dsh-py-codeact
A CodeAct agent loop for DeepSeek Harness using a persistent IPython session with bridged tools.
Install
$ dsh plugin --profile <name> add dsh-py-codeactPlugin Overview & Capabilities
AI-assisted organization based on the public repository snapshot. The content must be grounded in source evidence and does not replace compatibility or security verification.
Key Capabilities
- Persistent IPython REPL maintaining state, variables, and history across multiple execution cells
- Virtual `__dsh__.tools` import namespace with dynamically generated TypedDict signatures and async callables
- Direct integration with Model Context Protocol (MCP) servers under an unwrapped `mcp` namespace
- Exclusive CodeAct execution mode that limits the model's direct action space to Python while guarding tool dispatches
- Emits native `SUBTOOL` session events (`tool/code-dispatch-start` and `tool/code-dispatch`) for UI trajectory visibility
- Custom browser UI card rendering the Python code call with syntax highlighting via `@deepseek-ai/dsh-client-ui-tool`
- Built-in execution management supporting in-band cell interruption and SIGKILL recovery on runaway CPU loops
- Sandbox integration honoring `sandboxPolicy` and `workspace-write` containment constraints
Useful For
- Running complex multi-step data analysis with libraries like pandas where intermediate data frames remain in memory
- Orchestrating multiple DeepSeek Harness tools and MCP actions within idiomatic Python scripts and loops
- Reducing turn counts and token overhead by composing tool invocations programmatically inside code cells
- Providing rich REPL introspection features (magics, `%whos`, `%timeit`, `?` help) to agent workflows
Who It Fits
- Developers building advanced DeepSeek Harness agent presets with programmatic tool calling
- Data scientists and researchers running repetitive analytical workflows via LLMs
- Users orchestrating high-density MCP tool catalogues within isolated Python sessions
Documented Limitations
- Direct native writes to standard file descriptors (e.g. subprocesses writing directly to fd 1) bypass Python stdout capture
- Process containment provides sandboxed separation but is not a hard security boundary without strict host confinement
- Changing session working directories or sandbox policies retires the active kernel, clearing state and open handles
- Initial execution incurs a cold start while uv resolves the PEP 723 Python environment
- Pure CPU-bound blocking loops that never await require a SIGKILL escalation that restarts the interpreter and drops in-memory state
DSH Compatibility
Version-specific runtime evidence collected by DSH Plugin. A missing result means we have not tested that combination yet.
Security Signals
Objective signals discovered from package metadata and source inspection. These are not a guarantee that a plugin is safe.
package.json declares 0 runtime, 14 development, 10 peer, and 0 optional dependencies.
package.json declares DSH bundle metadata.
GitHub reports the repository license as MIT.
A root package.json was captured and can be inspected by the registry.
Public GitHub source metadata is available for this registry snapshot.
Source & Registry Notes
Public provenance, Registry classification, and the latest source check for this entry, kept separate from runtime verification.
- Source repository
- CNSeniorious000/dsh-py-codeact
- Registry source
- GitHub · dsh-plugin topic
- Registry classification
- Plugin
- Source checked
- 9c934d3 · 2026-10-25
This project is independently indexed from public source information. DSH Plugin is not affiliated with DeepSeek or the plugin author. Always check the author repository before installation.
Repository Activity
- GitHub stars
- 1GitHub stars
- Forks
- 2
- Open issues
- 11
- Last commit
- 2026-10-22
- Last release
- No release detected
Related DSH Plugins
Ranked by overlapping capabilities, use cases, plugin type, categories, and DSH profile.
Second-model AI auto-review for DeepSeek Harness approval requests with fail-closed security.
View plugin →Persistent multi-model agent teams with dynamic DAG orchestration, run monitoring, and token insights for DeepSeek Harness.
View plugin →Native SQLite-backed task management and project board bundle with Web UI and Agent workflow tools for DeepSeek Harness.
View plugin →Agent-native multi-market trading terminal and quantitative research plugin bundle for DeepSeek Harness.
View plugin →