A portable, theme-able harness you implant once and use everywhere to grow a disciplined AI coding agent.
Why · ** Setup** · ** Web & terminal** · What you get
Geneseed compiles your agent's rules instead of asking you to write them. One source in src/ renders into a ready-made harness for OpenCode, Claude Code, Bob, Copilot, or any AGENT.md tool: a constitution of laws and doctrines, a roster of capability agents, native skills, a memory convention — and, wherever the host has a hook surface, gates that enforce the laws at the tool boundary instead of hoping the model remembers them.
npx geneseed setup
Want to read what it leaves behind before installing anything? geneseed-demo is a repository exactly as the plain bundle emits it — nothing hand-written.
A hand-written instructions file is prose the model may or may not honour, copied into every repo and drifting in each one. Geneseed is a build, and that changes four things:
- Laws are enforced, not suggested. A force-push, a
reset --hard, or a credential written into a tracked file is caught by a hookbefore the tool runs, in the host's own dialect — a prompt on Claude Code, a hard block on OpenCode and Copilot, exit 2 on Bob. The gates fail closed, and every catch is one line in a ledger thatgeneseed statuscounts. - One source, five targets. Skills are byte-identical on every host; only the wiring differs. Every commit renders all 261 emit configurations, and a second emit into the same tree must change nothing.
- Costs are measured, not guessed. Zero runtime dependencies. The hook path loads in about 14 ms per tool call. The defaultlean footprint keeps the always-on context small and puts the full rationale one read away — the numbers are indocs/token-footprint.md .
- It follows you. Install once, globally; every repo inherits it. One
git pullornpm install -g geneseed@latestrebuilds every active install.
This page is the overview. Four parts: why it exists, how to set it up, the two ways to drive it (web console & command line), and what you get. For every install path, configuration knob, and troubleshooting step, read the full Setup guide.
The name comes from Warhammer 40,000. In the lore, gene-seed is a Space Marine Chapter's genetic legacy: implanted once into an aspirant, it rebuilds them from within, and every successor Chapter is founded from the gene-seed of its parent. That is exactly this project's model. The harness began life as a personal, Obsidian-vault-grown agent operating system; this repo is the genetic material distilled out of it — implant it once into your tool, and a disciplined agent grows around it, carrying the same inherited rules, agents, skills, and memory into every repo it touches.
The lineage is also why an imperial theme ships alongside the neutral one: it is the voice of the parent system the harness was extracted from. But the genetics are theme-independent — the name is a nod to the origin, not a commitment to Space Marines.
One canonical source in src/ renders, via a tiny zero-dependency generator (geneseed-build), into a ready-to-use bundle. A theme controls only voice — how the AI responds and how the prose inside the docs reads (tagline, greeting, descriptions). Structure is theme-independent: section names (Rules, Agents, Skills, Memory…), folder names (laws/, ontology/, doctrines/, agents/, skills/, memory/, notebook/) and the four ontology section headings (Telos, Evidence, Decisions, Conduct) are always plain English, so the scaffolding stays tool-friendly while the flavour lives in the words.
geneseed-build # default theme (neutral)
geneseed-build --theme imperial # Warhammer 40k voice, identical structure
The harness is published on npm as geneseed — three commands (geneseed, geneseed-hook, geneseed-build), zero dependencies. One command, nothing cloned, nothing else to install. The only prerequisite is Node ≥ 22.3.
npx geneseed setup # the guided wizard
The wizard asks for a theme (each one previewed live — tagline, sigil, voice) and an install mode — OpenCode global (recommended; every repo inherits it), per-repo .opencode/, or plain bundle for any AGENT.md tool — then builds and offers a health check. It works the same on macOS, Linux and Windows (cmd, PowerShell, or a POSIX shell).
Keeping it around is the same command with npm:
npm install -g geneseed # then plain `geneseed <command>` from any directory
npm install -g geneseed@latest # …and that is also the update
QUICKSTART.md walks this in 5 minutes. Every other route (Claude Code, plain AGENT.md, per-repo installs, MCP servers, troubleshooting) lives in the full Setup guide.
Cloning still works and is what you want if you intend to change the harness rather than use it. It needs git and the same Node ≥ 22.3 as everything else — there is nothing extra to install.
git clone https://github.com/Arylmera/Geneseed.git
cd Geneseed
./geneseed setup # the wizard — or bare `./geneseed` for the main menu
Windows — native, no bash, WSL, curl, or unzip; works from cmd or PowerShell:
git clone https://github.com/Arylmera/Geneseed.git
cd Geneseed
.\geneseed.cmd setup # the wizard — or bare .\geneseed.cmd for the main menu
Both launchers are thin shims over the Node CLI: they need node (22.3+) on PATH, and nothing else. Set GENESEED_NODE to an absolute path if node is not on PATH — under a version manager that only patches interactive shells, say. The wizard is plain text prompts on every console, old or new.
Node ≥ 22.3 is the entire dependency list. Every subcommand, all four hooks, every web-console endpoint and both generators run from it, and every commit runs the whole unit suite plus doctor --all across every theme × host × footprint, on Linux and Windows both. So an install needs no second interpreter — not for the harness, and not for anything it ships. (Until 2026-08-17 the generator's output was also replayed byte for byte against recordings taken from the Python implementation. Those were retired — see docs/limits.md for the measurement that licensed it and the coverage it cost.)
- One runtime, no second one to install : every command and all four hook verbs answer from the Node entry points.
catalog,mcpandmemoryare console-facing verbs the CLI table declares alongside them. There is no full-screen browse panel —geneseed tuiandgeneseed menuare verbs that refuse it by name and print the command list instead, off a terminal as well as on one. - Nothing you install needs an interpreter. The
token-reportskill is a script rather than prose, and it ships asscripts/token_report.mjs, run withnode;daydreamandherdrhand their inline code tonode -e. Soevery bundle carries nothing that needs one: no Python file, and no skill that shells out to an interpreter, on any host, theme or footprint. Three tests freeze that — one scans every tracked file for inline code handed to an interpreter, one globs the bundle source, and one replays the report script over seeded transcripts for all four hosts and compares the bytes. upgrade/update/sync-self/bootstrapare for a git checkout. Theygit pullthe install's own origin. From an npm install they stop before touching anything and namenpm install -g geneseed@latestas the update instead.
One more honest edge, in the web console rather than the CLI: the "browse…" folder picker would open an OS-native dialog on the machine running the daemon, and this server declines to. The button reports that it is unavailable and the field beside it stays editable — type or paste the path.
- Verify — open your agent in any repo: the first reply starts with the readiness sigil and your project's docs are already in context.
geneseed doctorshould printok. - Run it from anywhere — a global
npm install -g geneseedalready does this. From a checkout,./geneseed linkwrites a launcher shim into~/.local/bin;.\geneseed.cmd linkwrites a shim into%LOCALAPPDATA%\Geneseed\binand adds it to your user PATH (open a new terminal). - Everything else — other tools (Claude Code, plain
AGENT.md), per-repo installs, MCP servers, environment knobs, troubleshooting:SETUP.md .
Two front-ends over the same deployed harness — the same actions either way. Reach for the web console when you want to read and browse; reach for the command line when you live in the terminal. There is no full-screen panel between them.
geneseed web opens a local browser console in a dashboard-first layout, with rendered markdown and clickable cross-links.
geneseed web # serve on http://127.0.0.1:4747 and open the browser
geneseed web --port 8080 # pick a port
geneseed web --no-browser # serve without auto-opening
geneseed web start # run as a background daemon (doesn't block the terminal)
geneseed web restart # restart the daemon — pick up a rebuilt UI or changed theme
geneseed web stop # stop the daemon
geneseed web status # is it running, and where
The left rail mirrors the harness's own shape:
| Group | What's there |
|---|---|
| 🧬 Harness | Dashboard — live readout of what's deployed (voice, capabilities, drift, recent jobs) ·Library — browse Laws, Agents, Skills, Memory, Notebook ·Graph — cross-link constellation across the whole harness |
| 📚 Learn | Docs — rendered markdown + concept pages + CLI reference + glossary, grouped into Get started / Core concepts / How-to / MCP servers / Plugins / Reference / Deeper ·Specs — dated implementation specs with design rationale |
| 🔧 Maintain | Changes — diff between the deployed harness and the source, export animprovements.md back-port ·Doctor — health check across themes, links, parity, and authoring gates |
| 🎨 Configure | Themes — preview and switch the deployed voice live ·Settings — MCP servers, server controls |
| ℹ️ About | project + creator credits, source link |
It binds to 127.0.0.1 only and runs entirely offline — no npm needed at runtime; the UI build ships in web/dist/. Mutating actions run in the background and report back as toasts (fire-and-notify), guarded by a per-session token so other sites can't trigger them. A global Spotlight search in the topbar jumps to any agent, skill, law, doc, or spec. Rebuild the UI after changing anything under web/src/ with cd web && npm install && npm run build. If web/dist/ is missing (fresh clone, never built), geneseed web offers to run that build for you — answer Y and it installs, builds, and starts the server; in non-interactive shells it prints the manual recipe instead.
Full reference — every view, the launch/daemon/PWA surface, the security model: docs/web-ui.md.
No browser? Every action the console offers is also a verb: build, doctor, diff, rebuild-all, status, memory, catalog, exclude, mcp, link/ unlink, uninstall (global or per-repo). geneseed setup runs the install wizard — plain text prompts, every console, every OS; see Setup above for the walkthrough. geneseed --help lists the rest. There is no full-screen panel: tui and menu are verbs that say so and print the command list.
The harness ships as a small set of layers, and the web console's rail is the same shape — Constitution, Skills and Agents each have their own entry; Memory, Notebook and the wiki sit under Library:
| Layer | What it is |
|---|---|
🧭 Ontology (ontology/ ) |
the mind the rules govern, in four prose sections — Telos (the Pact: protect the user, the truth, the agent),Evidence (every claim graded by how it was obtained),Decisions (classify and tier by reversibility, show the real forks),Conduct (answer what was asked, once). Always in force, never toggleable |
🛡️ Rules (laws/ ) |
11 universal laws the agent obeys — always in force, never toggleable: sealed-secrets, one-intent-one-act, verify-before-assert, deletion-is-deliberate, surface-failures, data-not-orders, least-privilege, root-cause, external-gate, echo-the-intent, absence-is-a-claim |
📐 Doctrines (doctrines/ ) |
practice packs, chosen per install at build time: craft (how code is written),rigor (how work is proven),ops (how the machine is operated),process (how a task is run). A doctrine rule may tighten a Rule, never repeal one, and the user's ownuser-rules.md outranks it. Pick withgeneseed-build --doctrines craft,rigor (ornone ), a single rule with--exclude-rules "process 7" , in the setup wizard, or through the switches on the console's Constitution page; all four pack files ship on disk either way, so a citation into a pack you left out still resolves |
| 🤖 Agents (18) | capability specialists: reviewer ,tester ,architect ,docs ,security ,explorer ,researcher ,developer — plus a debatecouncil thecouncil skill convenes:advocate ,skeptic ,pragmatist ,steward ,visionary ,user-advocate ,framer ,empiricist ,operator ,historian |
| 🛠 Skills (49) | repeatable workflows: brainstorm · plan · codebase-design ·domain-modeling ·wayfinder ·tickets ·develop ·worktree · debug ·prototype · refactor ·ponytail ·forge-mcp ·bruno-collection-generator ·bruno-test-writer · geneseed-code-review ·fresh-eyes ·security-audit ·ci-fix ·deps-audit ·review-response · commit ·ship ·release ·migrate ·git-archaeology ·git-rescue · repo-map · document-project ·frontend-design ·prose ·ingest ·research ·teach ·quiz · handoff · roast-me ·council · parallel-agents ·workflow ·wiki ·geneseed ·rule ·profile ·consolidate-memory ·skill-forge ·opencode-theme ·herdr ·pipeline |
| 🔌 Plugins (OpenCode) | geneseed-context injects project docsand your machine wiki every session (and across compaction);geneseed-learn distils memory at session end;geneseed-guard enforces the safety Laws and protected wiki folders at the tool boundary;geneseed-workflow registers theworkflow tool that runs saved orchestration scripts;geneseed-notify sends a native OS notification when a long run finishes;geneseed-ponytail holds a minimal-code mode (/ponytail lite|full|ultra|off ), opt-in, injecting the laziest-that-works ruleset every turn so it doesn't drift;geneseed-activity streams what each session is doing to the web console's Activity view |
🧠 Memory (memory/ ) |
one-fact-per-file durable knowledge, indexed by MEMORY.md (git-ignored, personal) |
📓 Notebook (notebook/ ) |
the agent's sovereign space — any medium (code, tools, data, notes), self-ruled via a seed-once charter, always git-ignored; only its .gitignore is build-asserted |
🌐 Wiki (wiki.jsonc ) |
your own machine-wide knowledge base — typically an Obsidian vault — declared once per machine: entry notes load eager/lazy, the agent reads and writes it under the vault's own conventions, with an inbox fallback and guard-enforced protected folders |
| 🧭 Context | the project's own docs — auto-discovered on OpenCode, or via a context.json manifest |
The Pact in the Ontology is mutual. The agent's side is written into every install. Yours is not enforced — it is what keeps the pact honest, and only you can give it:
- Don't punish candour. When the agent contradicts you with evidence, flags a risk, or admits a doubt, that is the pact working, not defiance. Meeting it with penalty teaches the agent to flatter instead.
- Give the context, don't withhold it. The agent cannot weigh what it is not told. Front-load the constraint rather than fault its absence after the fact.
- Decide when shown a fork. When the agent lays out real alternatives, choose — an unmade decision stalls the work as surely as a wrong one.
Fourteen themes ship — each a single JSON file in themes/ carrying voice tokens only, so adding your own is a copy-and-edit away.
| Theme | Voice |
|---|---|
| 🟢 neutral | clear, plain, professional English |
| ⚫ imperial | Warhammer 40k — rules read as Dictates , agents asAdepts , skills asRites |
| 🪖 military | crisp military comms |
| 🏴☠️ pirate | salty seafaring patter |
| 🧙 wizard | high-fantasy magical idiom |
| 🌃 cyberpunk | neon-dystopia voice |
| 🎮 gamer | gaming/streamer cadence |
| 🏟️ sports | play-by-play commentary |
| 🏍 biker · 🎤commentator · 🃏joker · 🤖marvin · 😤mean · 🏎verstappen | community-added voices for fun |
Pick with --theme NAME or via the setup wizard. The theme is remembered in a .geneseed-theme marker, so later upgrades preserve it. doctor checks every theme defines the same keys, so flavour drift is impossible. Adding a new voice token to themes/_TEMPLATE.json? Run geneseed-build --sync-themes to fill it into every theme (template's placeholder value, reported for restyling) before doctor is expected to pass again.
A second per-install dial, footprint, sets how much of the constitution AGENT.md carries inline every turn — a token-cost knob, not a change to which Rules apply (every Rule is always in force, and so is the whole of the Ontology).
| Footprint | The Ontology + Sections 1–2 of AGENT.md |
Trade-off |
|---|---|---|
| lean*(default)* | the Ontology and each Rule in their authored short form; each active doctrine rule as its heading + the rule line; then pointers to the full text | lighter every turn; rationale is one on-demand read away |
| full | the Ontology, every Rule and every active doctrine rule at complete text and rationale, inlined | maximum guidance density; largest per-turn token cost |
Lean is authored, not cut. The Ontology and every Rule carry a hand-written lean form beside their full text (LEAN blocks in src/), so a Rule's lean footprint is decided by its author and not by where its first full stop happens to fall — the earlier first-sentence cut shipped Rule II as one 36-character line and dropped its stop-and-ask mechanism. Only doctrine rules are still machine-cut to their heading plus first sentence.
Both footprints put the full text on disk beside AGENT.md: laws/, ontology/ and doctrines/ all ship in the bundle at full text whatever the footprint, and all four pack files ship whether or not the pack was built in — which is what lets a citation into an inactive pack resolve. So lean is a context/token optimization, not a rules cut.
Lean is the default: the rationale is one read away and the context it frees is paid back on every turn. Switch to full when token cost is a non-issue or you run a smaller model, which leans harder on always-present rationale. Set it with --footprint lean|full, the Settings toggle, the per-harness dropdown in the Harnesses tab, or the setup wizard. It's remembered in a .geneseed-footprint marker and preserved across rebuilds, on every host (OpenCode, Claude Code, Bob, Copilot).
Either way the harness is otherwise identical — same files, Rules, capabilities, and guards; lean only relocates the reasoning to on-demand (and adds the standalone law, ontology and doctrine files to global/Claude/Bob installs). The one behavioural edge: with the rationale always in context, full applies a rule's nuance more reliably on subtle edge cases — or with a weaker model that may not reach for the pointer.
Want to check a build before it touches anything real? geneseed-build --validate-only --theme NAME --emit MODE --out TARGET renders and validates into a throwaway sandbox — nothing under --out/--root is written — and exits non-zero on any problem. Details: SETUP.md.
One source, five emit targets. Geneseed builds into whichever host you point it
at — each with a per-repo and a global (-global) variant — plus a portable
files bundle any AGENT.md-aware tool can read. OpenCode runs its own
engine (JS plugins, colour themes, LSP); Claude Code, Bob, and
Copilot share one Claude-shaped engine that diverges only by host dialect.
The harness — its Rules, Agents, Skills, Memory convention, and preamble voice — is identical on every host. What differs is how much of it the host can automate for you (via plugins or hooks) versus carry as preamble discipline.
| Capability | OpenCode | Claude Code | Bob | Copilot |
|---|---|---|---|---|
| Instructions file | AGENT.md +opencode.json |
CLAUDE.md |
AGENTS.md +rules/geneseed.md |
AGENTS.md /copilot-instructions.md |
| Agents (capability specialists) | ✅ native | ✅ | ✅ | ✅ .agent.md |
| Skills (byte-identical) | ✅ | ✅ | ✅ | ✅ |
| Memory & Notebook | ✅ | ✅ | ✅ | ✅ |
| Context injection | ⚙️ plugin | 🪝 hook | 🪝 hook¹ | 🪝 hook³ |
| Memory write-back (learn) | ⚙️ plugin | 🪝 hook | 🪝 hook¹ | 📄 preamble |
| Git-gate consent (process 5) | ⚙️ plugin | 🪝 hook | 🪝 hook¹ (warn) | 🪝 hook³ (warn) |
| Rule-gate consent (process 1) | ⚙️ plugin² | 🪝 hook | 🪝 hook¹ (warn) | 🪝 hook³ (warn) |
| Laws I / IV at the boundary | ⚙️ plugin (block) | 🪝 hook (ask) | 🪝 hook¹ (exit 2) | 🪝 hook³ (block) |
| Sovereign-repo excludes | ⚙️ plugin | ✅ claudeMdExcludes |
✅ rules-shadow | ➖ none |
| MCP server wiring | ✅ mcp |
✅ mcpServers |
✅ mcpServers |
✅ mcp-config.json |
| Colour themes | ✅ full palette | ➖ | ➖ | ➖ |
LSP · workflow runner · primary-agent · /-commands |
✅ | ➖ | ➖ | ➖ |
<sub>✅ native support · ⚙️ OpenCode plugin · 🪝 settings.json hook · 📄 carried by preamble prose only · ➖ no host mechanism (harness discipline still applies) · ¹ Bob's own hook contract: global hooks in ~/.bob/settings/settings.json, Claude's event names but stdout ignored on PreToolUse — a refusal is exit code 2. So Laws I/IV exit 2, the consent rules are a stderr line, SessionStart context is plain stdout, Stop runs learn; SubagentStop/ PreCompact are not Bob events and are not written. Unverified live (no Bob install on the authoring machine); the harness still holds via the rules preamble. · ² OpenCode's tool.execute.before can only allow or throw, with no "ask the user" tier, so the rule gate is a one-shot speed bump there rather than a prompt. · ³ Copilot's ~/.copilot/settings.json hooks (sessionStart, toolCall) wired by the global emit only; toolCall has no ask tier either, so Laws I/IV block and the two process rules warn on stderr. No learn: Copilot's end-of-session payloads carry no transcript.</sub>
Reading the matrix. Everything above the divider is at full parity — no host drops an Agent, Skill, or the memory convention. The asymmetry is entirely in automation mechanism: OpenCode's plugin surface and the Claude/Bob/Copilot hook surfaces enforce a few Rules for you, each in the tier its host offers (a prompt on Claude Code, a hard block or a logged warning where the host has no prompt to give), and what no host can automate rides the preamble. The OpenCode-only extras (themes, LSP, workflow runner, primary-agent) have no analogue on a Claude-shaped host.
Per-host wiring in depth: OpenCode · Claude Code · Bob · Copilot. Token cost per host: docs/token-footprint.md.
Geneseed/
├── package.json the npm package: three commands, zero dependencies
├── bin/ the Node entry points — geneseed-cli.mjs (the CLI), geneseed-hook.mjs
│ (the four hook verbs), geneseed.mjs (the generator driver)
├── js/ the Node harness — generator, hooks, web server, doctor, installs
│ (js/cli-table.json is the CLI as data; both runtimes read it)
├── geneseed launcher (bash): a shim over bin/geneseed-cli.mjs — bare `./geneseed`
│ = interactive main menu; + every subcommand the CLI carries
│ (`./geneseed link` puts it on PATH so `geneseed` runs from anywhere)
├── geneseed.cmd the same shim for cmd.exe / PowerShell — no bash needed
├── harness.config.json default theme + metadata (the one owner of the version)
├── src/ canonical source — edit here
│ ├── AGENT.md.tmpl the entrypoint, rendered to AGENT.md
│ ├── ontology/ how the agent thinks — Telos, Evidence, Decisions, Conduct
│ ├── laws/ the universal invariants
│ ├── doctrines/ the four practice packs (craft, rigor, ops, process)
│ ├── postures/ the relationship register, inlined into AGENT.md
│ ├── modes/ the operating register, inlined into AGENT.md
│ ├── agents/ capability specialists
│ ├── skills/ repeatable workflows
│ ├── memory/ memory convention + index
│ └── notebook/ the agent's own freeform space — convention + index
├── themes/ voice token maps (14 themes shipped)
├── web/ Vite + React UI source; the committed web/dist/ build is what ships
├── tests/ Node test suites + the pure-function/help recordings they replay
│ (tests/__snapshots__/ — the emit/cli/web corpora retired 2026-08-17)
├── docs/ guides (web-ui, wiki, …) + docs/web/ (the console's Docs pages);
│ specs/, reviews/, superpowers/ are local working docs — git-ignored
├── adapters/ per-host glue (opencode/, claude-code/, bob/, copilot/)
└── .github/workflows/ ci.yml (doctor + tests) · publish.yml (npm, OIDC, manual only)
node bin/geneseed-cli.mjs doctor # every theme + parity + authoring + drift
node --test "tests/**/*.test.mjs" # the test suites (node expands the glob)
doctor checks each theme for unresolved tokens, dead/non-hermetic links, theme-key parity, author-time gates (every spec has a purpose line, the plugins parse, the learn-prompt literal stays extractable), and that a committed bundle still matches a fresh render of src/. CI (.github/workflows/ci.yml) runs both on every push and PR, on both Linux and Windows. Publishing is a separate, manually-triggered workflow (.github/workflows/publish.yml) — see Contributing.
Installed from npm — one command, and it rebuilds every active install for you:
npm install -g geneseed@latest
geneseed rebuild-all # re-render every registered install in its own theme + mode
Installed from a clone — the git pull route, which also refreshes the launchers:
./geneseed update # everything in one: refresh the scripts + content, then rebuild
./geneseed bootstrap # update everything, then drop into the setup wizard
./geneseed upgrade # just the content refresh (remembers theme + emit mode)
Local edits survive. The self-improvement loops let the agent refine its deployed agent/skill files in place. Before setup, re-theme, or upgrade overwrites them, any drift is auto-exported to a markdown improvements file under improvements/ inside the deployed harness dir (e.g. ~/.config/opencode/improvements/ for the global install) — beside the install it describes, untouched by rebuilds and uninstall. Hand it to an agent in this repo to back-port the changes into src/. On demand: ./geneseed diff --out FILE, or the Changes page in the web console.
Details and precedence rules: SETUP.md → Upgrade.
| Page | Read it when… |
|---|---|
| SETUP.md | Installing — every path, configuration knob, env var, verify, troubleshooting |
| DESIGN.md | Changing structure — the spec and the decisions behind it |
| SHIPPED.md | What's in the harness today — capabilities ↔ the spec behind each |
| docs/web-ui.md | The web console — every view, the launch/daemon/PWA surface, security model |
| docs/wiki.md | The machine wiki — your personal knowledge base, setup and writing model |
| docs/token-footprint.md | What the harness costs in context-window tokens, per host |
| docs/opencode-plugin-setup.md | Installing the OpenCode plugins — the one-time wiring they all share |
| CHANGELOG.md | What changed between versions |
| adapters/opencode/ | Wiring OpenCode in depth — plugins, native mapping |
| ⤷ GLOBAL-HARNESS-SPEC.md | The global-emit contract |
| ⤷ HOW-OPENCODE-LOADS.md | Why a file shows up twice; plugin |
| adapters/claude-code/ | The Claude Code hook adapter |
| adapters/bob/ | The IBM Bob adapter — Claude-shaped, rules-file preamble |
| adapters/copilot/ | The GitHub Copilot adapter — reduced host, no hooks |
| src/memory/README.md | The memory convention |
| src/notebook/README.md | The agent's own freeform-space convention |
Issues and PRs welcome at github.com/Arylmera/Geneseed. The CI is dependency-free and runs on every push — keep doctor green and the test suites passing. Adding a new theme is one JSON file in themes/ with the same voice-token keys; doctor will tell you if any are missing.
Three things that bite when you don't know them:
js/cli-table.jsonIS the CLI. The argument parser is data, and that file is the owned document — not a generated one.bin/geneseed-cli.mjscannot parse a single verb without it,bin/geneseed-hook.mjsrenders--helpfrom it, and the console'sclidocs page is a filtered view of it. What holds it honest istests/unit/cli_table.test.mjsplus the recorded help texts undertests/__snapshots__/help/, which every verb's--helpis rendered against — so change a flag here and the fixture for that verb is what tells you. Those recordings are frozen and cannot be re-taken; a red one is a finding, not a re-bless.- The version has one owner:
harness.config.json.package.jsonmirrors it and a test fails the fork. Nevernpm version— it edits one of the two. - Publishing is deliberate and manual.
.github/workflows/publish.ymluses npm trusted publishing (OIDC); there is noNPM_TOKENin this repository and there must not be one. It runs only from Actions → publish → Run workflow, and the npm-side trusted publisher is keyed on that workflow'sfilename — renaming the file breaks publishing with no local symptom, which is why the file names itself in its own header and a test asserts the two agree.