Merge pull request #31 from huggingface/agent-improvements

Fix premature agent loop exit in long-running autonomous tasks

feat: add headless CLI mode with local filesystem tools and rich terminal rendering

Deduplicate file re-reads via content hashing for local and sandbox tools

feat: LLM-based pruning for complex tool outputs (hf_jobs)

Add deterministic pruning of old tool outputs before LLM compaction

Add per-tool error limits and per-turn request cap

feat: add doom-loop detection for repeated tool call patterns

feat: add research sub-agent tool, slim down main agent system prompt

feat: render cancelled tool state in UI after user interrupt

feat: kill sandbox processes and cancel HF jobs on user interrupt

fix: auto-retry on truncated tool calls + correct activity status on refresh

fix: live event streaming after reconnection + LiteLLM timeout

fix: clear pending_approval before tool execution starts

fix: gracefully handle MCP and OpenAPI failures during startup

refactor: simplify tool call validation and make interrupts cancel tool execution

fix: correct context_length init and emit tool_call events for malformed calls

fix: handle interrupt during streaming tool calls without corrupting context

fix: recover malformed tool calls instead of crashing session

feat: upgrade CLI with local tools, slash commands, and interrupt support

fix: properly close SDK message on error, show tool errorText

fix: patch dangling tool calls on error, move undo to context manager

remove lmnr dependency

feat: cleanup sandbox spaces on session end and increase cookie TTL

feat: resolve sandbox file paths for hf_jobs script display

fix: inject OAuth token into MCP server headers at runtime

fix: route all HF token usage through session.hf_token from OAuth

feat: implement cooperative agent cancellation with cancel button

feat: add hf_papers tool for paper discovery and reading

feat: restore sandbox tools and proactive compaction

feat: merge HF Space improvements

adding compaction

sandbox working and agent can use it

add create sandbox tool

sandbox start

log streaming

deleted search code

adding whoami to system prompt

Add hf_repo_files and hf_repo_git tools

Cleaning up tools

Remove utils_tools and disable private_hf_repo_tool

Fix: look for hardware_flavor at top level of tool_args, not nested

Merge configurable-permissions: add confirm_cpu_jobs, auto_file_upload, keep yolo mode

Merge branch 'dataset_tools'

Unify dataset tools into single comprehensive hf_inspect_dataset tool

fixing logs

Add yolo mode for auto-approving tool calls

cleaned comments

Added 2 dataset tools, one to list rows, the other one to download the dataset

Add auto_file_upload and confirm_cpu_jobs config settings