Commit Graph
8 Commits
Author SHA1 Message Date
retoor 67d4806d82 v1.44.0: extensive debug logging behind --debug
- Rules first in logging.md: R1-R12 plus a 27-event registry, every
  site validated against the source before implementation.
- DebugLog: file/stdout/stderr sinks, sentence/json formats,
  wipe-on-first-write for roots, append for TAI_CHILD, per-row
  commit, prune to 20000 rows, never raises.
- Vault logs table with six indexes plus FTS triggers, sealed at
  rest, read back through Store.log_query.
- Hooks: turns, completions (backend, model, ms, tokens, texts),
  tool dispatch, approvals, vault mutations, scheduler, serve,
  CAP turns, telegram updates, uncaught exceptions.
- Secrets never logged: redact plus LOG_MASKED_ARGS.
- 13 new tests (1222 + 16 green).
2026-10-18 21:53:03 +02:00
retoor 66ed0f9ea6 MCP both directions: client bridge plus --serve-mcp (1.42.0)
- client: stdio + Streamable HTTP transports, 2025-11-25
  handshake/list/call/ping, vault mcp_servers with sealed headers,
  per-profile enable, description-hash pins, tool cache
- mcp_normalize: pasted JSON/TOML configs (type repair, ${VAR}
  expansion from env then vault secrets, allow/deny lists);
  github URLs resolve via registry/package/README with live
  handshake verification before storing
- bridge: enabled tools enter lazy select_tools as
  mcp_<server>_<tool>, dispatch fallback, approval like non-safe
  shell, drift re-prompt, 64KB spill, per-call audit
- 7 agent tools: mcp_add/list/enable/disable/remove/tools/call
  (native count 42->60 in README, was already stale at 53)
- server: --serve-mcp KEY (+ --mcp-port/--mcp-bind) serves all 60
  native tools plus chat (agent turn, per-session history) and
  completions (stateless, current model) over single-threaded
  HTTP with Bearer auth, sessions, idle expiry; yolo documented
- rotation + plaintext-boot refusal cover sealed headers
- mcp.md contract with drift tests; 47 new tests; live loopback
  verified (client->serve over real HTTP)
2026-10-18 19:46:55 +02:00
retoor fa4e638a5c Vault-backed settings catalog with three-function API (1.41.0)
- SETTINGS_CATALOG: 28 env-mapped settings with type, default,
  required/nullable, strict options, preferred (reserved), encrypted
  flags, per-key bool word sets; routing split into
  CONFIG_SETTING_FIELDS (Config build) and SETTING_EXPORT_KEYS
- Store.get_or_create_settings / get_all_settings /
  upsert_settings: global+profile scopes, atomic strict validation,
  canonical forms, created/updated timestamps, env/db/default source
  reporting; encrypted values sealed at rest, rotation-aware,
  plaintext-boot refusal covers the table
- boot_settings in boot() and default_runner: Config second pass
  (env wins, CLI --no-opencode wins) plus env export for peripheral
  and applied keys; applied keys printed at boot
- settings.md contract with full catalog and exclusion rationale;
  16 SettingsTests incl. catalog self-check and doc drift guard
2026-10-18 14:15:35 +02:00
retoor 789f5206c4 docs: skills.md contract for the skill system and bundled set
- discovery/frontmatter contract incl. the bare->/| subset limit
- bundled table with upstream provenance and adaptation rules
- install target semantics, sidecar shape, maintenance workflow
- wired in AGENTS.md, README layout; drift test in BundledSkillsTests
2026-10-18 13:51:44 +02:00
retoor cc1c130b51 CAP protocol both directions: agent speak plus orchestrator drivers (1.39.0)
- Agent side: --cap-manifest plus --cap-speak stream-json binding with
  transcript-diff frames, diff/transcript artifacts, session config
- Orchestrator: PTY driver (settle heuristics on PtyProc/TuiVtScreen),
  stream-json, ACP-stdio (permission/elicit/fs/terminal serving) and
  A2A drivers (dual draft/current spelling, SSE mapping)
- Sessions, fleets: worktrees, master plans, mediated routing,
  budget refusal, audit log, research auto-answerer, --cap CLI
- cap.md contract with drift tests; research/cap-protocol archive
- Suites 1119 + 16 green
2026-10-18 13:17:35 +02:00
retoor af080fe68e Strict goal verification, approval hardening, retrieval eval, work limits, sandbox caps (1.38.0)
- Approval: re-prompt unknown answers (max 3), shell-only affirmative
  guidance recovery; destructive actions stay strict
- GoalVerifier subagent with strict JSON verdicts, bounded follow-up
  rounds, nonzero exit plus open turn on unverified one-shots
- Retrieval eval (Hit@K/MRR/nDCG/Context Precision), FTS cascade with
  title weighting, query/read event logging, --eval-retrieval CLI
- Supervisor caps (8 parallel, 64 records) and scheduler hardening
  (4 fires/tick, per-schedule lease, stated misfire policy)
- Sandbox box resource trio: 2g/2.0/256 at create, best-effort
  update for existing boxes, rm -f recovery pointer
- Research archive: 12 subjects, 667 sources, 27 conclusions;
  suites 979 + 16 green
2026-10-18 12:13:24 +02:00
retoorandClaude Sonnet 5 9dbd1fe667 Add plexer subsystem, container test infra, and last-resort opencode box backend
- Plexer: always-active terminal multiplexing. NativePlexerController/
  TmuxPlexerController share one interface, chosen by which env var is
  already set (TAI_PLEXER_SOCKET vs TMUX_PANE); NativePlexerBackend is
  the pty.fork()-based wrapper (raw mode, SIGWINCH resize forwarding,
  a Unix-socket control protocol, continuous session logging, a
  Ctrl+A prefix key distinct from tmux's own Ctrl+B). Auto-wraps new
  interactive shells via a .bashrc hook mirroring the existing
  command-not-found install pattern, gated on $TMUX/$TAI_PLEXER_ACTIVE/
  a $HOME/.tai-no-plexer bypass. New plexer_send/read/type/status
  tools, env-gated on a detected controller.
- tests/container.py + docker-compose.yml + tests/docker/Dockerfile:
  a disposable podman/docker image for the tests that must never touch
  a real machine (installing the plexer hook into a real ~/.bashrc, a
  live tmux session). Required suite (test_seal.py, test_tai.py,
  tests/plexer.py, tests/bashrc.py) gates the result; tests/rag.py
  runs as best-effort since its real-LLM dependency can't be satisfied
  inside a disposable container with no nested engine.
- tests/bashrc.py: real end-to-end hook install + bash -i auto-wrap,
  anti-recursion, tmux-active skip, non-interactive skip, a live
  TmuxPlexerController test, and an install/uninstall round-trip.
- tests/plexer.py: NativePlexerBackend under a real pty, plus a
  dedicated robustness pass (file permissions, >64KB payload framing,
  a malformed control request, a genuine self-deadlock fix in the
  write-to-child path).
- tai_test.py: the shared async process-driving framework (pipe/pty
  transport, group-wide kill-and-reap, a StdinBridge for --interactive)
  both test files build on.
- ChatClient last-resort backend: when there's no local opencode CLI
  but the sandbox box is already running, install opencode into it
  (official Node tarball, not apt -- apt's nodejs/npm pull in
  webpack/babel/jest as transitive deps) and route through it instead
  of giving up. Only reachable after the first retry round already
  failed everything else; never creates the box itself. TmuxPlexerController.status()
  fixed to use `list-panes` instead of `display-message -p`, which
  silently fell back to the default pane instead of erroring against
  a dead target.
- Fix 4 pre-existing test_tai.py bugs surfaced by running it in a
  clean environment for the first time: two stale mock signatures
  (subprocess.run's stdout= kwarg, Chat.complete's deadline= kwarg),
  one $TMUX-dependent catalog assertion, one tty-dependent color
  assertion.
- research/: a sourced research archive (Source Dossier/Citation
  Card/Conclusion tiers) backing the above design decisions, with its
  own audit.py consistency checker.
- .gitignore: fix a blanket *.md rule that was about to silently drop
  the entire research/ archive from version control; add __pycache__/
  and the one-off debug transcript this session started from.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
2026-10-10 04:36:05 +02:00
retoor ac189e9289 Add .gitignore for stray artifacts (screenshots, HTML scratch files, state) 2026-10-07 06:41:33 +02:00