run

A command for running one-shot or resumable tasks through DeepSeek Harness. It supports read-only runs, optional workspace edits, model and effort choices, and continuation of recorded sessions.

In plain words
What is it for?
Use it to launch a DeepSeek task, run it with write access when needed, select an available model or effort level, or resume the latest compatible session.
Why use it?
It gives a consistent way to start a task, choose whether files may be changed, or continue work after an earlier session.

Command

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add commands/cpj-dev/dsh-plugin-cc/run
Clone the repo
git clone --depth 1 https://github.com/cpj-dev/dsh-plugin-cc
Per session 18 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 518 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00018 $0.00518
Opus 5 $0.00009 $0.00259
Sonnet 5 $0.00004 $0.00104
Haiku 4.5 $0.00002 $0.00052

Measured 2d ago against content hash 39f2f0f132d9, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

run scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

plugins/dsh/commands/run.md · 28 lines

What it actually says

Run a task on DeepSeek Harness:

node "${CLAUDE_PLUGIN_ROOT}/scripts/dsh-bridge.mjs" run "$ARGUMENTS"

Flags (explain to the user when relevant):

  • Default: one-shot headless run, read-only sandbox. Crash-isolated, not resumable.
  • --write: allow workspace file edits (workspace-write sandbox).
  • --session: run through the shared broker and record a resumable dsh session id.
  • --resume (or --resume-last): continue the most recent recorded dsh session; new text becomes the follow-up prompt, empty means "continue". The resume is validated against the live broker runtime — if the broker was stopped or restarted since, the command errors and asks for a fresh --session instead of silently starting over (relay that to the user).
  • --model <name> / --effort low|high|max: per-run model selection (ignored on --resume; the broker keeps its startup model). Defaults: deepseek-v4-pro at effort max. low is in the official schema; medium is not. A vision id such as deepseek-v4-flash-vision-exp does not attach Claude chat images — this plugin sends text only.
  • --mode minimal|standard|anchored-standard: agent mode. Default standard (full dsh toolset from request #1, no overlay). minimal locks bash + str_replace_editor for the whole run. anchored-standard shows that pair until this session records a tool call or assistant reply, then restores the full catalog. Ignored on --resume; a live broker keeps its startup mode, and a --session run whose mode differs from the live broker errors with a /dsh:stop --broker hint — relay that to the user. /dsh:setup --mode <m> persists a machine default.
  • --background: queue the run, return a run id, check with /dsh:runs.

Before suggesting --resume, you may probe availability with:

node "${CLAUDE_PLUGIN_ROOT}/scripts/dsh-bridge.mjs" run-resume-candidate

Present the run output verbatim, including the footer line with the run id and session id.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 28 lines · 18 tokens per session scan A 39f2f0f132d9

Subscribe to this mod's changes

run is a command published in the GitHub repository cpj-dev/dsh-plugin-cc (106 stars, last pushed 5d ago), licensed MIT. It adds 18 tokens to every session and 518 once invoked, about $0.0001 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.