chore(release): bump version to v0.3.0; update README

- Version 0.3.0 marks first release with 100% unit test coverage,
  full CI enforcement (--cov-fail-under=100), and SENTINEL security
  patches across all Medium/High findings
- README title updated to v0.3.0

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

SENTINEL security audit + 5 hardening patches (Stage 4)

Patches applied from SENTINEL v2.3 audit of MultAI v0.2.26040636:
- F-001: Wrap prompt in <USER_PROMPT_START/END> delimiters in agent fallback task string
- F-003: Clear OS clipboard after paste injection to prevent residual exposure
- F-007: Pin engine/requirements.txt to exact versions (was >=, now ==)
- F-008: Log warning when agent fallback transmits prompt to external LLM API
- F-005: Add security comment recommending dedicated Chrome profile for CDP isolation

Findings: 0 Critical, 1 High (architectural CDP scope — mitigated by comment),
4 Medium (all patched), 3 Low, 2 Clean. Deployment: Deploy with mitigations.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

0.2.26040304 Alpha: report viewer polish, CI fix, gitignore cleanup

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

0.2.26040202 Alpha: login retry + Perplexity fix + platform-level fal…

…lback

- Login retry: needs_login platforms are retried after a 90s sign-in window,
  not permanently skipped; other platforms' results already collected during wait
- Perplexity: guard against "Computer" feature in configure_mode (model picker
  and research toggle both explicitly skip any option containing 'computer');
  inject_prompt prefers textarea (new Perplexity UI) over contenteditable
- Platform-level browser-use fallback: full_platform_run() added to
  AgentFallbackManager; kicks in for STATUS_FAILED platforms when API key set;
  uses 25 agent steps (DEEP) / 15 (REGULAR) for complete lifecycle retry

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

0.2.26040201 Alpha: redesign /consolidator as standalone skill + expo…

…se publicly

- Rename skill: multi-ai-consolidator → consolidator (fixes display name)
- Add Generic mode: consolidates any content sources (docs, transcripts, notes,
  URLs, pasted text) with auto-derived report structure per content type
- Preserve AI-Responses mode: CIR generation from orchestrator/specialist-skill
  archives unchanged — no behavioral change for existing MultAI workflow
- Add Phase 0 mode detection; Phase 3 structure determination; Phase 7 self-improve
- Expose /consolidator in README alongside /multai and /comparator
- Add CHANGELOG entry

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

0.2.26040105 Alpha: redesign /comparator as standalone skill + expose…

… publicly

SKILL.md rewrite (7 gaps addressed):
- Gap 1: Evidence discovery phase — CIRs, working-folder docs, LLM knowledge,
  user description; triage by confidence level; CIR not required
- Gap 2: build.json auto-derived from available evidence in Phase 5
- Gap 3: Phase 3 — interactive priority assignment (optional, 'auto' to skip);
  explains weights (Critical=5× … Low=1×) before asking
- Gap 4: Phase 4 generalised from "Process CIR" to "Process Evidence" —
  handles CIR Variant A/B, non-CIR docs, and LLM knowledge with confidence labels
- Gap 5: "compare X vs Y" as first-class operation with capability framework
  discovery (Phase 2) and full scoring flow
- Gap 6: Phase 7 — Markdown summary always produced: ranked scores table,
  category-by-category breakdown, key differentiators, shared capabilities,
  evidence quality
- Gap 7: Domain knowledge fully optional — Phase 1 skips gracefully if no
  domains/ file; Phase 8 creates it from scratch on first run

README: surface /comparator as a user-facing skill alongside /multai;
update project structure comment; expand comparison matrix section

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

0.2.26040104 Alpha: bump version

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

0.2.26040102 Alpha: adopt YYMMDDNN versioning scheme (semver-compliant)

Replace letter suffixes (A, B) with two-digit daily counter (01, 99).
Update plugin.json, pyproject.toml, CHANGELOG, and CICD-Strategy-and-Plan.md.
Keeps full semver validity for Claude plugin system version comparison.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

0.2.260401A Alpha: rename orchestrator skill to /multai

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

0.2.260331A Alpha: fix 7 orchestration issues

1. SKILL.md: ban Claude-in-Chrome; document --followup routing
2. Sign-in detection: is_sign_in_page() + STATUS_NEEDS_LOGIN (🔑)
3. Agent fallback broadened: navigation failure, click_send, configure_mode
4. Pre-flight: warn-only — never skip platforms based on rate budget
5. Global timeout: dynamic (max_plat + stagger + 60s buffer) so last
   staggered platform always gets its full per-platform timeout
6+7. Tab reuse: --followup flag reuses open conversations; new-topic
   runs reuse tabs but navigate to fresh conversation; tab state
   persisted in ~/.chrome-playwright/tab-state.json
   PLATFORM_URL_DOMAINS added for tab matching
Tests: UT-OR-12 (--followup), UT-CF-09 (URL domains), UT-CF-10 (needs_login)
Test count: 96 → 98

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>