captchaforge 0.2.39

Captcha detection and solving for Firefox and BiDi-driven browsers. Detection, vendor solver scaffolding, trusted cross-origin click delivery into nested OOPIFs, and stealth personas are implemented and tested; broad live-vendor solve rates are not yet benchmarked.
Documentation
/target/
**/target/
/Cargo.lock
**/*.rs.bk
.DS_Store

# Bench reporter outputs (regenerated by every captchaforge-bench run).
bench-results-*.csv
bench-results-*.html
bench-results-*.json
bench-results-*.md
bench-results-*.sarif
bench-results.json

# Trained model weights (large; ship via HuggingFace / Ollama / S3).
training/weights/
training/datasets/


# Python __pycache__ artefacts under tools/.
tools/__pycache__/

# ── Santh-internal artifacts (BEGIN, managed; do not edit between markers) ──
# Operator-private working notes, plans, audits, scoreboards, status
# trackers, and "depth" markers leak workflow context and must never
# ship publicly. The only governance artifacts that DO ship publicly are
# CODE_OF_CONDUCT.md, SECURITY.md, CONTRIBUTING.md, and LICENSE-*, those
# are explicitly allow-listed. Everything else internal is denied here.

# Internal working dirs
.internal/
jules_tickets/

# AI/agent tool artefacts (operator-private)
CLAUDE.md
AGENTS.md
AGENT.md
agents.md
GEMINI.md
.claude/
.agents/
.codex/
.cursor/
.cursorrules
.windsurf/
.windsurfrules
.aider*

# Plans, TODOs, audits, backlogs, scoreboards, "depth" markers
*DEPTH*
*_PLAN.md
*_PLAN_*.md
*_AUDIT.md
*_AUDIT_*.md
*_BACKLOG.md
*_BACKLOG_*.md
TODO.md
NOTES.md
SCRATCH.md
*_TODO.md
*_NOTES.md
ENGINE_GAPS.md
SCOREBOARD*.md
*_status_*.md
.audits/

# Allow-list: governance artifacts that ARE public
!CODE_OF_CONDUCT.md
!SECURITY.md
!CONTRIBUTING.md
!LICENSE
!LICENSE-APACHE
!LICENSE-MIT
# ── Santh-internal artifacts (END) ──
# Unified backlog (STANDARD.md)
/BACKLOG.md
/PLAN.md
/IMPROVEMENT_PLAN.md
/MASTER_PLAN.md
/*_PLAN.md
/*PLAN*.md