chore: replace python -m agents.validator with hawk in all agent mode instructions

README.md

@@ -518,62 +518,6 @@ tools: `tool_create`, `tool_list`, `tool_get`, `tool_update` (including enable/d
 the same engine that powers virtual tools. Self-evaluation depth is bounded so a tool cannot loop by
 calling itself.
 
-## Maintenance agents
-
-DevPlace ships a fleet of autonomous AI maintenance agents under `agents/` that
-keep the codebase consistent. Each agent owns one quality dimension, sweeps the
-whole repository for problems in that dimension, and either reports them or fixes
-them and verifies the build. The internals are documented in `AGENTS.md`; the
-public, newbie-friendly guide is at `/docs/maintenance-agents.html` and
-`/docs/maintenance-usage.html`.
-
-Every agent runs in two modes: `--fix` (default, autonomous) or `--check`
-(report only, non-zero exit on findings). The agents share one proven engine
-(`agents/agent.py`) and verify their own work with `agents/validator.py`, a
-dependency-free validator (Python `ast`/`py_compile`, JavaScript `node --check`,
-CSS brace balance, HTML and Jinja templates parsed with Jinja2).
-
-| Agent | Dimension |
-|-------|-----------|
-| `maestro` | Conversational conductor; runs the right agent or the whole fleet for you |
-| `security` | Route authorization, private-data and admin guards, input validation |
-| `audit` | Every state change leaves an audit-log entry |
-| `devii` | Devii capability parity and role-gated tool visibility |
-| `docs` | CLAUDE.md / AGENTS.md / README / `/docs` accuracy and role-aware visibility |
-| `fanout` | A feature is wired across every layer (form, schema, Devii tool, docs, SEO) |
-| `dry` | Duplication removed; shared helpers reused |
-| `style` | Naming, headers, typing, formatting rules |
-| `frontend` | ES6 modules, web components, CSS structure |
-| `seo` | Search metadata and sitemap coverage |
-| `test` | Integration-test coverage (writes tests, never runs the suite) |
-
-```bash
-make maestro                 # talk to the conductor; it runs the rest
-make validate                # validate the whole codebase (no external tools)
-make audit-agent             # fix one dimension autonomously
-make audit-agent CHECK=1     # report only
-make agents-all              # run the whole fleet
-make agents-all CHECK=1      # CI gate across the fleet
-make maintenance             # whole fleet, read-only, only git-changed files (fast)
-make maintenance-fix         # whole fleet, fix mode, only git-changed files (fast)
-```
-
-`make maintenance` and `make maintenance-fix` restrict the fleet to the files git
-reports as modified or new under `devplacepy/` and `tests/`, so checking and fixing
-work in progress is fast. In fix mode a hard write-allowlist guarantees no file
-outside that changed set is ever touched. The flag is `--changed` on the
-orchestrator and any single agent (`python -m agents.security --changed --check`).
-
-Every run is legible live: each agent prints a start banner (icon, name, a
-memorable codename, what it will do, and its report path), every line is prefixed
-with the wall-clock time and elapsed duration, each AI call streams its running
-dollar cost, and every file edit streams a unified diff. Reports land in
-`agents/reports/` as `<agent>-<codename>-<date>.{json,md}` (the orchestrator writes
-`fleet-<codename>-<date>.json`). Safety is enforced in code, not prompts: agents
-cannot edit anything under `agents/`, cannot shell out, and cannot modify more than
-a fixed number of files per run, and a run that exhausts its step budget is marked
-`incomplete` and exits non-zero rather than looking clean.
-
 ## Push notifications & PWA
 
 Authenticated users can receive native web push notifications, and the site is an