refactor!: remove dead packages with no importers

README.md

@@ -269,9 +269,6 @@ Features adopted from open-source agent projects. All are off by default unless
 | App verification | `AppVerify` tool | Boot-smoke check with readiness polling and evidence artifacts |
 | Media generation | `GenerateMedia` tool | Image/video generation with local persistence. Backend via `tool.SetMediaEngine`; an OpenAI-compatible client ships in `eyrie/client` (`ImageClient`), wired by the host (boundary-guarded — hawk routes through the eyrie facade) |
 | Voice transcription | Telegram voice notes + `stt` package | Transcribe Telegram voice/audio into the prompt. Backend via `stt.SetTranscriber`; an OpenAI-compatible client ships in `eyrie/client` (`AudioClient`), wired by the host |
-| Git-tree file snapshots | `internal/gitsnapshot` | Content-addressed tree capture/diff/preview/restore |
-| Turn-boundary rewind | `internal/filestate` | Per-prompt before/after snapshots with durable store |
-| Continual harness | `internal/intelligence/harness` | Versioned, evidence-backed refinement of supplemental prompts/memories/skills/subagents with rollback |
 | Bounded autonomous budgets | `internal/engine` (`AutonomousBudget`) | Track turns/tokens/time/continuations; report why a run stopped (budget vs gate-passed vs error) |
 | Agent family messaging | `internal/multiagent` (`FamilyMessenger`) | Direct parent/sibling/child messages with pending caps + rate limits |
 | Path reservations | `internal/multiagent` ledger | Detect overlapping-file changes between parallel branches |
@@ -281,7 +278,6 @@ Features adopted from open-source agent projects. All are off by default unless
 | Token-cheaper file views | `Read` tool `--minify` | Read-only, comment-stripped, whitespace-dense file view (Go via `go/parser`; other languages string-aware; never touches disk) — fewer tokens per read |
 | Classified provider hints | `internal/errhint` | Buckets provider errors (Auth/RateLimit/Connectivity/ModelNotFound/ContextOverflow) into a one-line fixable next step. Wired into TUI error rows (`friendlyErrorMessage`) and `hawk exec` CLI errors |
 | Atomic install transactions | `internal/installtxn` | Cross-process staged install/remove with rollback. Wired into skill install (atomic `SKILL.md` publish) |
-| Stale-lock reclaim | `internal/lockutil` | Race-correct atomic reclaim of O_EXCL lock files with live-restore (ready for O_EXCL lock sites) |
 | Test command discovery | `internal/testrunner` | Auto-detect test/verify commands (Go/npm/bun/pnpm/yarn/pytest/cargo) and parse runner output into structured results. Wired into `hawk verify` |
 | Circuit breaker | `internal/circuitbreaker` | Closed/open/half-open retry-storm protection with cooldown. Wired into auto-compaction (cooldown + half-open auto-retry) |
 | Smart turn routing | `internal/smartrouting` | Deterministic simple/strong turn classifier with fail-toward-strong safety. Wired into per-turn model selection (`settings.smart_routing`) |

docs/COMPETITIVE.md

@@ -16,7 +16,7 @@ Verified from source in this repo:
 - Custom providers: supported — `internal/config/settings.go:50` (`custom_providers`), `internal/config/eyrie_engine.go:32-50`.
 - Unwired backends: `internal/tool/computer_use.go:67-94` (`SetComputerBackend`, nil default), `internal/tool/media_generation.go:69-71` (`SetMediaEngine`, nil default).
 - Terminal detect covers kitty/ghostty/wezterm/alacritty names (`internal/ui/icons/detect_test.go:56`); Kitty graphics protocol not implemented (see `docs/plans/pi-adoption-plan.md:25`).
-- Bench infra exists (`internal/bench/suite.go`, `internal/feature/eval/`, `make bench`) but README publishes no numbers.
+- Bench infra exists (`internal/feature/eval/`, `make bench`) but README publishes no numbers.
 
 External star counts below are approximate web-search snapshots (2026-09-08), not repo-verified. Treat as order-of-magnitude traction, not exact rankings. hawk is pre-release (`VERSION`: `0.0.1`, `README.md:40-44` source-build primary) — it competes on architecture, not stars.
 

docs/design/ECOSYSTEM-MARKETPLACE.md

@@ -231,7 +231,7 @@ Concrete reuse map — what is *already there* and what each piece becomes:
 | Hook bundling | `ManifestHook{Event,Command,Priority}` | `manifest_v2.go:32` |
 | MCP server bundling | MCP client/loader | `hawk/internal/mcp/mcp.go`, `hawk/internal/tool/mcp_tool.go` |
 | Sub-agent bundling | Persona system (YAML frontmatter MD) | `hawk/internal/multiagent/agents/` (cf. `TOP20_COMPARISON.md:62`) |
-| Rule/skill discovery precedence | `DefaultRuleSources`, `RuleDiscoverer` | `hawk/internal/context/rules.go:20`, `:46` |
+| Rule/skill discovery precedence | `DefaultRuleSources`, `RuleDiscoverer` | `hawk/internal/plugin/` rule discovery |
 | Trust — static scan | `malware_check.go` (blocked/suspicious regex) | `hawk/internal/plugin/malware_check.go:19` |
 | Trust — unicode/homoglyph | `audit.go` (`AuditFinding`, severities) | `hawk/internal/plugin/audit.go:12-22` |
 | Registry generators/validators | Python tooling | `starling/tools/{update_registry,registry_schema,validate_skill,package_skill,sync_marketplace}.py` |
@@ -366,7 +366,7 @@ The estimate is bounded on the low side because the hardest plumbing — the reg
 - `hawk/internal/plugin/malware_check.go:19`
 - `hawk/internal/plugin/audit.go:12`
 - `hawk/internal/tool/skill.go:62,91`
-- `hawk/internal/context/rules.go:20,46,97`
+- `hawk/internal/plugin/` rule discovery
 - `hawk/cmd/skills_cmd.go:16-211`
 - `hawk/internal/mcp/mcp.go`, `hawk/internal/tool/mcp_tool.go`
 - `TOP20_COMPARISON.md:65,74,83,181,227,232,240,241`

docs/design/HAWK-CLOUD-SAAS.md

@@ -315,7 +315,7 @@ This is the crux: **what is reusable today vs. net-new.**
 | Sandbox executor interface | `internal/sandbox/container.go:18-21` (`containerExecutor`: `Exec`, `Running`) | `CloudSandbox` implements the same interface → drop-in. Callers don't know if they're on local Docker or a cloud microVM. |
 | Sandbox lifecycle manager | `internal/sandbox/snapshot_sandbox.go:52-228` (`Create/Pause/Resume/Snapshot/Restore/List/Cleanup`) | Existing pause/resume/snapshot semantics map cleanly onto E2B/Daytona pause+snapshot APIs; the manager abstraction guides the `CloudSandbox` API shape. |
 | Messaging gateways | `internal/daemon/gateway.go`, `telegram.go`, `discord.go`, `slack.go` | Already forward to `/v1/chat` via `forwardToHawk` (`gateway.go:17`) with bearer auth. In cloud they forward to the tenant-scoped chat endpoint with the org's key — minimal change. |
-| Cron engine | `internal/system/cron/cron.go` | Foundation for cloud scheduled runs (Routines), per `TOP20_COMPARISON.md:48`. Out of scope here but shares the worker plane. |
+| Cron engine | (not yet implemented) | Cloud scheduled runs (Routines) are planned, per `TOP20_COMPARISON.md:48`. Out of scope here but shares the worker plane. |
 
 ### Net-new (must build)
 
@@ -387,7 +387,7 @@ credits hit zero.
 - **M2.3** JetBrains extension (ACP).
 - **M2.4** Org-scoped provider config (BYO keys / BYO model endpoints), aligning with eyrie multi-tenant proxy (`TOP20_COMPARISON.md:96`).
 - **M2.5** Audit logging, IT-managed policy tiers, self-hosted/on-prem distribution.
-- **M2.6** Cloud Routines (scheduled/triggered runs) on the worker plane via `internal/system/cron/cron.go` (`TOP20_COMPARISON.md:48`).
+- **M2.6** Cloud Routines (scheduled/triggered runs) on the worker plane (`TOP20_COMPARISON.md:48`).
 
 ---