dsh-plugin-opencode-omo

@royenheart/dsh-plugin-opencode-omo

A DeepSeek Harness plugin that adds an opencode-omo agent preset (mode) to the web profile. The mode replicates the behavior of opencode + the omo plugin (oh-my-openagent), scoped to this mode only — other presets (standard/…) keep the default dsh loop, sandboxed fs, and no omo hooks.

What the mode provides

• opencode + omo system prompt — the real opencode default.txt persona (tone, style, proactiveness, conventions, code style, task guidance) + omo's Sisyphus orchestrator identity, declared as the complete system prompt: the dsh harness identity and runtime-context snapshot are suppressed for this mode. The loop shim additionally prepends opencode's live environment block (exact model id, working dir, workspace root, git, platform, date, recomputed every step).
• omo role picker in the composer — in dsh's existing conversation.input.left tool-row slot (after the access/plan chips): sisyphus, hephaestus (Deep Agent), prometheus (Plan Builder), atlas (Plan Executor), sisyphus-junior, athena/athena-junior/council-member, metis, momus, oracle, librarian, explore, multimodal-looker. Selecting a role swaps the session's complete system prompt and applies that role's configured model.
• Global "Role Settings" in the dsh settings panel (settings.section): per-role primary model dropdown (follow current / fixed) under a centered "Primary model" label; a dsh-style circle "+" button opens a fallback model list below the role box (repeatable additions, cancel/close adds nothing), persisted in opencode-omo-roles settings. On request failure the loop shim advances through the role's fallback chain before the harness retry policy runs.
• opencode toolchain (complete) — persistent bash, read/write/edit/read_image, apply_patch, glob/grep, todo_write, skill, web_fetch/web_search, lsp, exit_plan_mode (plan), ask_user_question. tool-surface.mjs overwrites the model-visible descriptions/parameters with opencode's tool/*.txt text and shims read/edit/write to opencode's parameter names.
• omo task() surface — task-shim.mjs registers the omo-style task(category/subagent_type/load_skills/run_in_background/task_id) invocation, mapping it onto dsh named subagents + generic delegation.
• omo multi-role subagents — oracle (read-only advisor), librarian (external docs/code search), explore (codebase grep), metis (pre-planning), momus (plan reviewer), multimodal-looker (media), plus generic subagent/subagent_fork + workflow/ralph.
• omo context injection — AGENTS.md/CLAUDE.md walk-up + skills/ + omo's rules-injector (.omo/rules, .cursor/rules, .github/instructions, copilot-instructions.md).
• omo hooks — comment-checker (rejects AI-slop comments on write/edit), hashline (read tagging N#HH|content + hashline_edit stale-ref guard).
• per-mode execution backend — local filesystem (dsh-fs-local) + persistent PTY shell, isolated from other modes' sandboxed fs/shell.
• native-seam loop shim — no dsh-side driver seam. driver.mjs is an ordinary preset plugin using the shipped seams: a dynamic ctx.systemPrompt.section({ complete: true }) recomputes opencode's env block and the selected omo role prompt per assembly; system-prompt/assemble applies opencode's model tool gating; agent/inbox/claimed, agent/pre-step, agent/request, and agent/request-error provide ultrawork detection, maxSteps, role model routing, and fallback retry. Other presets are untouched by construction.

Layout

cordis.patch.yml                 # bundle patch: self host row
install.py                       # idempotent install/uninstall (incl. user preset root)
src/                             # host + client plugin halves (role registry, settings, picker UI)
lib/                             # built host/client bundles (npm run build)
scripts/build.sh                 # typecheck + tsdown build
presets/opencode-omo/
  agent.cordis.yml               # the composition (tools, roles, hooks, LSP)
  preset.yml                     # display metadata
  persona.md                     # opencode default.txt + omo Sisyphus persona
  roles/*.md                     # subagent personas (also main-role prompts)
  roles/prompts/*.md             # primary-role complete prompts (hephaestus/prometheus/atlas/…)
  skills/                        # omo shared skills
  driver.mjs                     # native-seam loop shim (prompt/route/fallback/maxSteps/ultrawork)
  rules.mjs                      # rules-injector
  comment-checker.mjs            # comment-checker hook
  apply-patch.mjs                # apply_patch tool
  hashline.mjs                   # omo hashline read-tagging + hashline_edit

Install

lib/ is generated locally and is not committed. install.py always builds the repository's own toolchain first (npm install when the toolchain is missing, then npm run build) and only reports an error when npm itself is missing:

python3 install.py install --profile web              # install (idempotent)
python3 install.py uninstall --profile web            # remove

install.py symlinks the package into ~/.dsh/profiles/<profile>/node_modules/, edits the profile's package.json (adds/removes the dependency + bundle entry), and publishes the preset through dsh's native user preset root as a real directory under $DSH_HOME/.agent-presets/opencode-omo (entries symlinked into the package, so updates stay live):

Manual alternative — the package is a dsh bundle: it declares dsh.bundle.patch and ships the preset. dsh plugin reconciles dsh.profile.bundles from the installed package automatically:

dsh plugin --profile web add link:/path/to/dsh-plugin-opencode-omo

The preset still needs its user-root publication (the bundle patch cannot create $DSH_HOME/.agent-presets entries):

mkdir -p "$DSH_HOME/.agent-presets/opencode-omo"
for f in /path/to/dsh-plugin-opencode-omo/presets/opencode-omo/*; do
  ln -s "$f" "$DSH_HOME/.agent-presets/opencode-omo/"
done

Restart dsh and select opencode-omo from the mode picker.

Optional LSP server. The preset preconfigures typescript-language-server for the lsp tool. install.py checks PATH and warns when it is missing; the preset then self-disables its lsp-stdio row so the mode still mounts (LSP queries fail gracefully instead of blocking the whole preset). Install it to enable LSP:

npm install -g typescript-language-server typescript

then restart dsh.

web_fetch provider. The preset enables dsh's native web_fetch tool, so the bundle patch also registers the native HTTP fetch provider (@deepseek-ai/dsh-web-fetch-http) into the shared ctx.web; install.py links that package into the profile module tree. web_search keeps using the existing DeepSeek search provider (DEEPSEEK_API_KEY). Other presets keep fetch: false, so their tool surface is unchanged.

Required dsh-side changes

This release depends on one dsh-side patch. Apply it to deepseek-harness for full maxSteps fidelity; the patches are under patches/, split by feature:

• patches/0001-agent-pre-step-assistant-prefill.patch — adds optional assistantPrefill to PreStepDecision; the loop appends it to the request after the derived history and logs it on request/header (request-only, never a session message). The plugin uses it to restore opencode's MAX_STEPS_PROMPT assistant-role semantics. Runtime compatibility: the host plugin scans the installed @deepseek-ai/dsh-agent-loop bundle for the compiled assistantPrefill marker. When the patch is absent, maxSteps automatically degrades to an equivalent synthetic user message (nothing is silently dropped) and the /roles response carries the warning to the browser; the web client shows it once per page load via the native @deepseek-ai/dsh-client-ui-primitives Toast (4 seconds, non-blocking).

cd /path/to/deepseek-harness
git apply /path/to/dsh-plugin-opencode-omo/patches/0001-agent-pre-step-assistant-prefill.patch
npm run build:lib:host
npx vitest run packages/core/agent-loop/tests/interception.spec.ts

• Provider-visible format/toolChoice is still a proposal (see DSH_CHANGE_PROPOSALS.md); omo's regular path does not use it and the standalone structured-output plugin covers the common route.

Everything else runs on unmodified dsh seams: the preset is published through $DSH_HOME/.agent-presets and the composer picker occupies the existing conversation.input.left slot (the client registers through ctx.slots.inject(), so it waits for the declaring parent regardless of out-of-tree bundle apply order).

Bench experiments (equivalence validation)

Runners and reproduction notes live in tests/benches/; reports are written to docs/exps/. The scientific methodology (paired design, McNemar/bootstrap/TOST, A/A noise floor, trace alignment, cache and latency protocols) is documented in docs/exps/2026-08-16-scientific-bench-methodology.md. Design highlights:

• Run an isolated-port dsh (opencode-omo mode, isolated $DSH_HOME) and the machine's installed opencode + oh-my-openagent (isolated XDG_CONFIG_HOME).
• Both use deepseek-official/deepseek-v4-pro (dpsk v4 pro); the API key comes from the DEEPSEEK_API_KEY environment variable. Scripts hardcode no machine paths or secrets.
• Tiered benches: L1 HumanEval, L2 MBPP, L3 SWE-bench-verified-mini (sampled), comparing pass@1, CoT/reasoning exposure, tool-call chains (read/edit/write/bash/test/subagents), and final patches/answers.
• Performance: bench_metrics.mjs normalizes token usage (including cacheRead) and tool timestamps on both sides; eval_perf.mjs / eval_swe_perf.mjs produce cache hit rate, TTFT, step duration, and per-tool timing offline (see report §5).
• MBPP fix: MBPP rows have no entry_point; the scripts infer the function name from the first assert fn(...) or the reference def fn. The old implementation hardcoded Function name: undefined, which was the root cause of MBPP behavioral divergence (see report §0 and §6.4).
• Raw bench data is downloaded to tests/benches/.data/ and is not versioned; fetch-benches.sh reproduces the downloads and setup-homes.sh reproduces both isolated homes.

Latest results (dpsk v4 pro; L1 30×3, corrected L2 5×2, legacy L3 sample)

level	dsh pass@1	opencode+omo pass@1	per-item agreement
HumanEval (30 tasks × 3 repeats)	1.00	1.00	1.00
MBPP (corrected entry_point, 5 tasks × 2 repeats)	1.00	1.00	1.00
SWE-bench-verified-mini sample (sphinx-doc__sphinx-10323)	same patch	same patch	byte-identical git diff

Full report: docs/exps/2026-08-15-opencode-omo-equivalence-bench.md; raw transcripts are under tests/benches/.runs/.

Alignment status (audited against reference/opencode + reference/oh-my-openagent)

• Aligned: opencode default persona (complete system prompt + live env block whose provider/model now follow the same per-step route as the actual request — session live model selection or the role primary/fallback — so prompt and request cannot split; workspace root now derived as the git root); opencode tool families + gpt apply_patch/edit-write tool gating enforced on BOTH the model-visible schema and execution (tools/pre-execute deny mirror); opencode maxSteps + verbatim MAX_STEPS_PROMPT; verbatim opencode plan.txt / plan-mode.txt with dynamic ${planInfo} and the plan→build BUILD_SWITCH reminder; omo role catalog/display names; sisyphus/hephaestus/atlas/sisyphus-junior + specialist subagents; comment-checker/hashline/rules-injector hooks; generated Sisyphus routing sections; extracted omo Sisyphus model-family templates (GPT-5.5/GPT-5.4/claude-opus-4-7/claude-opus-4-8/claude-fable-5/gemini/kimi-k3/kimi-k2-7/kimi-k2-6/glm-5-2, with the dynamic Sisyphus fallback for unknown families) plus hephaestus GPT variants, all 8 atlas variants, and specialist model variants (oracle/metis/momus); omo-default per-role PRIMARY model resolution (provider-scope ordered) and fallback chains that start AFTER the primary; omo role sampling defaults (sisyphus/hephaestus GPT effort medium, atlas temperature 0.1); omo-style retryable-error gating before fallback advance; reasoning-effort selectors in role settings; ultrawork keyword override; /start-work, /remove-ai-slops, /refactor, /stop-continuation, /handoff, /hyperplan, /team-mode commands; composer role picker + global per-role model/fallback settings; omo skills published as user-dsh so the third-party skills-manager can manage them. The omo rules-injector text is now folded into the complete system prompt (driver.mjs + rules.mjs) instead of being dropped by suppressRuntimeContext(); approved plans are persisted at .opencode/plans/<created>-<session>.md; specialist subagent personas now load the extracted reference prompt files (oracle/librarian/explore/metis/momus/multimodal-looker).
• MCP: separate plugin dsh-plugin-mcp-support mounts native @deepseek-ai/dsh-mcp-client servers from its bundle-row config or the persisted mcp-support settings namespace.
• Structured output: separate plugin dsh-plugin-structured-output provides opencode-style /json-schema + StructuredOutput validation on native seams (no dsh-side format field). Its visibility is opt-in per preset via Settings → 结构化输出工具 (Structured output); no mode is enabled by default.
• Partial: extracted family templates keep dynamic sections filled by dsh-native data rather than omo's builder output; structured output is tool-enforced rather than tool_choice: required; hooks are regex/simplified ports; AGENTS.md injection is dsh-native; child subagents inherit the session model because dsh child headers/descriptors do not carry the subagent role id (primary-role sampling defaults ARE applied).
• Requires one dsh-side patch (see patches/, audited 2026-08-15): PreStepDecision.assistantPrefill for opencode's MAX_STEPS_PROMPT. Provider-visible format/toolChoice remains a proposal; omo's regular path does not use it and the standalone structured-output plugin covers the common route.

Remaining gaps

1. dsh-side (patch provided): PreStepDecision.assistantPrefill. Without the patches/ patch the maxSteps prompt is degraded to a synthetic user message; after applying the patch this gap is closed.
2. dsh-side (proposal, medium): GenerateOptions.format / toolChoice. omo's regular path does not use them; the standalone structured-output plugin covers the common route.
3. Child subagent per-role sampling cannot reliably resolve the role id (dsh child headers/descriptors do not carry it); primary-role sampling defaults ARE applied and children inherit the session model.
4. Plan files: dsh itself does not persist them; the plugin writes .opencode/plans/* after exit_plan_mode approval. A first-class plan-file seam remains an optional improvement.
5. team-mode TUI, comment-checker CLI, and hashline diff enhancer remain non-LLM/editing experience differences.