Skip to content
Chimera

Chimera 0.6.0 — Wave-13 + Wave-14 Capability Roundup

Chimera 0.6.0 — Wave-13 + Wave-14 Capability Roundup

Section titled “Chimera 0.6.0 — Wave-13 + Wave-14 Capability Roundup”

Released: 2026-05-07

The 0.6.0 release rolls up wave-13 (15 G-tracks, 6 E-tracks, 3 B-tracks) and wave-14 (10 tracks) on top of the seven-CLI family that 0.5.0 shipped. Every change is purely additive — no breaking API changes, no removed flags, no renamed slash commands. The v0.5-era surface is preserved bit-for-bit.

What’s new is breadth-and-depth filling on the seven-CLI surface: a new apply_patch write tool, 18+9 lifecycle hook events with event-filter matching, the five-mode permission standard alongside the legacy four-choice surface, a content-addressed file-undo store, declarative permission rules, the Ctrl-X chord layer, plan mode, instruction-file ingestion (AGENTS.md / CLAUDE.md), 13 algorithm skills, the swe-agent agent preset, three new benchmarks, and the register_argument auto-promotion safety net wired across every CLI.

Tools and hooks

Section titled “Tools and hooks”
  • apply_patch tool. Diff-format file editing for agents that prefer unified-diff output to a structured Edit tool. Lives at chimera/tools/apply_patch.py. (W13 G1.)
  • Four more high-leverage tools. write_guard, notebook_edit, worktree, cron. (W13 G13.)
  • 27 lifecycle hook events. 18 standard events plus 9 fine-grained wave-14 additions covering compaction lifecycle, subagent start/stop, structured tool-call telemetry, and session-state transitions. (W13 G4 + W14-7.) The HookMatcher now accepts an event filter so a single matcher can listen on a subset of an event category rather than firing for every event in that category.

Permission system

Section titled “Permission system”
  • Five-mode permission standard. read-only, suggest, auto, yolo, strict wired across every CLI’s --permission-mode flag on top of the legacy four-choice surface (default, acceptEdits, bypassPermissions, plan). Both sets accepted; legacy values keep their semantics. (W13 G3.)
  • Declarative permission rules. ~/.chimera/permissions.json loaded at startup. Rule shape: {tool, action, arg_key?, arg_pattern?, description?}. Manage via /permissions list|add|remove in the otter REPL. (W13 G6.)

Editor / TUI ergonomics

Section titled “Editor / TUI ergonomics”
  • Ctrl-X chord layer. Vi-style chord prefix (Ctrl-X p = plan mode toggle, Ctrl-X s = shell mode, etc.). (W13 G7.)
  • Plan mode. Boot any CLI’s REPL in planner posture; the agent emits a plan rather than acting. Toggle mid-session with /plan. Plans persist to ~/.chimera/plans/ for /resume. (W13 G7 + stoat W14-3.)
  • File-undo store. Content-addressed per-turn snapshots under ~/.chimera/snapshots/<session>/ so otter’s /undo rewinds conversation and files. (W13 G5.)
  • Multi-step /undo --steps N. The snapshot store could always do multi-step; v0.6 wires the slash dispatcher to walk it. Forms: /undo --steps 3 / /undo -n 3 / /undo 3. The symmetric /redo --steps N works the same way. Drains to baseline cleanly when N exceeds the stack. (W14-9.)
  • Bracketed paste, sessions slash, mode chords. stoat REPL UX. (W14-3.)
  • 12-slash palette. badger gained 12 new slash commands. (W14-4.)
  • 5 print-mode flags. weasel’s -p print mode gained five new flags for output shape, attachment, and structured emission. (W14-5.)
  • Quality layer. shrew gained model profiles, output parser, quality monitor, and skill injector. (W14-6 + W13 G11/G12.)

Project ingestion

Section titled “Project ingestion”
  • Instruction-file ingestion. Every CLI now ingests AGENTS.md and CLAUDE.md from project root and merges them into the system prompt. (W13 G2.)
  • 13 algorithm skills. Curated SKILL.md files under chimera/skills/algos/ covering common algorithm patterns. (W13 G12.)

Subagents and slash

Section titled “Subagents and slash”
  • Subagent registry. chimera/agents/registry.py extended with a proper register/get/list/load_directory surface; swe-agent agent preset added. (W13 G8.)
  • /resume slash + /diff slash. Top-level /resume and /diff slash commands across all 7 CLIs. (W13 G9.)
  • help-long auto-promote across all 7 CLIs. register_argument is now wired in mink, otter, ferret, weasel, shrew, stoat, badger. Future contributors who paste a verbose help= string can no longer silently bloat --help past the 50-line ceiling. (W13 E6 + W14-9.)
  • cc-alias deprecation prep. Clearer stderr banner naming both the rename version and the v0.7.0 removal version. (W13 E4.)

Benchmarks and evaluation

Section titled “Benchmarks and evaluation”
  • ProgramBench inference loop. Full inference loop harness for ProgramBench, replacing the wave-13 scaffold. (W14-8.)
  • MultiSWE-bench scaffold. Runnable adapter, not a stub. (W13 B2.)
  • Three more benchmark scaffolds. See chimera/eval/benchmarks/. (W13 B3.)
  • Provider catalog refresh. 7 model families recognised by the pricing/cost subsystem; PyPI release prep groundwork. (W13 E2 + E5.)

CLI-by-CLI

Section titled “CLI-by-CLI”
CLIWave-13 highlightsWave-14 highlights
mink5-mode permissions; settings keys (G14)9 hook events + settings apply (W14-7)
otterapply_patch, file-undo, perm rules, plan modeTier-1 polish: worktree, stats, export, import, skills (W14-2)
ferretcodex flags, instruction files (G15)Codex-style subcommands (W14-1)
weaselJS/TS exec, RPC streaming, CodingAgent default (G10)5 print-mode flags (W14-5)
shrewskill injector (G11), 13 algo skills (G12)Quality layer: 4 features (W14-6)
stoatchord+plan mode (G7)Hooks engine + UX features (W14-3)
badgerparity refresh + provider chainSlash palette: 12 commands (W14-4)

Quality

Section titled “Quality”
  • Tests: 7665 passing, 62 skipped (uv run pytest --ignore=tests/integration -q --tb=line 2>&1 | tail -1 → 7665 passed, 62 skipped, 13 warnings in 138.43s, measured 2026-05-07). Up from the v0.5 baseline of 3922 with wave-13 + wave-14 coverage (multi-step undo, hooks, apply_patch, plan mode, file-undo, declarative permissions, print-mode flags, badger slashes, ProgramBench, etc.). The per-CLI cross-section (tests/cli/ tests/otter/ tests/weasel/ tests/shrew/ tests/stoat/ tests/badger/ tests/mink/ tests/ferret/) reports 3166 passing, 10 skipped post-W14-9 (verified → 3166 passed, 10 skipped, 1 warning in 88.26s). Live-integration tests under tests/integration/ (526 collected) are excluded from these counts because they spend real provider tokens.
  • Lint: uv run ruff check chimera/ → All checks passed!.
  • Mypy: uv run mypy chimera/ → Success: no issues found in 639 source files.
  • Trademark scrub: all 7 per-CLI scrubs clean — bash scripts/all_trademark_scrub.sh exits 0 (passed: 7 (mink otter ferret weasel shrew stoat badger); failed: 0).

Upgrading from 0.5.x

Section titled “Upgrading from 0.5.x”

See Migrating from v0.5 to v0.6 for the step-by-step guide. The summary:

  • No breaking API changes. Every flag, every slash command, every public API works exactly as in v0.5.
  • No code-level migrations required. The AgentPreset.build() → CodingAgent.from_preset(...) rewrite from v0.5 is still the only recommended migration; it remains a DeprecationWarning until v0.7.0.
  • Per-CLI --version bumped from 0.5.0 to 0.6.0. Scripts pinning exact-string match against 0.5.0 need a one-line bump.
  • Optional adoption: apply_patch tool, declarative permission rules, instruction-file ingestion, plan mode. None auto-launch without an opt-in surface (a flag, a file, or a tool registration).

Deprecation timeline (unchanged from v0.5)

Section titled “Deprecation timeline (unchanged from v0.5)”
SymbolStatus in v0.6Removed in
chimera cc subcommand aliasDeprecationWarning (stderr line)v0.7.0
AgentPreset.build()DeprecationWarning (Python warning)v0.7.0
chimera.agents.presets.agent_styles moduleDeprecationWarning via AgentPreset.build()v0.7.0
claude_code preset keyDeprecationWarning (use coding_agent)v0.7.0

Nothing on that list moved. The v0.5-to-v0.6 gap is one minor version, not the larger v0.5-to-v0.7 horizon.

Section titled “Links”