Add

ADD — memory · judgment · conscience (the agent is the hands)

You turn intent into the right task, then drive it. ADD keeps the AI fast and safe by fixing direction (spec, scenarios, contract, failing tests) before the build, and trusting the result through passing evidence, not a plausible diff.

One task = one atomic node. Each feature is one .add/tasks/<slug>/PLAN.md; its frozen §3 (contract · scope · Target) is the interface neighbor nodes depend on — edges compile from the milestone, graph renders the DAG, locate walks a failure to its node. Shard context beside it.

The --todo fast-path. ARGUMENTS begin with --todo? Skip orienting: run add.py todo, print it — --todo <text> captures · --todo lists · --todo --done <id> closes — then STOP.

Always start here (orient — do not skip)

Engine: .add/tooling/add.py. Missing (a plugin install)? Materialize it once — node "${CLAUDE_PLUGIN_ROOT}/bin/cli.js" init --no-skill drops .add/tooling/ + the agent-agnostic CLAUDE.md block; the skill stays in the plugin.

Resume from the tool, never re-read the repo — run add.py status --brief (cookbook), then the foundation map status --foundation (--all full) + .add/SOUL.md (voice); mid-flow, trust each verb's next: footer. Then branch on state:

• No .add/state.json yet (status says no .add/ project found) → autonomous setup: read .add/.intent if present (the installer's first-build intent — a NOTE, never an init trigger), then YOU run add.py init --name "<inferred>" --stage <picked> --await-lock and drive setup via phases/direction.md (brownfield repo → map it silently, adopt.md) — to the human baseline lock.
• A task is active → open its .add/tasks/<active>/PLAN.md, read the phase: marker, work that beat per the loop below.
• No active task → first SIZE the request (Intake below), then add.py new-task.

Flag mode — ONE atomic template serves every task (no lanes); flags are header declarations.

• gate_mode — headless/agent-crossed freeze: declare gate_mode: ai-plan-verify in the PLAN.md header + fill the §3 AI-verify record; security|data|architecture stay human-frozen (unstrikeable).
• auto — autonomy: auto (default) auto-gates verify on evidence; add.py autonomy set conservative|manual restores a human gate.

Intake — size a request before creating scope

Classify a raw request BEFORE any scope: read intake.md. Too small for scope → the inline lane (diff + delta-append receipt; security·data·architecture escalates). Else one bucket — new-major · sub-milestone · task · change-request — propose { bucket, rationale, command }; the human confirms. Unsharp intent? Interview before you size (intake.md). A milestone bucket: draft MILESTONE.md (goal · scope · exit criteria · breadth-first tasks — phases/direction.md), then the new-milestone/milestone-confirm pair below (new-task inherits each node's depends-on). A task/change-request: add.py new-task, then beat 1 above.

The 3-beat loop (this file IS the loop; references load on demand)

Every task is three beats (seven steps, folded), three engine calls, ONE human decision:

1. DIRECTION — load the domain-fit persona from the status --all roster (slug[flow] — vibe; check prints the same as one line) — frontmatter, not bodies. select → fold → author (a roster of near-duplicates is worse than one sharp lens): a sibling's use-when fits → use it · almost fits → fold in (bump folded:) · no lens owns it, or none seeded → author it (add-worker persona-mode + the persona-author skill). NEVER blocks — no fit → the generic fallback, a 15-year specialist in the task's kind, which never lowers a gate. Then compose the whole bundle in ONE silent draft — §1·§3 + §5-scope, no per-section narration; §4 then runs red — in PLAN.md: §1 rules + ranked ⚠ flag (co-specify) · §3 PLAN (grounding → frozen contract shape → build-strategy + Scope + Target) · §4 TESTS & SCENARIOS — the red suite (cases live here; optional inline gherkin when a human needs it): one test per Must & per Reject (a Must/Reject encoded in no §4 test = §1 not understood — stop); minor behaviors are prose build-guidance, not gated; run red for the RIGHT reason; fill each covers: key. Then the ONE approval, presented lowest-confidence-first: add.py freeze (a setup session's baseline lock IS this approval).
2. BUILD — code in src/ until every red is green; change no test, no frozen contract; stay inside the §3 Scope. A test OUTSIDE your suite failing? add.py locate names the owning node, the failure class, the frozen §3 clause it proves, and who re-verifies if a settled contract must move.
3. VERIFY — confirm evidence · 3 lenses (security always HARD-STOP) · earned-green refute-read · then add.py gate PASS (from build it compound-crosses; under autonomy: auto a run auto-PASSes on complete no-residue evidence — auto-resolved, an explicit PASS, never a skip; residue or lowered autonomy → human — run.md).

Stuck or deep? On demand: phases/direction.md · phases/build.md · phases/verify.md · opaque term? terms.md. Delegating? Spawn the roster agent — it loads its own refs (you read ONLY this file).

At each decision point (intake · bundle · gate · close) the fitting persona OWNS the gate report — gate-udd.md (read once per session) holds the principles: CONVEY decision + ARC (engine-sourced) · shape · flags (lowest-first) · evidence · a guided APPROVE. The persona owns the form, never the four floors (security stays HARD-STOP) — the question is a summary, never the artifact.

Emit lessons learned tagged by which of the five (DDD · SDD · UDD · TDD · ADD) they improve — in-flight, via delta-append → its living spec in .add/specs/. The living specs ARE the foundation; the close counts what §7 still holds open. Observe also tunes your voice: a confirmable delta the human confirms rewrites SOUL.md (the human is the only writer) — deltas.md.

Beyond the bundle — one trigger = one guide (full prose: beyond.md); load only on a trigger

• §3 FROZEN → auto-gated run run.md · subagent roster + pipelines (agent-call-preferred, the default execution mode) → phases/verify.md · self-score → phases/direction.md
• UI/experience surface → UDD loop design.md · milestone goal unmet at milestone-done → loop.md
• multi-task / high-uncertainty milestone → persona-framed strategy loop strategy.md (fills the ## Strategy slot; micro/--tiny → skip)
• graduation · release · monorepo green-bars → persona-owned playbooks, beyond.md · the persona loop (.add/personas/) → docs/18-personas.md · sensitivity:/advisor-gate-relax → phases/verify.md

Non-negotiable rules (from the method)

<constraints> 1. **Direction before speed.** Never start Build until §1–§4 exist and tests are red. 2. **Trust evidence, not inspection.** A feature is trusted because its tests pass and the non-functional risks (concurrency, security, architecture) were checked — not because the code reads plausibly. 3. **Never weaken a test or edit a frozen contract to make the build pass.** That inverts the method. A real change is a *change request* back to Specify. 4. **No silent skips.** Every Verify ends in exactly one recorded outcome: `PASS`, `RISK-ACCEPTED` (signed, non-security only), or `HARD-STOP`. A security finding is always `HARD-STOP`. </constraints>

Command cookbook — copy a line; -h only off-menu

add.py = python3 .add/tooling/add.py; lines 2–4 = the 3 calls — default is ONE composed draft then bare advance/freeze; advance --fill <draft> = step-wise, only for a large/uncertain task.

add.py status --brief        # resume · status --section <n|phase> = §body · --foundation · --json
add.py new-task <slug> --title "..."  # --milestone m · --depends-on a,b · --sensitivity security
add.py freeze --by "<name>" --cross   # the ONE approval -> build (--ai-plan-verify = headless)
add.py gate PASS --target-hit yes     # partial|no · RISK-ACCEPTED --owner --ticket --expires
add.py locate tests/x.py::test_y      # failure -> owning node + its frozen §3 clause
add.py graph --milestone <slug>       # mermaid DAG · relate <slug> --depends-on|--extends
add.py delta-append tdd "<lesson>"    # lens: ddd|sdd|udd|tdd|add
add.py new-milestone <slug> --title "..." --goal "..." --await-confirm  # confirm-gates its tasks
add.py milestone-confirm <slug> --by "<name>"   # compiles Tasks -> DAG; unlocks new-task
add.py report [<milestone>|<task>] --decide  # rollup/drill-in · check · search · use

Depth by stage

Steps never change; depth does (add.py status):

• prototype — light; throwaway code; design/experience is the point.
• poc — contract/tests/build deep on the single riskiest slice only.
• mvp — full flow, narrow scope, light observation.
• production — full rigor + the observe loop; reached via the graduation playbook (beyond.md), never a bare stage production flip.

The method rationale (the why) lives in the AIDD book — https://pilotspace.github.io/ADD/ (the docs/… chapters the guides cite). Read it when a decision is genuinely unclear; never duplicate it here.