Closes gaps that allowed an agent to expose Docker's Remote API to the
internet by writing to /etc/docker/daemon.json.
Terminal tool (approval.py):
- chmod: now catches 666 and symbolic modes (o+w, a+w), not just 777
- cp/mv/install: detected when targeting /etc/
- sed -i/--in-place: detected when targeting /etc/
File tools (file_tools.py):
- write_file and patch now refuse to write to sensitive system paths
(/etc/, /boot/, /usr/lib/systemd/, docker.sock)
- Directs users to the terminal tool (which has approval prompts) for
system file modifications
* feat: GPT tool-use steering + strip budget warnings from history
Two changes to improve tool reliability, especially for OpenAI GPT models:
1. GPT tool-use enforcement prompt: Adds GPT_TOOL_USE_GUIDANCE to the
system prompt when the model name contains 'gpt' and tools are loaded.
This addresses a known behavioral pattern where GPT models describe
intended actions ('I will run the tests') instead of actually making
tool calls. Inspired by similar steering in OpenCode (beast.txt) and
Cline (GPT-5.1 variant).
2. Budget warning history stripping: Budget pressure warnings injected by
_get_budget_warning() into tool results are now stripped when
conversation history is replayed via run_conversation(). Previously,
these turn-scoped signals persisted across turns, causing models to
avoid tool calls in all subsequent messages after any turn that hit
the 70-90% iteration threshold.
* fix: replace hardcoded ~/.hermes paths with get_hermes_home() for profile support
Prep for the upcoming profiles feature — each profile is a separate
HERMES_HOME directory, so all paths must respect the env var.
Fixes:
- gateway/platforms/matrix.py: Matrix E2EE store was hardcoded to
~/.hermes/matrix/store, ignoring HERMES_HOME. Now uses
get_hermes_home() so each profile gets its own Matrix state.
- gateway/platforms/telegram.py: Two locations reading config.yaml via
Path.home()/.hermes instead of get_hermes_home(). DM topic thread_id
persistence and hot-reload would read the wrong config in a profile.
- tools/file_tools.py: Security path for hub index blocking was
hardcoded to ~/.hermes, would miss the actual profile's hub cache.
- hermes_cli/gateway.py: Service naming now uses the profile name
(hermes-gateway-coder) instead of a cryptic hash suffix. Extracted
_profile_suffix() helper shared by systemd and launchd.
- hermes_cli/gateway.py: Launchd plist path and Label now scoped per
profile (ai.hermes.gateway-coder.plist). Previously all profiles
would collide on the same plist file on macOS.
- hermes_cli/gateway.py: Launchd plist now includes HERMES_HOME in
EnvironmentVariables — was missing entirely, making custom
HERMES_HOME broken on macOS launchd (pre-existing bug).
- All launchctl commands in gateway.py, main.py, status.py updated
to use get_launchd_label() instead of hardcoded string.
Test fixes: DM topic tests now set HERMES_HOME env var alongside
Path.home() mock. Launchd test uses get_launchd_label() for expected
commands.
Root cause: terminal_tool, execute_code, and process_registry returned raw
subprocess output with ANSI escape sequences intact. The model saw these
in tool results and copied them into file writes.
Previous fix (PR #2532) stripped ANSI at the write point in file_tools.py,
but this was a band-aid — regex on file content risks corrupting legitimate
content, and doesn't prevent ANSI from wasting tokens in the model context.
Source-level fix:
- New tools/ansi_strip.py with comprehensive ECMA-48 regex covering CSI
(incl. private-mode, colon-separated, intermediate bytes), OSC (both
terminators), DCS/SOS/PM/APC strings, Fp/Fe/Fs/nF escapes, 8-bit C1
- terminal_tool.py: strip output before returning to model
- code_execution_tool.py: strip stdout/stderr before returning
- process_registry.py: strip output in poll/read_log/wait
- file_tools.py: remove _strip_ansi band-aid (no longer needed)
Verified: `ls --color=always` output returned as clean text to model,
file written from that output contains zero ESC bytes.
Models occasionally copy ANSI escape sequences from terminal output
or display formatting into file content, breaking shebangs and
injecting binary characters into scripts.
Strip ANSI codes (CSI, OSC, simple escapes) from:
- write_file content
- patch old_string, new_string, and V4A patch content
The check is fast (skips entirely if no ESC byte present).
Reported by Andi Jaeger.
* fix: prevent infinite 400 failure loop on context overflow (#1630)
When a gateway session exceeds the model's context window, Anthropic may
return a generic 400 invalid_request_error with just 'Error' as the
message. This bypassed the phrase-based context-length detection,
causing the agent to treat it as a non-retryable client error. Worse,
the failed user message was still persisted to the transcript, making
the session even larger on each attempt — creating an infinite loop.
Three-layer fix:
1. run_agent.py — Fallback heuristic: when a 400 error has a very short
generic message AND the session is large (>40% of context or >80
messages), treat it as a probable context overflow and trigger
compression instead of aborting.
2. run_agent.py + gateway/run.py — Don't persist failed messages:
when the agent returns failed=True before generating any response,
skip writing the user's message to the transcript/DB. This prevents
the session from growing on each failure.
3. gateway/run.py — Smarter error messages: detect context-overflow
failures and suggest /compact or /reset specifically, instead of a
generic 'try again' that will fail identically.
* fix(skills): detect prompt injection patterns and block cache file reads
Adds two security layers to prevent prompt injection via skills hub
cache files (#1558):
1. read_file: blocks direct reads of ~/.hermes/skills/.hub/ directory
(index-cache, catalog files). The 3.5MB clawhub_catalog_v1.json
was the original injection vector — untrusted skill descriptions
in the catalog contained adversarial text that the model executed.
2. skill_view: warns when skills are loaded from outside the trusted
~/.hermes/skills/ directory, and detects common injection patterns
in skill content ("ignore previous instructions", "<system>", etc.).
Cherry-picked from PR #1562 by ygd58.
---------
Co-authored-by: buray <ygd58@users.noreply.github.com>
Keep Docker sandboxes isolated by default. Add an explicit terminal.docker_mount_cwd_to_workspace opt-in, thread it through terminal/file environment creation, and document the security tradeoff and config.yaml workflow clearly.
- Add 'emoji' field to ToolEntry and 'get_emoji()' to ToolRegistry
- Add emoji= to all 50+ registry.register() calls across tool files
- Add get_tool_emoji() helper in agent/display.py with 3-tier resolution:
skin override → registry default → hardcoded fallback
- Replace hardcoded emoji maps in run_agent.py, delegate_tool.py, and
gateway/run.py with centralized get_tool_emoji() calls
- Add 'tool_emojis' field to SkinConfig so skins can override per-tool
emojis (e.g. ares skin could use swords instead of wrenches)
- Add 11 tests (5 registry emoji, 6 display/skin integration)
- Update AGENTS.md skin docs table
Based on the approach from PR #1061 by ForgingAlex (emoji centralization
in registry). This salvage fixes several issues from the original:
- Does NOT split the cronjob tool (which would crash on missing schemas)
- Does NOT change image_generate toolset/requires_env/is_async
- Does NOT delete existing tests
- Completes the centralization (gateway/run.py was missed)
- Hooks into the skin system for full customizability
- treat git diff --cached --quiet rc=1 as an expected checkpoint state
instead of logging it as an error
- downgrade expected write PermissionError/EROFS/EACCES failures out of
error logging while keeping unexpected exceptions at error level
- add regression tests for both logging behaviors
Stray print() in write_file_tool exception handler leaked debug output
to stdout. Replaced with logger.error() which is already set up in
the file.
Authored by memosr.
Co-authored-by: memosr <memosr@users.noreply.github.com>
Follow-up to PR #705 (merged from 0xbyt4). Addresses several issues:
1. CONSECUTIVE-ONLY TRACKING: Redesigned the read/search tracker to only
warn/block on truly consecutive identical calls. Any other tool call
in between (write, patch, terminal, etc.) resets the counter via
notify_other_tool_call(), called from handle_function_call() in
model_tools.py. This prevents false blocks in read→edit→verify flows.
2. THRESHOLD ADJUSTMENT: Warn on 3rd consecutive (was 2nd), block on
4th+ consecutive (was 3rd+). Gives the model more room before
intervening.
3. TUPLE UNPACKING BUG: Fixed get_read_files_summary() which crashed on
search keys (5-tuple) when trying to unpack as 3-tuple. Now uses a
separate read_history set that only tracks file reads.
4. WEB_EXTRACT DOCSTRING: Reverted incorrect removal of 'title' from
web_extract return docs in code_execution_tool.py — the field IS
returned by web_tools.py.
5. TESTS: Rewrote test_read_loop_detection.py (35 tests) to cover
consecutive-only behavior, notify_other_tool_call, interleaved
read/search, and summary-unaffected-by-searches.
file_tools.py creates its own Docker sandbox when read_file/search_files
runs before any terminal command. The container_config was missing
docker_volumes, so the sandbox had no user volume mounts — breaking
access to heartbeat state, cron output, and all other mounted data.
Matches the existing pattern in terminal_tool.py:872.
Missed in original PR #158 (feat: add docker_volumes config).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Combine read/search loop detection with main's redact_sensitive_text
and truncation hint features. Add tracker reset to TestSearchHints
to prevent cross-test state leakage.
Terminal output was already redacted via redact_sensitive_text() but
read_file and search_files returned raw content. Now both tools
redact secrets before returning results to the LLM.
Based on PR #372 by @teyrebaz33 (closes#363) — applied manually
due to branch conflicts with the current codebase.
Add contextual [Hint: ...] suffixes to tool results where they save
real iterations:
- patch (no match): suggests read_file/search_files to verify content
before retrying — addresses the common pattern where the agent retries
with stale old_string instead of re-reading the file.
- search_files (truncated): provides explicit next offset and suggests
narrowing the search — clearer than relying on total_count inference.
Other hints proposed in #722 (terminal, web_search, web_extract,
browser_snapshot, search zero-results, search content-matches) were
evaluated and found to be low-value: either already covered by existing
mechanisms (read_file pagination, similar-files, schema descriptions)
or guidance the agent already follows from its own reasoning.
5 new tests covering hint presence/absence for both tools.
- Block file reads after 3+ re-reads of same region (no content returned)
- Track search_files calls and block repeated identical searches
- Filter completed/cancelled todos from post-compression injection
to prevent agent from re-doing finished work
- Add 10 new tests covering all three fixes
When context compression summarizes conversation history, the agent
loses track of which files it already read and re-reads them in a loop.
Users report the agent reading the same files endlessly without writing.
Root cause: context compression is lossy — file contents and read history
are lost in the summary. After compression, the model thinks it hasn't
examined the files yet and reads them again.
Fix (two-part):
1. Track file reads per task in file_tools.py. When the same file region
is read again, include a _warning in the response telling the model
to stop re-reading and use existing information.
2. After context compression, inject a structured message listing all
files already read in the session with explicit "do NOT re-read"
instruction, preserving read history across compression boundaries.
Adds 16 tests covering warning detection, task isolation, summary
accuracy, tracker cleanup, and compression history injection.
Add Daytona to image selection, container_config guards, environment
factory, requirements check, and diagnostics in terminal_tool.py and
file_tools.py. Also add to sandboxed-backend approval bypass.
Signed-off-by: rovle <lovre.pesut@gmail.com>
- Implemented functionality to load ephemeral prefill messages from a JSON file, enhancing few-shot priming capabilities for the agent.
- Introduced a mechanism to load an ephemeral system prompt from environment variables or configuration files, ensuring dynamic prompt adjustments at API-call time.
- Updated the CLI and agent initialization to utilize the new prefill messages and system prompt, improving the overall interaction experience.
- Enhanced configuration options with new environment variables for prefill messages and system prompts, allowing for greater customization without persistence.
- Updated the authorization logic to include a per-platform allow-all flag for improved flexibility.
- Revised the order of checks to prioritize platform-specific allow-all settings, followed by environment variable allowlists and DM pairing approvals.
- Added global allow-all configuration for broader access control.
- Improved handling of allowlists by stripping whitespace and ensuring valid entries are processed.
- Introduced logging functionality in cli.py, run_agent.py, scheduler.py, and various tool modules to replace print statements with structured logging.
- Enhanced error handling and informational messages to improve debugging and monitoring capabilities.
- Ensured consistent logging practices across the codebase, facilitating better traceability and maintenance.
- Updated the tool name from "search" to "search_files" across multiple files to better reflect its functionality.
- Adjusted related documentation and descriptions to ensure clarity in usage and expected behavior.
- Enhanced the toolset definitions and mappings to incorporate the new naming convention, improving overall consistency in the codebase.
- Improved the caching mechanism for ShellFileOperations to ensure stale entries are invalidated when environments are cleaned up.
- Enhanced thread safety by refining the use of locks during environment creation and cleanup processes.
- Streamlined the cleanup of inactive environments to prevent blocking other tool calls, ensuring efficient resource management.
- Added error handling and messaging improvements for better user feedback during environment cleanup.
- Introduced a new script, `kill_modal.sh`, to facilitate stopping running Modal apps, including the ability to stop all apps or specific swe-rex sandboxes.
- Enhanced user experience with clear usage instructions and feedback during the stopping process.
- Improved error handling to ensure smooth execution even if some apps fail to stop.
- Introduced new environments: Terminal Test Environment and SWE Environment, each with default configurations for testing and software engineering tasks.
- Added TerminalBench 2.0 evaluation environment with comprehensive setup for agentic LLMs, including task execution and verification.
- Enhanced ToolContext with methods for uploading and downloading files, ensuring binary-safe operations.
- Updated documentation across environments to reflect new features and usage instructions.
- Refactored existing environment configurations for consistency and clarity.
- Updated `.gitignore` to exclude `testlogs` directory.
- Refactored `handle_web_function_call` in `model_tools.py` to support running async functions in existing event loops, improving compatibility with Atropos.
- Introduced a thread pool executor in `agent_loop.py` for running synchronous tool calls that internally use `asyncio.run()`, preventing deadlocks.
- Added `ToolError` class to track tool execution errors, enhancing error reporting during agent loops.
- Updated `wandb_log` method in `hermes_base_env.py` to log tool error statistics for better monitoring.
- Implemented patches in `patches.py` to ensure async-safe operation of tools within Atropos's event loop.
- Enhanced `ToolContext` and `terminal_tool.py` to utilize the new async handling, improving overall tool execution reliability.
- Introduced file manipulation capabilities in `model_tools.py`, including functions for reading, writing, patching, and searching files.
- Added a new `file` toolset in `toolsets.py` and updated distributions to include file tools.
- Enhanced `setup-hermes.sh` and `install.sh` scripts to check for and optionally install `ripgrep` for faster file searching.
- Implemented a new `file_operations.py` module to encapsulate file operations using shell commands.
- Updated `doctor.py` and `install.ps1` to check for `ripgrep` and provide installation guidance if not found.
- Added fuzzy matching and patch parsing capabilities to improve file manipulation accuracy and flexibility.
