Feature Map

Feature Map

This is the current main-branch feature inventory for R1.

Mission Runtime

Feature	Benefit	Status	Reference
Plan/execute/verify workflow	Keeps agent output tied to explicit execution and verification gates	Done	`app/`, `workflow/`, `verify/`
Adversarial review posture	Favors reviewable output over ungoverned autonomous edits	Done	`critic/`, `convergence/`, `engine/`
Evidence model	Persists what happened instead of trusting session memory	Done	`ledger/`, `bus/`, `session/`
Anti-truncation enforcement	Refuses end-turn while plan items unchecked or truncation phrases are emitted; layered machine-mechanical defense against LLM self-reduction.	Done	`internal/antitrunc/`, `internal/agentloop/antitrunc.go`, `internal/supervisor/rules/antitrunc/`, `cmd/r1/antitrunc_cmd.go`, `docs/ANTI-TRUNCATION.md`

Deterministic Skills

Feature	Benefit	Status	Reference
Deterministic skill manufacture	Turns reusable workflows into governed artifacts	Done	`internal/skillmfr/`
Registry and selection	Lets runtime behavior map to explicit skill assets	Done	`internal/skill/`, `internal/skillselect/`
Flagship deterministic runtimes	Shows concrete packaged uses of the deterministic runtime path	Done	April 30 main-branch skill runtime commits

Skill Pack Lifecycle

Feature	Benefit	Status	Reference
`r1 skills pack init`	Creates repo-local packs without hand-building directories	Done	`cmd/r1/skillspackcmd.go`
`info`, `install`, `list`, `publish`, `search`, `update`	Makes pack inspection, activation, discovery, and refresh operational	Done	`cmd/r1/skillspackcmd.go`
`sign` and `verify`	Adds integrity controls to pack distribution	Done	`cmd/r1/skillspackcmd.go`
`serve` HTTP registry	Exposes published packs through a stable read-only registry	Done	`cmd/r1/skillspackserver.go`
Runtime signed-pack verification	Prevents runtime registration from ignoring pack integrity	Done	April 30 main-branch signed-pack verification commit

Runtime Helper Surfaces

Feature	Benefit	Status	Reference
Ledger audit runtime	Lets deterministic flows query ledger-backed audit evidence	Done	April 30 main-branch ledger audit runtime commit
Skill execution audit runtime	Makes runtime execution behavior inspectable	Done	April 30 main-branch execution audit runtime commit
Metrics collection runtime	Exposes runtime metrics snapshots to deterministic flows	Done	`cmd/r1-mcp/metrics_runtime.go`
Timeout and cancellation hooks	Keeps deterministic runtime calls bounded and cancellation-aware	Done	`cmd/r1-mcp/backends.go`
Oneshot runtime cost metadata	Makes runtime cost visible to callers and operators	Done	April 30 main-branch oneshot cost metadata commit

Agentic Test Harness

Feature	Benefit	Status	Reference
`r1.*` MCP catalog (38 tools across 10 categories)	One namespace; every UI action reachable through MCP	Done (catalog only; back-end pending specs 1-7)	`internal/mcp/r1servercatalog.go`
Slack-style envelope + stokerr/ taxonomy	Predictable wire shape; no raw Go errors leak	Done	`internal/mcp/envelope.go`, `internal/mcp/stokerr_map.go`
`r1 mcp serve --print-tools [--markdown]`	Lint + docs generator have a stable input	Done	`cmd/r1/mcp.go`
`internal/tui/teatest_shim.go`	Bubble Tea drivable through MCP without a terminal emulator	Done (in-process driver; teatest swap pending dep)	`internal/tui/teatest_shim.go`
`A11yEmitter` + JSONPath evaluator	Synthetic a11y trees + structural assertions	Done	`internal/tui/a11y.go`, `internal/tui/jsonpath.go`
`*.agent.feature.md` parser + dispatcher	Gherkin-shaped tests dispatched to MCP catalog	Done	`tools/agent-feature-runner/`
8 seed feature fixtures across all 10 categories	Coverage gate per spec 8 §10	Done	`tests/agent/{tui,web,cli,mission,worktree}/`
`lint-view-without-api` + allowlist	UI without API is a build break	Done (Go scanner active; React + Tauri scanners blocked on specs 6/7 merge)	`tools/lint-view-without-api/`
`make agent-features[-update,-drift-check]`, `make lint-views`, `make docs-agentic`, `make storybook-mcp-validate`	One-line CI/local recipes	Done	`Makefile`
`docs/AGENTIC-API.md` + D-A1..D-A5 acceptance	External-agent contract + decisions log	Done	`docs/AGENTIC-API.md`, `docs/decisions/index.md`

Status

Done

governed mission runtime
deterministic skill substrate
full pack lifecycle including signing, verification, and HTTP serving
runtime metrics/audit/timeout/cancel/cost helper surfaces
agentic test harness wire surface (38 r1.* tools, parser/dispatcher,

TUI shim, lint scanner, 8 seed fixtures, AGENTIC-API.md, D-A1..D-A5)

In Progress

broader runtime-wide adoption of deterministic skills
agentic test harness back-end wiring (depends on specs 1-7 merging

the cortex/lanes/TUI/r1d/web/desktop sources)

Scoped

stronger publishing and release-check loops for pack distribution

Scoping

more outward-facing superiority reporting and proof surfaces

Potential-On Horizon

cross-product distribution and exchange of governed deterministic

skills

Pages in this directory