelease versions. `skip-version-check` label is Surgeon-only.
+
+## Why
+
+The repo had no documented versioning policy. This caused two incidents:
+
+- **PR #640:** Prerelease version `0.9.1-build.4` silently broke workspace resolution. The semver range `>=0.9.0` does not match prerelease versions, causing npm to install a stale registry package instead of the local workspace link. Four PRs (#637–#640) patched symptoms before the root cause was found.
+- **PR #116:** Surgeon set versions to `0.9.1-build.1` instead of `0.9.1` on a release branch because there was no guidance on what constitutes a clean release version.
+
+## Skill Reference
+
+Full policy documented in `.squad/skills/versioning-policy/SKILL.md`.
+
+## Impact
+
+- All agents must follow the versioning policy when touching `package.json`
+- Surgeon charter should reference this skill for release procedures
+- CI pipeline enforces the policy via automated gate
@@ -0,0 +1,45 @@
+# Decision: Restore always-on coordinator inline-dispatch gate (v0.10.0 regression)
+
+- **Author:** Procedures (Prompt Engineer)
+- **Date:** 2026-06-24
+- **Requested by:** Tamir Dresher
+- **Branch:** squad/fix-coordinator-inline-dispatch-gate (worktree)
+- **Status:** Proposed — applied in working tree, awaiting Lead (Flight) review; NOT committed.
+
+## Context
+
+Matthew Wan reported the main Squad coordinator "does a lot of work on its own instead of using his roster of agents." Worked in v0.9.4, broke in v0.10.0. Dina Berry worked around it by hand-editing the coordinator charter.
+
+## Root cause
+
+Commit `afe78188` (#1035, "context overflow sentinel + coordinator size reduction") cut the coordinator template and moved the concrete inline-dispatch gate and dispatch mechanics out of the always-on prompt into lazy-loaded reference files (`client-compatibility-reference.md`, `spawn-reference.md`). The leftover inline guidance ("inline work is last-resort fallback only") was too soft to hold the line, so the coordinator began executing domain work itself.
+
+## Decision
+
+Re-establish a strong, **always-on** dispatch gate without breaking the legitimate Direct/Lightweight response modes or the size-reduction spirit. Three minimal edits (~14 lines) to the canonical template `.squad-templates/squad.agent.md`, propagated to all 5 synced copies via `npm run sync-templates`:
+
+1. **Inline-dispatch gate** (Client Compatibility): inline permitted ONLY in Direct Mode, or when NEITHER `task` NOR `runSubagent` exists; otherwise MUST dispatch.
+2. **STOP gate** (How to Spawn an Agent): about to emit a domain artifact with no spawn call this turn → STOP and dispatch (exceptions: Direct Mode / no spawn tool).
+3. **VS Code `runSubagent` micro-playbook** re-inlined (~5 lines) so how-to-dispatch is always-on.
+
+## The 5 synced copies (canonical first)
+
+1. `.squad-templates/squad.agent.md` (canonical — edited here)
+2. `templates/squad.agent.md.template`
+3. `.github/agents/squad.agent.md`
+4. `packages/squad-cli/templates/squad.agent.md.template`
+5. `packages/squad-sdk/templates/squad.agent.md.template`
+
+## Validation
+
+- New regression/parity test `test/coordinator-inline-dispatch-gate.test.ts` (deterministic, subprocess-free): RED before (8/8 fail), GREEN after (8/8 pass).
+- Full `npm run test`: 18 failed files vs 31 baseline (fewer); all failures are pre-existing flaky subprocess/timeout e2e tests (3 of 4 "new vs baseline" pass in isolation; the 4th is `squad doctor` at 40x10 exiting null = timeout, unrelated to template content). No new deterministic failures.
+- `npm run lint` (tsc noEmit both packages): clean.
+
+## Changeset
+
+Required. The `changelog-gate` regex governs `.squad-templates/`, `templates/`, and `packages/squad-(sdk|cli)/templates/` — not just `src/`. Added `.changeset/fix-coordinator-inline-dispatch-gate.md` (patch, squad-cli + squad-sdk).
+
+## Rollout
+
+Existing installs require `squad upgrade` to pick up the regenerated coordinator agent.
@@ -0,0 +1,4 @@
+### 2026-03-26: Copilot git safety rules
+**By:** RETRO (Security)
+**What:** Added mandatory Git Safety section to copilot-instructions.md: prohibits staging the entire working tree with a bare-dot `git add` (i.e. `git add` followed by just `.`), requires feature branches and PRs, adds pre-push checklist, defines red-flag stop conditions.
+**Why:** Incident #631 — @copilot used destructive staging on an incomplete working tree, deleting 361 files.
@@ -34,6 +34,18 @@ Squad publishes two npm packages: `@bradygaster/squad-sdk` and `@bradygaster/squ
 
 ## Rules (Non-Negotiable)
 
+### 0. Verify `dev` and `main` Share Ancestry BEFORE Starting Any Release-Prep Work
+
+> Source: 2026-08-13 orphaned-history incident, PR #1699. `dev`'s history was silently reset on 2026-07-13 (a dependabot bump became a brand-new root commit with zero shared history with `main`). Nothing in normal release-prep work (version bump, CHANGELOG entry, changeset consolidation, CI) surfaces this — `dev` builds and tests green the whole time because none of it touches `main`'s history. The break is invisible until the very last step: opening the `dev` → `main` promotion PR, which GitHub then reports as flatly unmergeable, with `git merge-tree` refusing outright and no reviewable diff.
+
+**Before touching version numbers or the CHANGELOG, run:**
+
+```bash
+git merge-base upstream/dev upstream/main
+```
+
+If this returns a commit SHA (exit 0), proceed normally. **If it returns nothing (exit 1), STOP.** The promotion PR will be unmergeable no matter how clean `dev`'s own release content is — fix ancestry first via a separate `chore: restore shared ancestry` PR (merge `main` into `dev` with `--allow-unrelated-histories`; do not rebase if `dev` has a `non_fast_forward` ruleset, since rebase requires a force-push and rewrites every commit under every open PR). See `.squad/decisions/inbox/data-restore-dev-main-ancestry.md` for the full worked example and the diff-based verification technique used to prove a dev-wins conflict resolution was safe.
+
 ### 1. Coordinator Does NOT Publish
 
 The coordinator routes work and manages agents. It does NOT run `npm publish`, trigger release workflows, or make release decisions. Brady owns the release trigger. If an agent or the coordinator is asked to publish, escalate to Brady.
@@ -132,6 +144,7 @@ Set this environment variable in all CI build steps to prevent the build script
 | GITHUB_TOKEN can't trigger downstream workflows (v0.9.4) | squad-npm-publish.yml never fires | Manual `gh workflow run` or use PAT/GitHub App token (see below) |
 | Lockfile integrity check rejects workspace packages (v0.9.4) | False failures in squad-npm-publish.yml | Only validate packages resolved from npm registry (`startsWith('https://')`) (PR #1044) |
 | `prebuild` version bump breaks workspace linking (v0.9.4) | Local builds fail after bump-build.mjs runs | `git checkout -- package.json packages/*/package.json` then fresh install |
+| `dev` and `main` silently drift to unrelated histories (2026-08-13) | `dev` → `main` promotion PR is flatly unmergeable, no reviewable diff | Run `git merge-base upstream/dev upstream/main` BEFORE starting release-prep; see Rule 0 above |
 
 ## v0.9.4 Incident Learnings
 

@@ -22,6 +22,7 @@
     "Artsdalen",
     "Aspire",
     "astro",
+    "astropy",
     "Automator",
     "Baer",
     "benchmarkfn",
@@ -117,6 +118,7 @@
     "monorepo",
     "MSAL",
     "MSMQ",
+    "mwaskom",
     "myaccount",
     "mycompany",
     "myorg",
@@ -142,6 +144,8 @@
     "PRD",
     "Pruthviraj",
     "Pydantic",
+    "pydata",
+    "pylint",
     "pypi",
     "pyproject",
     "pytest",
@@ -180,7 +184,10 @@
     "Sturtevant",
     "subagent",
     "subagents",
+    "swebench",
     "swnger",
+    "Sympy",
+    "sympy",
     "Tamir",
     "tamirdresher",
     "telemetry",
@@ -221,6 +228,7 @@
     "Worf",
     "worktree",
     "worktrees",
+    "xarray",
     "ZUAER"
   ],
   "dictionaries": [