What is dsh-delegate-runtime?

A contract for using a command-line bridge to DeepSeek Harness, a tool that reviews code and carries out tasks. It defines the available commands, options, environment, and failure behavior.

Use it when checking setup, reviewing or critiquing code, running tasks, resuming sessions, importing conversations, or viewing stored run results.

It prevents commands from being assembled incorrectly or failures from being handled inconsistently. The bridge's output is intended to be shown directly to the user.

dsh-delegate-runtime — Skill by cpj-dev

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 skills/cpj-dev/dsh-plugin-cc/dsh-delegate-runtime

Any agent

npx skills add cpj-dev/dsh-plugin-cc --skill dsh-delegate-runtime

Clone the repo

git clone --depth 1 https://github.com/cpj-dev/dsh-plugin-cc

Made for: Claude Code, Codex.

Per session 47 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 881 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.

Model	Per session	Once invoked
Fable 5	$0.00047	$0.00881
Opus 5	$0.00023	$0.00441
Sonnet 5	$0.00009	$0.00176
Haiku 4.5	$0.00005	$0.00088

Measured 2d ago against content hash 5c94d91d25d4, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

dsh-delegate-runtime 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/skills/dsh-delegate-runtime/SKILL.md · 47 lines

How it starts

The opening of the file, as written. The whole thing — 47 lines — stays where its author put it; the contents beside it link to each section on GitHub.

dsh bridge runtime contract

Every DeepSeek Harness interaction goes through one script:

node "${CLAUDE_PLUGIN_ROOT}/scripts/dsh-bridge.mjs" <subcommand> [flags] [text]

stdout is the user-facing result; render it verbatim. A nonzero exit means the run failed and stderr carries the reason.

Subcommands

Subcommand	Purpose	Key flags
`check`	readiness probe	`--json`
`setup`	install the pinned npm CLI (or `--harness <built-checkout>`) and create/verify the multi-turn `cc` profile	`--harness <path>`, `--mode`, `--json`
`review`	read-only code review	`--base <ref>`, `--scope`, `--model`, `--effort`, `--mode`, `--background`
`critique`	structured adversarial critique	same as review
`run`	task execution	`--write`, `--session`, `--resume`, `--fresh`, `--model`, `--effort`, `--mode`, `--background`, `--prompt-file`
`run-resume-candidate`	is a resumable dsh session available?	`--json`
`import`	weak-import this conversation into a resumable session	`--source <jsonl>`
`runs`	list runs / one run's status	`[run-id]`, `--all`
`show`	stored result of a finished run	`[run-id]`
`stop`	kill a run's process tree / the broker	`[run-id]`, `--broker`

Facts that shape correct usage

Sandbox: default runs are read-only; --write grants workspace-write. There is no mid-run approval — permissions are decided before launch.
Fresh runs are one-shot headless dsh sessions and are NOT resumable. Only --session, --resume, and import (broker paths) record resumable session ids.
--model/--effort work on one-shot runs; a --resume keeps the broker's startup model. Plugin defaults when omitted: model deepseek-v4-pro, effort max (one-shot and broker alike).
Agent mode defaults to standard (full toolset from request #1, no overlay). --mode minimal locks bash + str_replace_editor for the whole run; --mode anchored-standard shows that pair until this session records a tool call or assistant reply, then restores the full catalog. DSH_CC_MODE and the persisted /dsh:setup --mode default sit between flag and built-in. A live broker keeps its startup mode — a --session run resolving a different mode errors and names /dsh:stop --broker.
--background returns immediately with a run id; the actual work continues in a detached worker. Poll runs <id>, fetch show <id>.
stop kills processes; a mid-turn broker stop discards all in-memory dsh sessions for the workspace.
Environment: DSH_BINARY overrides the dsh executable; DSH_CC_SESSION_ID/DSH_CC_TRANSCRIPT_PATH are exported by the SessionStart hook — never set them manually.

Read the full file on GitHub · 47 lines

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.

2d ago First seen · 47 lines · 47 tokens per session scan A 5c94d91d25d4

Subscribe to this mod's changes

dsh-delegate-runtime is a skill published in the GitHub repository cpj-dev/dsh-plugin-cc (106 stars, last pushed 5d ago), licensed MIT. It adds 47 tokens to every session and 881 once invoked, about $0.0002 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.

Building agents ai-agent claude-code claude-code-plugin dsh dsh-plugin dsh-plugins

odai

Skill "odai" from orziz/odai, covering 精神内核, 当前判断, 按表现分配支撑, 共同行动边界 and 完成.

orziz/odai · 57 tokens

cordis-plugin-development

Create, modify, debug, or extend dynamic Cordis Plugins, including Host Services and Events, Client Slot and theme UI, Package-private Client-to-Host calls, dynamic Tools, version updates, approval failures, and runtime diagnostics. Use this Skill to route a user request to the correct platform and Inspect Provider…

deepseek-ai/deepseek-harness · 80 tokens

editing-cordis-compositions

Use when creating, changing, or validating a Cordis composition for this harness — writing or editing an agent preset, adding or removing a plugin row, deciding whether something belongs to the host composition or to one session, checking whether a preset you authored actually mounts, or diagnosing a row that mounted…

deepseek-ai/deepseek-harness · 69 tokens

dsh-web-community-plugin-developer

Develop a DSH community plugin and register it in the dsh-web Community Plugins index — author the plugin in the contributor's own repository following the official cordis bundle standard, add its entry to packages/dsh-community-plugins/community.json, regenerate the index with scripts/community-index, rebuild and…

zhu1090093659/dsh-web · 123 tokens

submit-dsh-plugin

验证并提交 DeepSeek Harness 插件到 imsai-sh/awesome-deepseek-harness-plugins 社区目录。适用于插件作者要求收录、发布或提交自己的插件，创建目录 JSON，修复目录提交 PR，或者发起合规 PR。检查公开仓库、dsh-plugin topic、dsh.bundle.patch、作者测试证据、双语元数据和单文件差异，并在获得授权后执行 fork、push 和创建 PR。.

imsai-sh/awesome-deepseek-harness-plugins · 103 tokens

ribao

日报、周报、工作总结、状态更新：基于可核证事实形成完整汇报。用户要求汇报一段时期的工作、沿用既有汇报模板、整理进展与风险，或 odai 判断专业汇报工艺会改善结果时使用。.

orziz/odai · 64 tokens