Skip to content

spec: v0.19 — The Agentic Turn#53

Draft
codejunkie99 wants to merge 1 commit into
masterfrom
spec/v0.19-agentic-turn
Draft

spec: v0.19 — The Agentic Turn#53
codejunkie99 wants to merge 1 commit into
masterfrom
spec/v0.19-agentic-turn

Conversation

@codejunkie99
Copy link
Copy Markdown
Owner

@codejunkie99 codejunkie99 commented May 25, 2026

Draft spec for converting agentic-stack from a static portable brain into an active multi-agent runtime.

Adds spec.md covering eight new subsystems:

  1. Plans layer (.agent/plans/) — a fifth memory layer for intent.
  2. Multi-agent bus (.agent/bus/) — append-only JSONL coordination across harnesses.
  3. Eval runner (.agent/evals/) — turns accepted lessons into regression-gated behavior contracts.
  4. Hybrid retriever (.agent/retriever/) — fastembed + sqlite-vec + BM25 with context-pack assembly.
  5. Skill graph — typed requires/provides frontmatter and a resolver/typed call.
  6. Cross-project federation (~/.agentic-stack/global/) — promote lessons accepted in 3+ projects.
  7. Speculative execution (.agent/spec/) — N approaches in parallel via git worktrees, scored, merge winner.
  8. Background autonomous actions (.agent/act/) — sandboxed proposals workflow with human approval.

Design invariants: local-first, append-only, permissioned via existing protocols/permissions.md, cross-platform, backward compatible (every new feature is opt-in behind .features.json flags).

Includes new protocol schemas, adapter changes (one-line bus.py announce at session start), CLI surface, storage/perf targets, migration plan from v0.18, 6-week phased rollout, risks, and a definition of done.

Not for merge as-is — this is the design surface to review, edit, and decompose into implementation PRs.

Note

Add v0.19 agentic runtime specification to spec.md

Adds spec.md defining the v0.19 "Agentic Turn" runtime design. The document covers the new .agent/ directory layout, core subsystems (plans, bus, evals, retriever, spec, act), CLI surfaces, adapter protocols, permissions model, storage/performance targets, migration path, rollout plan, and definition of done.

📊 Macroscope summarized 69aac89. 1 file reviewed, 2 issues evaluated, 0 issues filtered, 1 comment posted

🗂️ Filtered Issues

Comment thread spec.md
│ ├── messages.jsonl # append-only event log
│ ├── claims.jsonl # who owns what subgoal
│ ├── locks/<resource>.lock # advisory file locks
│ └── inbox/<agent_id>/ # per-agent unread pointer
Copy link
Copy Markdown

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

🟢 Low spec.md:45

The .agent/bus/inbox/<agent_id>/ directory is listed in the layout (line 45) as a "per-agent unread pointer," but section 2.2's bus subsystem spec never defines what files belong in this directory, how unread pointers are structured, read, or written, or which CLI commands interact with them. This leaves an unimplemented component in the specified directory layout that implementers cannot build from the spec.

Consider either adding inbox semantics to section 2.2 (file format, read/write protocol, CLI commands) or removing inbox/<agent_id>/ from the directory layout if it's deferred to a future release.

🚀 Reply "fix it for me" or copy this AI Prompt for your agent:
In file spec.md around line 45:

The `.agent/bus/inbox/<agent_id>/` directory is listed in the layout (line 45) as a "per-agent unread pointer," but section 2.2's bus subsystem spec never defines what files belong in this directory, how unread pointers are structured, read, or written, or which CLI commands interact with them. This leaves an unimplemented component in the specified directory layout that implementers cannot build from the spec.

Consider either adding inbox semantics to section 2.2 (file format, read/write protocol, CLI commands) or removing `inbox/<agent_id>/` from the directory layout if it's deferred to a future release.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant