docs: changelog for v2.7.14-backlog-hardening + v2.7.15-git-diff-panel

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

CHANGELOG.md

@@ -2,6 +2,14 @@
 
 All notable changes per release tag. Most recent on top, ordered by tag creation date (which matches the git history). Tag names follow `vMAJOR.MINOR.PATCH-slug` — the slug describes what shipped, so the tag name alone is enough to recall the batch.
 
+## v2.7.15-git-diff-panel — 2026-06-03
+
+A Files / Git tab in the right-side file panel (the file-browser sidebar) that shows the project repository's git diff and lets the user stage, unstage, commit, and discard whole files in-session — modeled on Paseo's diff view, scoped and planned through the `plan-a-feature` → `plan-implementation` skills, then built and audited via `paseo-epic` in an isolated worktree. Two comparison modes (Uncommitted vs HEAD, and the current branch vs its base — the upstream tracking branch else `origin/HEAD`), auto-selected by repo dirty-state on first open and pinned after an explicit choice; per-file expand/collapse with lazy Shiki `lang:'diff'` highlighting, +/- stats, and binary/too-large placeholders. All git read and write logic lives in `apps/server` (new `git_diff.ts` + routes on `projects.ts`) — the read-only-server posture governs the assistant's tools, not the user's own actions, and the container already mounts `/opt` read-write while `project_bootstrap` already commits via `execFile`. Every write uses the safe `execFile` argv pattern (never a shell string) with `--` operand separators, per-file `pathGuard` + realpath symlink-escape validation, server-derived `-c` commit identity (the request body is `.strict()` and carries no author fields), and the write endpoints are deliberately absent from the assistant tool registry. Reads are bounded (30s deadline, 10MB); an index lock or an in-progress merge/rebase/cherry-pick/bisect surfaces as "repository busy" and disables writes. The panel stays current via a client `git_diff_refresh` sessionEvent (no new wire contract) coalesced across tab-open, mutations, turn completion, and pending-change apply; discard is an irrecoverable hard-delete behind a plain confirm distinguishing a tracked revert from an untracked delete. New `git_diff` pure-helper + temp-repo integration tests (59 cases); server 630 tests green, web tsc clean. Pairs with `v2.7.14-backlog-hardening` (shipped together).
+
+## v2.7.14-backlog-hardening — 2026-06-03
+
+Five independent items from the second external-code-review backlog (`boocode_code_review_v2.md`), each built and audited as its own phase via `paseo-epic`. **External task-cancel** now actually works: Stop on an opencode/goose/qwen/claude task aborts the running child via a per-task `AbortController` registry reachable from the cancel route and finalizes the assistant message as `cancelled` — fixing two latent bugs (catch blocks left the message `streaming`; warm success-paths wrote `complete` on an aborted turn); warm pools/worktrees are preserved (abort the prompt only, never the pooled process) and the native boocode path is unchanged. **Parser prune**: the tool-call parser drops to its two load-bearing exports (eight zero-caller symbols unexported, a gate test added for the `<invoke>`-as-text fallback) with no live-path behavior change, and placeholder-rejection logging moves to pino. **BooChat stall-timeout**: a 90s per-chunk deadline wraps native inference's `fullStream` via `AbortSignal.any` so a hung local stream finalizes the message instead of hanging — no retry, since re-running re-emits already-streamed deltas (a pure `classifyStreamError` helper is added). **view_session_history**: a read-only MCP tool returning the newest-N transcript (role≠system) in chronological order. **Retire :9502**: the unused `apps/coder/web` fallback SPA is removed (package, static-serve block, build step, Dockerfile copy, `@fastify/static`), keeping every API/WS/health/MCP route. F1 added an optional `status` field to the shared `message_complete` contracts frame (so a deploy rebuilds `@boocode/contracts` first, as the sequence already does). Server 630 / coder 360 tests green.
+
 ## v2.7.13-contracts-ssot — 2026-06-02
 
 Creates `@boocode/contracts` (`packages/contracts`), a new workspace package that becomes the single source of truth for every cross-app wire contract — reversing the decision recorded in `v2.5.12-provider-lifecycle-phase4` that declined a shared types package as not worth the Docker/build-order risk at solo scale; a live `AgentSessionConfig` drift that had since appeared between `apps/coder` and `apps/web` justified the investment. Six contracts are now defined exactly once: the `WsFrameSchema` Zod runtime schema, the provider snapshot types (`ProviderSnapshotEntry` and family), the Zod provider-config schemas, `MessageMetadata` + `ErrorReason`, `AgentSessionConfig`, and `WorktreeRiskReport`; both Zod-backed contracts use `z.infer` so validator and type derive from the same definition and cannot drift independently. All four consumers — `apps/server`, `apps/web`, `apps/coder`, and the fallback SPA `apps/coder/web` — import via `workspace:*` through a per-subpath exports map consuming built dist only (no tsconfig project references); the hand-synced copies and their parity tests (`provider-types-parity.test.ts`; the ws-frames byte-parity assertion) are deleted while the KNOWN_FRAME_TYPES drift test and broker fail-closed tests are preserved. Build order is inverted in the root build script, Dockerfile, and coder deploy docs; `apps/coder/web`'s migration also removed dead `pending_change_*` reducer arms (no frame publisher exists for these — pending changes are HTTP-delivered), closing a latent missing-default-arm crash, and reconciled field-type conflicts with the canonical `WsFrame`; zod is pinned to a single version across the workspace. Server 543 / coder 293 / contracts 11 tests passing; human smoke verified on the live stack 2026-06-02.