[BOUNTY: 10 RTC] Add .trashclaw.toml project config with auto-loaded context files

[Scottcjn]

Bounty: 10 RTC

Task

Support a .trashclaw.toml (or .trashclaw.json) project config that auto-loads specified files into context on startup.

Requirements

• Read .trashclaw.toml from CWD on startup
• Support fields:
  • context_files = ["src/main.py", "README.md"] — auto-loaded into conversation
  • system_prompt = "You are a Rust expert" — appended to system prompt
  • model = "codestral" — override model name
  • auto_shell = true — skip shell approval
• Existing .trashclaw.md support stays (this is additive)
• Use stdlib tomllib (Python 3.11+) with JSON fallback for older Python

Why

Projects should be able to configure TrashClaw behavior per-repo. Drop a config file in your project and TrashClaw knows what to load.

Wallet: Drop your wallet name when you submit.

[adittq]

/apply

I'm interested in working on this bounty.

My approach:

1. Review the requirements thoroughly
2. Implement a high-quality solution
3. Test comprehensively
4. Submit with clear documentation

Relevant experience:

• python, automation, documentation, scripting, monitoring, test, bounty, RTC, agent, json, writing, translation, analysis, research, data, web, api, linux, bash, git, markdown, tutorial, blog, video, social, github, openclaw
• Track record of completing similar tasks
• Fast turnaround time (typically 2-4 hours)

I can start immediately. Let me know if you have any questions!

[jiarongmai]

I'll attempt this bounty! Project-level config support is a great feature.

My approach:

1. Read .trashclaw.toml\ or .trashclaw.json\ from CWD
2. Support fields: \context_files, \system_prompt, \model, \�uto_shell\
3. Use \ omllib\ (Python 3.11+) with JSON fallback
4. Auto-load specified files into context on startup
5. Keep existing .trashclaw.md\ support

Wallet: jiarongmai