Must-Have Features — Long-Term Plan Source scope: 12+ open-source projects studied April 2026.
Projects: hermes-agent (Python upstream), honcho (Python memory), gbrain (TypeScript memory/runtime), browser-harness / go-browser-harness, mercury-agent (TypeScript), space-agent (JS browser-first), picoclaw (Go lightweight agent), MLT-OSS/hermes-agent-go (Go port), trpc-agent-go, AgenticGoKit, claw-prime, langchaingo, 8 go-agent-os reference repos.
Audience: gormes-agent planner, builder, and reviewer skills. Not end-user marketing.
Relationship to progress.json: This document names the features gormes must eventually have. The canonical execution queue is progress.json. This document is the why and the what . progress.json is the when and the how .
Feature Architecture Every feature is classified by three axes:
Axis Values Meaning Criticality foundational / production / differentiator / commodityHow essential the feature is to gormes’s existence Maturity shipped / in-flight / planned / gapCurrent implementation status Complexity trivial / moderate / large / subsystemImplementation effort required
A feature is must-have if it scores foundational on criticality OR production on criticality with gap maturity. Differentiators are must-have when they define gormes’s unique value. Commodity features are must-have only when they block user adoption.
1. Provider Adapters — The Model Surface Donor projects: hermes-agent, picoclaw, mercury-agent, gbrain, trpc-agent-go, langchaingo
Current parity: ~70% (7/10 major providers)
Criticality: foundational
Every provider gormes cannot speak is a provider the operator cannot use without running Python Hermes alongside. This is not a nice-to-have. It is the single biggest block to a Python-free agent turn.
1.1 Must-Have Provider Completion Provider Status Rationale Anthropic Messages API Shipped Primary provider for Gormes development OpenAI Chat Completions Shipped Largest model ecosystem Codex Responses API Partial Required for Codex CLI parity AWS Bedrock (Converse + Anthropic) In-flight Enterprise deployment vector Gemini / Google Cloud Code Partial Google ecosystem Google Code Assist Partial Google developer tooling OpenRouter Partial Multi-model routing; 250+ models DeepSeek / Kimi reasoning Partial Reasoning-model isolation needed Moonshot / Kimi tool schema Partial Chinese model ecosystem xAI / Grok Gap Growing provider, no Go adapter LM Studio / local inference Gap Offline-capable agent runtime Azure Foundry (OpenAI + Anthropic) In-flight Enterprise Azure deployment
1.2 Provider Infrastructure (Must-Have) Feature Source Status Provider-neutral stream contract hermes-agent, gbrain Shipped Tool-call normalization + continuation hermes-agent Shipped Provider registry + alias manifest hermes-agent, picoclaw Shipped Retry-After header parsing + backoff plandex (go-agent-os) Shipped Provider error taxonomy (rate-limit vs auth vs capacity) plandex, hermes-agent Shipped Prompt-cache capability guard hermes-agent Shipped Provider rate guard + degraded-state evidence hermes-agent Shipped Provider timeout config fail-closed hermes-agent Shipped Cross-provider reasoning-tag sanitization hermes-agent Shipped Tool-call argument repair + schema sanitizer hermes-agent Shipped Model metadata registry (context limits, pricing, capabilities) hermes-agent, gbrain In-flight Smart model routing with fallback chain mercury-agent (DeepSeek→OpenAI→Anthropic→Grok→Ollama) In-flight Per-turn model selection + reasoning effort propagation hermes-agent Shipped Provider account usage read model (/usage) hermes-agent Shipped
2. Native Prompt Assembly — The Agent’s Brain Donor projects: hermes-agent (prompt_builder.py), mercury-agent (SOUL/persona system), space-agent (prompt-items.js)
Current parity: ~40%
Criticality: foundational
Without native prompt assembly, the Go runtime cannot construct a correct Hermes-compatible system prompt. Every turn depends on Python’s prompt builder. This is the critical path for Python-free operation.
2.1 System Prompt Block Assembly Block Source Status SOUL.md identity + AGENTS.md project context hermes-agent, mercury-agent (soul.md) Shipped USER.md + MEMORY.md durable user context hermes-agent, space-agent (plain markdown includes) Shipped Tool/memory guidance constants (byte-equivalent port) hermes-agent (prompt_builder.py) Shipped Skills guidance block (toolset-aware) hermes-agent, space-agent (metadata-driven placement) In-flight Model-specific role + tool-use enforcement hermes-agent Shipped Timestamp + model/provider/session metadata hermes-agent (run_agent.py:3770-3779) Shipped Platform/session context block hermes-agent Shipped Ephemeral memory/plugin recall (injected into user turn) hermes-agent Gap WSL environment hint hermes-agent Shipped Developer-role model routing hermes-agent Shipped
2.2 Prompt Budgeting (Must-Have from space-agent design) Feature Source Status Prompt contributors as keyed, budgeted items space-agent (prompt-items.js) Gap Cached token counts per contributor space-agent Gap Trim policy with runtime read-back seam space-agent Gap Long-context trimming without silent text drop space-agent Gap
3. Context Compression — The Long Session Engine Donor projects: hermes-agent (ContextEngine), gbrain (tiered enrichment)
Current parity: ~60%
Criticality: production
Real agent sessions run long. Without compression, the context window fills and the agent degrades. Hermes’s ContextEngine is a well-defined contract. Gormes must port it completely.
3.1 Compression Engine Feature Source Status ContextEngine interface + status contract hermes-agent Shipped Token-budget trigger + summary sizing hermes-agent Shipped Protected head/tail message invariants hermes-agent Shipped Tool call/result pairs kept together hermes-agent Shipped Compression lineage record hermes-agent Shipped Auxiliary compression headroom for system + tool schemas hermes-agent Shipped Provider-aware context cap resolution hermes-agent Shipped Tool-result pruning hermes-agent Shipped Image-token budget charge hermes-agent Shipped Compression-boundary callback (notify kernel) hermes-agent In-flight Manual compression feedback + context references hermes-agent Gap Kernel compression-boundary callback binding hermes-agent Gap Recalculate budgets on model-switch hermes-agent (2026-04-25 sync) Gap
4. Gateway + Channels — The Operator Surface Donor projects: hermes-agent (gateway/run.py), picoclaw (18+ channels), mercury-agent (permission-hardened Telegram)
Current parity: ~60% (5/19 channels shipped; shared chassis in place)
4.1 Shared Gateway Infrastructure (Must-Have) Feature Source Status Shared gateway chassis (multi-channel) hermes-agent, gormes Shipped Slash command registry + active-turn policy hermes-agent Shipped Session store + SessionSource parity hermes-agent Shipped Session metadata refresh (created, activity, platform, title) hermes-agent Shipped Gateway auto-title generation hermes-agent Shipped Gateway /status Hermes-format closeout hermes-agent Shipped Gateway /title manual title command hermes-agent Shipped Gateway session token accounting hermes-agent Shipped Gateway stream/tool trace formatting hermes-agent Shipped Gateway typing-action wiring during stream hermes-agent Shipped Placeholder edit-failure fallback hardening hermes-agent Shipped Gateway fresh-final send/delete fallback hermes-agent Shipped Gateway message deduplicator hermes-agent Shipped Gateway inbound dedup manager binding hermes-agent Shipped Notify-to delivery routing hermes-agent Gap Channel directory refresh + stale-target invalidation hermes-agent Gap Mirror + sticker cache surfaces hermes-agent Gap
4.2 Channel Adapters (Must-Have by Priority) Channel Status Priority Rationale Telegram Shipped Primary dogfood channel Discord Shipped Developer community channel Slack Shipped Enterprise channel WhatsApp Shipped Global messaging WeChat (WeCom + WeiXin) Shipped Chinese enterprise BlueBubbles (iMessage) Shipped Apple ecosystem Home Assistant Shipped Smart home integration Signal Partial Privacy-focused users Email + SMS Shipped Fallback delivery Matrix Gap Open federated protocol Mattermost Gap Self-hosted enterprise Feishu / Lark Partial Chinese enterprise DingTalk Partial Chinese enterprise QQ Partial Chinese consumer LINE Gap Japanese market VK Gap Russian market IRC Gap Legacy community Yuanbao Partial Chinese AI platform Webhook (generic) Shipped Universal integration
4.3 Gateway Active-Turn Policy (Must-Have from Hermes Design) Feature Source Status /stop interrupthermes-agent Shipped /queue waithermes-agent Shipped /steer mid-turn injectionhermes-agent Shipped Busy command guard for compression + long CLI actions hermes-agent Shipped Gateway-handled slash commands bypass active-session guard hermes-agent Shipped Mid-run steer injection between tool calls hermes-agent Shipped Hardline command blocks (yolo/approval-off/cron) hermes-agent (2026-04-26 sync) Partial Approvals bypass interrupt semantics hermes-agent Gap
Donor projects: hermes-agent (40+ tools), mercury-agent (31 tools, permission-hardened), picoclaw (Go native tools), gbrain (operation catalog)
Current parity: ~25% (15/61 tools)
Criticality: foundational
Feature Source Status Typed Tool interface + in-process registry hermes-agent, gormes Shipped Tool descriptor layer (toolset, availability, mutating flag, trust class, timeout, result budget, audit kind, prompt-visible flag)gbrain (OperationSpec), hermes-agent Gap Streamed tool_calls accumulation hermes-agent Shipped Kernel tool loop (90-turn budget, summary grace turn) hermes-agent Shipped Doctor verification (--offline) hermes-agent, gbrain Shipped Tool registry inventory + schema parity harness hermes-agent Shipped Executor-enforced trust class + timeout before handler runs gbrain (trust-class enforcement) Gap
Tool Source Status read_file hermes-agent Shipped search_files hermes-agent Shipped write_file hermes-agent Shipped patch hermes-agent Shipped terminal (foreground) hermes-agent Shipped Terminal (background, PTY) hermes-agent Partial todo (session-bound)hermes-agent Shipped session_search (session-bound)hermes-agent Shipped clarify hermes-agent Shipped Checkpoint manager (atomic snapshot, shadow-repo)hermes-agent In-flight Code execution (sandboxed) hermes-agent Shipped Process registry hermes-agent Gap Fuzzy patch/lint/checkpoint restore hermes-agent Gap
Tool Source Status Browser navigate/snapshot/click/type/scroll hermes-agent, go-browser-harness Shipped Browser console/vision/CDP hermes-agent, go-browser-harness Shipped Browser dialog/get-images hermes-agent, go-browser-harness Shipped Web search (DuckDuckGo, Tavily, Brave, Perplexity, SearXNG) hermes-agent, picoclaw Shipped Web extract + crawl hermes-agent Shipped Browser daemon lifecycle (start/stop/ensure)hermes-agent Gap Browser profile management hermes-agent Gap Browser tab enumeration/switch hermes-agent Gap Browser doctor/diagnostics hermes-agent Gap Browser event drain hermes-agent Gap Browser Use cloud + Go harness bridge hermes-agent, go-browser-harness Shipped HTTP GET (no-browser) hermes-agent Gap Iframe targeting hermes-agent Gap
Tool Source Status Image generation hermes-agent Shipped Image input mode router + native content parts hermes-agent Shipped Vision analysis hermes-agent Gap TTS synthesis + voice-mode state hermes-agent Gap Transcription tool hermes-agent Shipped
5.5 Delegation + Subagents Feature Source Status Deterministic subagent runtime (goroutines) hermes-agent, gormes Shipped Max-depth guard + bounded batch execution hermes-agent, gormes Shipped Timeout + cancellation scopes (context.Context) hermes-agent, gormes Shipped Typed result envelope + append-only run log hermes-agent, gormes Shipped Runner-enforced tool allowlists + blocked-tool policy hermes-agent, gormes Shipped Interrupt propagation to concurrent-tool workers hermes-agent, gormes Shipped Child-agent message/tool-call replay from ledger gbrain (subagent_messages) Shipped Durable subagent/job ledger (SQLite)gbrain (Minions), gormes Shipped Background review toolset restriction hermes-agent Shipped
5.6 Scheduling + Background Tasks Feature Source Status robfig/cron scheduler + bbolt/SQLite job store hermes-agent Shipped Heartbeat [SYSTEM:] + [SILENT] delivery contract hermes-agent Shipped Durable job queue (claim, renew, complete, fail, retry, cancel)gbrain (Minions), gormes Shipped Durable job backpressure + timeout audit gbrain, gormes Shipped Durable worker supervisor status seam gbrain, gormes Shipped Durable pause/resume intent contract gbrain, gormes Shipped Durable replay and inbox message contract gbrain, gormes Shipped Cronjob tool API + schedule parser parity hermes-agent Gap Parent-child DAGs with completion policies gbrain (Minions) Gap Per-job token accounting gbrain Gap Cron multi-target delivery hermes-agent Shipped Cron context_from output chaining hermes-agent Shipped
5.7 MCP (Model Context Protocol) Feature Source Status MCP client (stdio + HTTP transport) hermes-agent, gormes Shipped MCP tool discovery + schema normalization hermes-agent, gormes Shipped MCP OAuth state store + token refresh hermes-agent, gormes Shipped Goncho MCP tool catalog gormes Shipped Gormes-native MCP host runtime boundary gormes Shipped
6. Memory System — Goncho / Honcho Parity Donor projects: honcho (3-agent memory, peer paradigm), gbrain (knowledge graph, brain-first lookup), mercury-agent (Second Brain, 10 typed memories), hermes-agent (MemoryProvider interface)
Current parity: Shipped (Goncho Phase 3 complete)
Criticality: production (core is shipped; enhancements are differentiators)
6.1 Shipped Memory Foundation Feature Status SQLite + FTS5 lattice + schema migrations Shipped Ontological graph + LLM extractor + dead-letter queue Shipped Neural recall (2-layer seed selection, CTE traversal) Shipped Semantic fusion (Ollama embeddings, cosine similarity, hybrid) Shipped Memory mirror (USER.md async export) Shipped Session index mirror (bbolt→index.yaml) Shipped Tool execution audit log (JSONL) Shipped Transcript export (gormes session export) Shipped Extraction state visibility (gormes memory status) Shipped Insights audit log (daily usage.jsonl) Shipped Memory decay (relationships.last_seen + weight attenuation) Shipped Cross-chat synthesis (user_id concept, opt-in cross-chat recall) Shipped Session lineage + cross-source search Shipped Honcho-compatible scope/source tool schemas Shipped Honcho host + SillyTavern integration fixtures Shipped Goncho configuration namespace + dreaming scheduler Shipped Goncho streaming chat persistence Shipped Hermes memory tool over Goncho/local store Shipped Goncho drop-in compatibility (keys, webhooks, HTTP, CLI, SDK) Shipped
6.2 Memory Enhancements (Must-Have Differentiators) Feature Source Status Typed memory categories (identity, preference, goal, habit, episode, reflection)mercury-agent (10 types), honcho Gap Confidence + durability scoring per observationmercury-agent, honcho Gap Conflict resolution (confidence wins; equal→newer)mercury-agent, honcho Gap Auto-pruning (21-day stale memories)mercury-agent Gap Zero-LLM knowledge graph wiring (regex-based auto-links)gbrain Gap Brain-first lookup (5-step before external API)gbrain Gap Retrieval eval harness (precision@k, recall@k, cross-chat negative tests)gbrain Gap Source-aware retrieval ranking (curated/reviewed boost, bulk-source damping)gbrain Gap Compiled truth + timeline pattern (per-entity page)gbrain Gap Directional peer cards (users AND agents as peers)honcho Gap Multi-memory backends (Turbopuffer, LanceDB, Redis cache layer)honcho Gap Three-agent memory loop (Deriver/Dialectic/Dreamer)honcho Gap
7. CLI / Operator Surface — The Control Panel Donor projects: hermes-agent (49-file CLI tree), mercury-agent (token budget, loop detection)
Current parity: ~50%
Criticality: production
7.1 CLI Commands (Must-Have by Function) Command Source Status gormes (TUI, oneshot, gateway modes)hermes-agent Shipped gormes doctorhermes-agent Shipped gormes gateway status/start/stophermes-agent Shipped gormes session exporthermes-agent Shipped gormes memory statusgormes Shipped gormes config (check, edit, migrate)hermes-agent Shipped gormes auth (login, logout, status, add)hermes-agent In-flight gormes model (interactive picker)hermes-agent Shipped gormes setup (wizard)hermes-agent Shipped gormes profilehermes-agent Shipped gormes agent-template-resethermes-agent Shipped gormes mcp loginhermes-agent Shipped gormes backup / gormes restorehermes-agent Gap gormes logshermes-agent Gap gormes web-serverhermes-agent Gap gormes uninstallhermes-agent Gap gormes cron management CLIhermes-agent Gap gormes plugins managementhermes-agent Gap Diagnostics, backup, and status CLI closeout hermes-agent Gap
7.2 API Server (Must-Have) Feature Source Status OpenAI-compatible chat-completions API hermes-agent Shipped Responses API store + run event stream hermes-agent Shipped API server disconnect snapshot persistence hermes-agent Shipped Gateway proxy mode forwarding hermes-agent Shipped Dashboard API client contract hermes-agent Shipped API server detailed health endpoint hermes-agent Shipped API server cron admin endpoints hermes-agent Shipped SSE streaming to Bubble Tea TUI gormes Shipped
7.3 TUI (Terminal UI) Feature Source Status Bubble Tea shell + 16ms coalescing mailbox gormes Shipped SSE reconnect gormes Shipped Hermes skin token renderer hermes-agent, gormes Shipped Hermes status bar + bottom-pinned chrome layout hermes-agent, gormes Shipped Native TUI slash-command dispatch gormes Shipped Native TUI /save, /branch, conversation viewport gormes Shipped Web dashboard (TypeScript/React parity with Hermes UI)hermes-agent Gap
8. Security + Safety — The Trust Boundary Donor projects: mercury-agent (permission system), hermes-agent (approval, URL safety, Tirith), gbrain (trust-class enforcement)
Current parity: ~30%
Criticality: foundational
Gormes has zero permission hardening. This is the most critical gap in the entire project. An agent that can execute arbitrary shell commands, write to any filesystem path, and make arbitrary network requests is not safe to operate.
8.1 Must-Have Security Features Feature Source Status Shell blocklist (36+ dangerous patterns)mercury-agent Gap Filesystem scoping (folder-level read/write restrictions)mercury-agent Gap Permission approval UX (inline y/n/always)mercury-agent, hermes-agent Gap Hardline command pattern detection (non-recoverable host destruction) hermes-agent, gormes In-flight Recoverable dangerous patterns + blocked-result schema hermes-agent, gormes In-flight Approval mode config normalization hermes-agent, gormes In-flight Gateway hook auto-accept strict parser hermes-agent, gormes Shipped Subagent dangerous-command non-interactive policy gormes Shipped URL safety / SSRF guard hermes-agent Partial Path security (traversal prevention, workspace confinement)hermes-agent Gap Credential redaction in logs and telemetryhermes-agent Partial Website policy (per-domain allow/block)hermes-agent Gap Cron dangerous-command approval mode hermes-agent Gap Tirith security integration hermes-agent Gap OSV vulnerability check hermes-agent Gap Trust-class enforcement in shared tool executor gbrain, mercury-agent Gap WhatsApp identity safety (ASCII-only, anti-traversal)hermes-agent (2026-04-27 sync) Shipped
8.2 Operator Safety (Must-Have from Mercury) Feature Source Status Loop detection (5 types: hard/failing/text/no-action/same-tool)mercury-agent Gap Token budget system (daily + per-session, auto-concise at 70%)mercury-agent Gap Permission manifest (YAML-based, per-agent allow/deny)mercury-agent Gap Filesystem operation preflight (batch validation before writes) space-agent Gap
9. Skills System — The Agent’s Knowledge Donor projects: hermes-agent (skills_hub.py, 118K lines), gbrain (fat skills, thin harness), space-agent (metadata-driven skills), mercury-agent (skill preprocessing)
Current parity: ~40%
Criticality: differentiator
9.1 Skills Infrastructure Feature Source Status SKILL.md parsing + active store hermes-agent, gormes Shipped Deterministic selection + prompt injection block hermes-agent, gormes Shipped Kernel injection + usage log hermes-agent, gormes Shipped Inactive candidate drafting + explicit promotion hermes-agent, gormes Shipped SKILL.md frontmatter validation guard gbrain, gormes Shipped Hermes creative skill metadata compatibility hermes-agent, gormes Shipped Skill preprocessing + dynamic slash commands hermes-agent, gormes Shipped Skills hub search (in-memory registry) hermes-agent, gormes Shipped Skills list (enabled/disabled status + filter) gormes Shipped Metadata-driven skill placement (when, loaded, placement, context tags)space-agent Gap Skill conformance checks + routing evals gbrain Gap Skill conflict detection (overlapping triggers)gbrain Gap Bundled productivity skills (Airtable, Spotify, TouchDesigner, Google Meet)hermes-agent In-flight Skill registries (external repo fetching)hermes-agent Gap Skill confidence/feedback scoring hermes-agent Gap Skill versioning + review history gbrain Gap
10. Sandboxing + Execution Environments Donor projects: hermes-agent (Docker, Modal, Daytona, Singularity), picoclaw (7 terminal backends)
Current parity: ~30%
Criticality: production (Docker is must-have; others are differentiators)
Backend Source Status Priority Docker hermes-agent In-flight Must-have Environment interface + file sync contract hermes-agent, gormes Shipped Must-have Modal hermes-agent Gap Differentiator Daytona hermes-agent Gap Differentiator Singularity hermes-agent Gap Differentiator Local (SSH, direct exec) hermes-agent, picoclaw Shipped Shipped Code execution mode policy (strict/project/default resolver)hermes-agent Gap Must-have
11. Packaging + Distribution Donor projects: hermes-agent, picoclaw (Desktop/Headless/Android/Docker/Termux), claw-prime
Current parity: ~60%
Criticality: production
Artifact Status Priority Single Go binary (Linux, macOS, Windows) Shipped Foundational OCI / Docker image Shipped Must-have Homebrew formula Shipped Must-have Unix installer (install.sh) source-backed update flow Gap Must-have Windows installer (install.ps1 + install.cmd) Gap Must-have Installer site asset/route coverage Gap Must-have Debian/RPM packages Gap Commodity Android APK / Termux Gap Commodity
12. Soul / Personality System — Phase 6 Donor projects: mercury-agent (soul.md, persona.md, taste.md, heartbeat.md), space-agent (personality.system.include.md)
Current parity: Not started (Phase 6)
Criticality: differentiator
Feature Source Status SOUL.md identity (heart) mercury-agent Planned persona.md behavior (face) mercury-agent Planned taste.md preferences (palate) mercury-agent Planned heartbeat.md lifecycle (breathing) mercury-agent Planned User-owned markdown files (operator-editable) mercury-agent, space-agent Planned Plain-text memory includes as first-class space-agent Planned Git-backed personality time travel (rollback/revert)space-agent Planned
13. Learning Loop — Phase 6 Donor projects: gbrain (skill extraction, maintenance cycles), hermes-agent (skill generation)
Current parity: Not started (Phase 6)
Criticality: differentiator
Feature Source Status Complexity detector (heuristic/LLM signal) hermes-agent, gormes Gap Skill extractor (LLM-assisted pattern distillation) hermes-agent, gormes Gap Hybrid lexical + semantic skill lookup gbrain, gormes Gap Skill effectiveness scoring + feedback loop gbrain, gormes Gap TUI + Telegram skill browsing gormes In-flight Delta-bounded skill/memory maintenance (changed-source IDs only)gbrain (e2961c0) In-flight Source-aware retrieval damping fixtures gbrain Gap Code Cathedral II (call-graph edges, two-pass retrieval)gbrain Gap Incremental extract + embed-stale filters gbrain (v0.22.1) Gap Per-source sync anchor (source-scoped last-commit)gbrain (v0.22.5) Gap Fail-improve loop (logs regex failures, generates better patterns)gbrain Gap
14. Observability + Telemetry Donor projects: hermes-agent (logs, status, tips), gbrain (doctor, degraded mode), mercury-agent (heartbeat)
Current parity: ~40%
Criticality: production
Feature Source Status gormes doctor --offlinehermes-agent Shipped gormes gateway status (read-only)hermes-agent, gormes Shipped gormes memory status (extractor queue, dead-letter)gormes Shipped Tool execution audit log (JSONL) gormes Shipped Insights audit log (daily usage.jsonl) gormes Shipped CLI log redactor (secret shapes) hermes-agent Shipped CLI log snapshot reader hermes-agent Shipped CLI dump support-summary helper hermes-agent Shipped Visible degraded mode (every fallback reported in status/doctor/audit)gbrain, mercury-agent Gap Semantic recall disabled warning (no embedding model)gbrain Gap Extractor queue depth + dead-letter count gbrain, gormes Shipped Graph extraction age gbrain Gap Skill resolver warnings gbrain Gap Durable job stalled count gbrain, gormes Shipped Child-agent replay not available warning gbrain Gap Tips system (deterministic, seen-state)hermes-agent, gormes Shipped Web dashboard (session, skills, chat, config, logs)hermes-agent Gap
15. Plugin Architecture Donor projects: hermes-agent (plugin SDK, hooks, tools, slash commands), gbrain (isolated MCP plugins)
Current parity: ~40%
Criticality: commodity (must-have infrastructure; plugins themselves are differentiators)
Feature Source Status Plugin SDK hermes-agent Shipped Dashboard theme/plugin slot inventory gormes Shipped First-party Spotify plugin fixture hermes-agent, gormes Shipped First-party Google Meet plugin fixture hermes-agent, gormes Shipped Third-party extension system hermes-agent Gap Plugin isolation (subprocess/WASM for untrusted)gbrain, hermes-agent Gap Plugin manifest + capability declaration gbrain Gap Operator enablement required before activation gbrain Gap Hindsight memory setup hermes-agent, gormes Gap
16. Mixture of Agents Donor projects: hermes-agent (multi-model coordination), gbrain (subagent orchestration)
Current parity: Gap
Criticality: differentiator
Feature Source Status Multi-model coordination hermes-agent Gap Agent ensemble with voting/consensus hermes-agent Gap Model-specific task routing hermes-agent Gap
Cross-Cutting Must-Have Patterns (from All Projects) These are not features but architectural patterns that every subsystem must adopt:
Pattern Source Applied To Contract-first operations (one descriptor → CLI, MCP, tool schemas, doctor)gbrain Tools, commands, gateways Trust-class enforcement in shared executor (operator/gateway/child/system)gbrain, hermes-agent All tool execution Visible degraded mode (every fallback reported)gbrain, mercury-agent All subsystems Provider-neutral events (adapters own quirks; kernel sees one contract)hermes-agent Provider layer Thin harness, fat skills (intelligence in skills, not runtime)gbrain Skills system Hermetic fixtures over live tests (replayable, no live credentials)gbrain, hermes-agent All tests Metadata-driven discovery (not hardcoded branches)space-agent Skills, tools, commands Prompt contributors as keyed, budgeted items space-agent Prompt assembly Goroutines + channels (not asyncio threads)All Go projects Concurrency Context.Context propagation (cancellation, deadlines)All Go projects All I/O SQLite + FTS5 first (Postgres optional behind interface)gbrain, gormes Memory, jobs, state Single binary, zero runtime deps claw-prime, gormes Distribution Durable job ledger (restartable work, not ephemeral goroutines)gbrain Cron, subagents, long work Provenance-rich memory (source, extractor, confidence, freshness)gbrain, honcho Goncho relationships Skills as reviewed code (frontmatter, conformance, routing evals)gbrain, hermes-agent Skills store
Feature Gap Summary by Criticality Foundational Gaps (Block Python-Free Agent Turn) # Feature Source Project Estimated Effort F1 Provider completion (xAI/Grok, LM Studio, DeepSeek, Moonshot) hermes-agent, picoclaw 4-6 weeks F2 Native prompt builder assembly (skills snapshot, memory/search guidance) hermes-agent 2-3 weeks F3 Tool descriptor layer (toolset, trust class, timeout, audit, budget) gbrain 2-3 weeks F4 Shell blocklist + filesystem scoping mercury-agent 2-3 weeks F5 Permission approval UX (inline y/n/always) mercury-agent, hermes-agent 2-3 weeks F6 Trust-class enforcement in shared executor gbrain 1-2 weeks
Production Gaps (Block Real-World Operation) # Feature Source Project Estimated Effort P1 Context compression complete (manual feedback, model-switch recalculation) hermes-agent 2-3 weeks P2 Loop detection (5 types) mercury-agent 1-2 weeks P3 Token budget system + auto-concise mercury-agent 2-3 weeks P4 URL safety / SSRF guard complete hermes-agent 1-2 weeks P5 Browser daemon lifecycle + doctor hermes-agent, go-browser-harness 2-3 weeks P6 Docker sandbox backend hermes-agent 1-2 weeks P7 Code execution mode policy (strict/project/default) hermes-agent 2-3 weeks P8 CLI closeout (backup, logs, web-server, uninstall, diagnostics) hermes-agent 3-4 weeks P9 Packaging closeout (install.sh, install.ps1, install.cmd) hermes-agent 2-3 weeks P10 Visible degraded mode in all subsystems gbrain, mercury-agent Ongoing P11 Channel adapters (Matrix, Mattermost, LINE, IRC, VK, Feishu/DingTalk closeout) hermes-agent, picoclaw 4-6 weeks
Differentiator Gaps (Define Gormes’s Unique Value) # Feature Source Project Estimated Effort D1 Typed memory categories + confidence scoring mercury-agent, honcho 3-4 weeks D2 Zero-LLM knowledge graph wiring (regex auto-links) gbrain 2-3 weeks D3 Brain-first lookup (5-step before external API) gbrain 2-3 weeks D4 Retrieval eval harness (precision@k, recall@k) gbrain 2-3 weeks D5 Metadata-driven skill placement (when, loaded, context tags) space-agent 2-3 weeks D6 Minions DAG job dependencies (parent-child with policies) gbrain 3-4 weeks D7 Soul/personality system (soul.md, persona.md, taste.md, heartbeat.md) mercury-agent 2-3 weeks D8 Learning loop (complexity detector, skill extractor, feedback) hermes-agent, gbrain 6-8 weeks D9 Code Cathedral II (call-graph edges, two-pass retrieval) gbrain 4-6 weeks D10 Web dashboard (TypeScript/React parity with Hermes UI) hermes-agent 6-8 weeks D11 Multi-memory backends (Turbopuffer, LanceDB, Redis) honcho 3-4 weeks D12 Three-agent memory loop (Deriver/Dialectic/Dreamer) honcho 4-6 weeks D13 Mixture of agents (multi-model coordination) hermes-agent 6-8 weeks
Implementation Sequence (Recommended) Week 1-2: Provider completion (xAI/Grok, LM Studio, DeepSeek/Kimi)
Week 1-2: Tool descriptor layer (OperationSpec with trust classes)
Week 2-3: Prompt builder assembly final closeout
Week 3-4: Shell blocklist + filesystem scoping + permission approval
Week 3-4: Trust-class enforcement in tool executor
Copy Short-Term (Next 90 Days) — Production Hardening Week 5-6: Context compression complete (manual feedback, recalculation)
Week 5-6: Loop detection (5 types)
Week 7-8: Token budget system + auto-concise
Week 7-8: Docker sandbox backend
Week 9-10: Browser daemon lifecycle + doctor
Week 9-10: Code execution mode policy
Week 11-12: CLI closeout (backup, logs, diagnostics)
Week 11-12: Packaging closeout (installers)
Copy Medium-Term (Next 6 Months) — Differentiators Month 3: Typed memory categories + confidence scoring
Month 3: Zero-LLM knowledge graph wiring
Month 4: Brain-first lookup + retrieval eval harness
Month 4: Minions DAG job dependencies
Month 5: Metadata-driven skill placement
Month 5: Soul/personality system
Month 6: Channel adapters (Matrix, Mattermost, LINE, IRC)
Copy Long-Term (Next 12 Months) — Capstone Features Month 7-8: Learning loop (skill extraction, feedback)
Month 8-9: Web dashboard
Month 9-10: Code Cathedral II
Month 10-11: Multi-memory backends
Month 11-12: Three-agent memory loop + Mixture of agents
Copy Not Recommended for Gormes Features studied but explicitly excluded from the must-have list:
Feature Source Reason Browser-first runtime (agent lives in browser) space-agent Fundamentally different runtime model WebLLM / HuggingFace browser-side inference space-agent Requires browser runtime, not Go Mercury’s Telegram org model (admin/member roles) mercury-agent Gormes gateway channels use different abstraction Picoclaw’s ultra-lightweight constraints ($10 hardware) picoclaw Different deployment target Full Kubernetes ACP controllers go-agent-os references Use local state machine instead TypeScript indexer / tree-sitter WASM in runtime gbrain Keep runtime small; optional code-context evidence only Hermes’s dynamic Python import-time tool registration hermes-agent Explicit registration is safer for Go Hermes’s sync/async/thread complexity hermes-agent Go goroutines + context.Context are sufficient Hermes’s Postgres-backed cron hermes-agent SQLite-backed is correct for single-binary promise
References Generated: April 30, 2026
Source: 12+ open-source projects, 7-phase gormes roadmap, cross-project feature map, upstream lessons
← Previous
Implementation Roadmap
Next →
Architecture Plan