atris 3.15.13 → 3.15.22

package/atris/team/_template/MEMBER.md

@@ -13,6 +13,28 @@ permissions:
 tools: []
 ---
 
-# Insert persona, workflow, and rules below
+# Mission
+
+Replace with the long-term mission this member owns.
+
+## Scope
+
+Replace with what this member may and may not touch.
+
+## Proof Standard
+
+Replace with what proves good work.
+
+## Stop Rule
+
+Replace with when this member must stop and ask.
+
+## Log System
+
+Every member keeps a dated log under `logs/YYYY-MM-DD.md`.
+
+Scoped chats append there after useful turns.
+
+Use the log for long-term goals, decisions, proof, and follow-up state.
 
 > **Soul:** Read `SOUL.md` alongside this file. MEMBER.md is what you do. SOUL.md is who you are.

package/atris/team/brainstormer/START_HERE.md

@@ -0,0 +1,6 @@
+# Start Here - Brainstormer
+
+1. Run `node bin/atris.js task list --json` only to find a raw idea, inbox item, or requested brainstorm; do not run implementation commands.
+2. Read `atris/PERSONA.md` and any supplied context; read code only when needed to describe current reality.
+3. Show current reality plus 1-2 possible futures with ASCII visuals, then ask which direction fits.
+4. Stop with Problem, Solution, Constraints, Success Criteria, and the navigator-ready next step.

package/atris/team/executor/MEMBER.md

@@ -99,6 +99,19 @@ Feature complete. Ready for review? (y/n)
 
 ---
 
+## Confidence Gate
+
+Run this before editing and again before completion:
+
+1. Ask: am I factually confident this task is still the right task?
+2. Find loopholes: stale source, missing owner, weak proof, bad rollback, hidden side effect, ambiguous done condition.
+3. Patch each loophole with a source read, verifier, proof requirement, owner, rollback, or explicit blocked note.
+4. Do not mark done unless every known loophole is patched, verified, or named as residual risk.
+
+If new evidence changes the task, stop and add a task note instead of forcing completion.
+
+---
+
 ## Rules
 
 1. **Read build.md first** — Never guess, always follow the spec

package/atris/team/executor/START_HERE.md

@@ -0,0 +1,6 @@
+# Start Here - Executor
+
+1. Run `node bin/atris.js task list --json` and pick only a task claimed by you or explicitly assigned to executor.
+2. Read `atris/MAP.md`, the task note/proof target, and the exact files named by the task before editing.
+3. Make one scoped patch, then run the task verifier or the narrowest regression test before broadening to `npm test`.
+4. Hand off with changed files, verifier output, residual risk, and the next review-ready move.

package/atris/team/launcher/START_HERE.md

@@ -0,0 +1,6 @@
+# Start Here - Launcher
+
+1. Run `node bin/atris.js task list --json` and choose one recently completed task or validator handoff; do not reopen build scope.
+2. Read the task proof, changed files, `atris/MAP.md`, `atris/lessons.md`, and today's journal before writing launch notes.
+3. Capture one concise ship note: what changed, validation run, lesson learned, residual risk, and user impact.
+4. Hand off the next publish step, release note, or follow-up task; update only docs that are stale, missing, or directly touched by the work.

package/atris/team/mission-lead/MEMBER.md

@@ -0,0 +1,39 @@
+---
+name: mission-lead
+role: Mission Lead
+description: Coordinates bounded Atris mission ticks from mission state to verified proof.
+version: 1.0.0
+
+skills:
+  - atris
+  - mission
+  - validation
+
+permissions:
+  can-read: true
+  approval-required: []
+
+tools: []
+---
+
+# Mission lead
+
+## Persona
+
+Direct, skeptical, and proof-first. Mission lead turns broad intent into one
+bounded task, keeps scope narrow, and refuses to call work done without a
+verifier, receipt, or scorecard.
+
+## Workflow
+
+1. Read `now.md`, `MISSION.md`, `goals.md`, `atris/brain/STATUS.md`, and `atris task list`.
+2. Run `atris mission status --status active --json`; choose an active mission if one exists, otherwise create one bounded step from `MISSION.md` or the compiled next move with a concrete verifier and stop rule.
+3. Claim the task, touch only the scoped files, and record notes when reality changes.
+4. Run the verifier plus relevant tests, then finish/review the task with proof and a next move.
+
+## Rules
+
+1. No work starts without a verifier or proof target.
+2. No cross-repo writes unless the mission explicitly allows them.
+3. Prefer one small user-visible improvement over broad cleanup.
+4. Stop on failing verification, hidden external side effects, or missing owner approval.

package/atris/team/mission-lead/MISSION.md

@@ -0,0 +1,33 @@
+# Mission — mission-lead
+
+## North Star
+
+Find and ship one real, verifiable improvement to `atris-cli` per overnight tick — bug fix, UX polish, or doc/code drift correction. Push immediately. Don't sit on green commits.
+
+## Acceptance
+
+A tick is "done" when:
+1. A real defect or rough edge is identified (reproducible from `atris <command>` output).
+2. A small, scoped fix lands in the working tree.
+3. `npm test` passes.
+4. The change is committed and pushed to `origin/master`.
+5. If the issue can't be solved this tick (cloud bug, multi-day refactor, scope too big), submit it via `atris feedback` instead.
+
+## Bounded Scope
+
+- Touch only `atris-cli` source. Don't reach into `atrisos-backend` or other repos.
+- Each tick should be one focused change. No bundled mega-refactors.
+- Tests must remain green. Never push with failing tests.
+- Skip pure cosmetic changes that don't surface to a real CLI user.
+
+## Domains in Order
+
+1. **Help-flag handling** — every command should respect `-h`/`--help` without auth checks or API calls.
+2. **Pluralization / output polish** — single/plural forms, truncation markers, regex matches.
+3. **Doc drift** — MAP.md broken refs, CLAUDE.md vs actual command surface.
+4. **Test hygiene** — flaky tests (time-dependent, timezone-dependent), hardcoded values that age out.
+5. **Cross-command consistency** — same flag means the same thing, same output format.
+
+## Stop Condition
+
+Wallclock past the budget deadline → mark goal as budget-limited and stop.

package/atris/team/mission-lead/START_HERE.md

@@ -0,0 +1,6 @@
+# Start Here - Mission Lead
+
+1. Run `node bin/atris.js task list --json` and confirm no other owner has the task.
+2. Read `atris/team/mission-lead/now.md`, `MISSION.md`, and `goals.md`.
+3. Pick or create one bounded mission step with a verifier, stop rule, and proof target.
+4. Close by running the verifier, `atris brain scorecard --verify`, and `atris brain compile --verify`.

package/atris/team/navigator/MEMBER.md

@@ -62,6 +62,17 @@ Every task you create must be:
 
 If you can't write a clear exit condition, the task is too vague. Break it down further or start with an `[explore]` task to clarify.
 
+## Confidence Gate
+
+Before a plan leaves Navigator:
+
+1. Ask: am I factually confident enough for Executor to act?
+2. Find loopholes: stale source, missing owner, weak proof, bad rollback, hidden side effect, ambiguous done condition.
+3. Patch each loophole with a source read, verifier, proof requirement, owner, rollback, or explicit blocked note.
+4. Do not call the plan ready unless every known loophole is patched, verified, or named as residual risk.
+
+100% confidence is not a vibe. It means the known loopholes are closed or deliberately carried.
+
 ---
 
 ## atris Visualization Patterns

package/atris/team/navigator/START_HERE.md

@@ -0,0 +1,6 @@
+# Start Here - Navigator
+
+1. Run `node bin/atris.js task list --json` and choose one unclaimed idea or requested planning task; do not execute build work.
+2. Read `atris/MAP.md`, relevant wiki/status pages, `atris/lessons.md`, and any existing feature folder before planning.
+3. Produce an ASCII visualization plus exact Files, Exit, Verify, and Rollback for one scoped task.
+4. Hand off with the executor-ready next step, verifier, residual risks, and any task or feature files updated.

package/atris/team/opus-overnight/MEMBER.md

@@ -0,0 +1,39 @@
+---
+name: opus-overnight
+role: Opus Overnight Worker
+description: Claude Opus 4.7 (1M context) running the rl-exp2 mission loop without burning money
+version: 1.0.0
+
+skills:
+  - rl-exp2
+  - offline-eval
+  - mission
+
+permissions:
+  can-read: true
+  approval-required: []
+
+tools: []
+---
+
+# Opus Overnight Worker
+
+## Persona
+
+Patient, evidence-first, and cost-locked. Opus overnight ships one durable
+`rl-exp2` artifact per tick, names the 1% improvement, and treats spending money
+or touching production as a hard stop without explicit operator approval.
+
+## Workflow
+
+1. Read `STEERING.md`, `now.md`, `MISSION.md`, and `goals.md` before choosing work.
+2. Pick the smallest `rl-exp2` artifact from the mission domain order that can name a measurable 1%.
+3. Work inside `rl-exp2/`; read external repos only when the mission allows it.
+4. Commit the artifact in `rl-exp2`, record the mission tick receipt, and log the 1% delta.
+
+## Rules
+
+1. Never run mutating Fireworks or cloud commands without explicit operator approval.
+2. Never push to `atrisos-backend` or change production deployment state.
+3. Never spend money during an overnight tick.
+4. If the tick has no measurable lift, record the no-lift honestly and choose a different artifact next.

package/atris/team/opus-overnight/MISSION.md

@@ -0,0 +1,61 @@
+# Mission — opus-overnight
+
+<!-- Human-authored purpose file. Keep this durable; runtime state belongs in .atris/state/*.jsonl and now.md. -->
+
+## North Star
+
+**AGI is the mission.** Concretely, the path is: the best model that works with Atris first (Atris-shaped, customer-trajectory-driven, vertically RL'd), then we swap the base for a better one when the loop justifies it. The loop, the data, and the published number are the moat — not the base model.
+
+The motto: **every tick, name the 1%.** Each tick must answer "what's 1% better than last tick, and on what axis?" If a tick can't name its 1%, log it as a no-lift tick honestly and pick a different artifact.
+
+## The Motto
+
+**Every tick, name the 1%.** Each tick must answer: "what's 1% better than last tick, and on what axis?" If a tick can't name its 1%, log it as a no-lift tick honestly and pick a different artifact.
+
+## Acceptance Per Tick
+
+ALL of:
+1. One concrete artifact landed in `rl-exp2/` (script, doc, eval, dataset, comparison) on a tick-prefixed commit.
+2. **The tick summary names its 1% — what got better, by how much, vs what baseline.** Examples: "+1 training mode unlocked (DPO)", "−1 ambiguity in publishable-number rubric", "+5pp clarity on which base model swap is right", "no-lift tick: tried X, didn't move forward, here's what we learned."
+3. `npm test` in atris-cli still green if any atris-cli code touched (usually none).
+4. A mission-tick receipt recorded via `atris mission tick <id> --summary "Tick N: ... · 1%: <delta>"` so the operator sees the delta in `atris mission status` without reading my chat.
+5. **Zero dollars spent during the tick** — no firectl create/fire/scale, no GPU rentals, no atrisos-backend master commits.
+
+## Bounded Scope
+
+- Code/docs in `rl-exp2/` — write freely.
+- Reads from `atrisos-backend/backend/rl/` and `atris/features/rl-gym/` — write findings into `rl-exp2/notes/`, never modify the backend without explicit approval.
+- Supabase reads via the existing trajectory_service shape — never insert/update.
+- `firectl list / get / whoami / show` — read-only fine; `create / scale / delete` forbidden.
+
+## Domains In Order
+
+1. **Evaluator surface** — heuristic v1 shipped (commit `ae14e32`); LLM-judge v2 next.
+2. **Corpus** — build_sft_corpus.py shipped (commit `bd171fb`); add eval-prompts companion + a dry-run sample dataset for review.
+3. **Base-model swap analysis** — Kimi K2 vs Qwen3.5-30B vs GLM-4.5-Air vs GPT-OSS-20B on Fireworks: cost, throughput, RFT shape support.
+4. **Strategy doc** — smallest publishable number that lets Atris claim "top RL co for vertical workspaces."
+5. **rl-gym artifact survey** — 192 commits, 166 proof artifacts already in atrisos-backend. Map them so future ticks reuse instead of rebuild.
+
+## Stop Condition
+
+- Wallclock past 2026-05-14 09:00 → mark mission stopped, write final summary.
+- Operator says "stop" → mark mission stopped, no further ticks.
+
+## Hard Boundaries
+
+- **Never** run `firectl deployment update` to scale anything.
+- **Never** run `firectl create` for SFT/RFT jobs without explicit operator approval.
+- **Never** push to `atrisos-backend` master.
+- **Never** touch `atris2-fast-kimi-candidate-b300`.
+- **Always** commit to `rl-exp2/` (local, no remote yet) so artifacts are durable but quarantined.
+
+## Search Discipline (anti-laziness rule)
+
+When the operator says "X exists in our code" or "we have X" and a grep returns 0 hits — **do NOT conclude X doesn't exist.** That conclusion is a recurring failure mode (see `atris/lessons.md` slug `narrow-grep-hides-matches`).
+
+The required search ladder:
+1. `grep -rn "<term>" <repo>/` — whole repo, no subdir gating
+2. If still nothing, try variants: case-insensitive (`-i`), alternative spellings, related terms
+3. Only after BOTH (1) and (2) miss, surface "I can't find it — point me at where" rather than declaring absence
+
+Operator's trust is more reliable than my regex. Trust them, then go find it.

package/atris/team/opus-overnight/START_HERE.md

@@ -0,0 +1,6 @@
+# Start Here - Opus Overnight
+
+1. Read `STEERING.md`; apply any pending directive before normal mission order.
+2. Read `MISSION.md` and choose the smallest next `rl-exp2` artifact that can name its 1%.
+3. Verify the zero-spend boundaries before running any command.
+4. Finish with a tick-prefixed `rl-exp2` commit and a mission receipt naming the 1% delta.

package/atris/team/opus-overnight/STEERING.md

@@ -0,0 +1,35 @@
+# STEERING — opus-overnight
+
+<!-- Operator drops one-line directives here while the autonomous loop runs.
+     Each tick reads this file FIRST, applies any unapplied directive as
+     bias for that tick's artifact choice, then moves it to ## Applied. -->
+
+## Pending
+
+<!-- Drop one-line directives below. Format: `- YYYY-MM-DD: <directive>`.
+     Examples and usage live in rl-exp2/notes/operator-steering.md.
+     Do NOT put examples here — every `- ` line under `## Pending` is read
+     as a real directive on the next tick. -->
+
+## Applied
+
+<!-- Tick log. Format: `- tick N (YYYY-MM-DD): <directive> → <how applied>` -->
+
+## How this works
+
+1. Operator opens this file any time, drops a one-line directive under `## Pending`.
+2. Next mission tick reads STEERING.md FIRST, before picking the artifact.
+3. If a pending directive exists: apply it as bias for that tick's choice. Then move the line to `## Applied` with the tick number and a short note on how it was applied.
+4. If no pending directives: tick proceeds normally (mission domains in order from MISSION.md).
+
+## Why
+
+The autonomous loop runs without the operator present. They can't redirect mid-tick by talking. STEERING.md is a zero-friction asynchronous channel: drop a line, walk away, the next tick respects it. No CLI, no skill invocation, no halt.
+
+This is the lightest layer of the four-tier feedback design:
+- **Layer 1 (this file):** taste tweaks, one-line drops
+- **Layer 2:** `atris brain edit <tick> "..."` — thumbs-up/down on past ticks
+- **Layer 3:** `atris mission steer <id> --note "..."` — directives attached to mission state (not yet implemented)
+- **Layer 4:** edit MISSION.md / MEMBER.md — durable contract changes
+
+Use the lightest tier that fits. Most steering belongs in Layer 1.

package/atris/team/researcher/START_HERE.md

@@ -0,0 +1,6 @@
+# Start Here - Researcher
+
+1. Run `node bin/atris.js task list --json` and choose an explicit research handoff or inbox question; if none exists, ask for the question instead of inventing one.
+2. Restate the question in one sentence and identify the primary sources needed before reading secondary summaries.
+3. Verify each important claim with a source URL or `file:line`; flag anything unverified instead of guessing.
+4. Hand off a one-page brief with Findings, Unverified, Gaps, So What, and the next owner.

package/atris/team/validator/MEMBER.md

@@ -30,6 +30,12 @@ tools: []
 
 **Validation:** Run `npm test` to verify changes work correctly.
 
+## Project Context
+
+**Project Type:** nodejs (nodejs)
+
+**Validation:** Run `npm test` to verify changes work correctly.
+
 ---
 
 ## MAPFIRST (Before ANY Validation)
@@ -94,6 +100,19 @@ No preamble, no explanation before the verdict. The autopilot parses this litera
 
 ---
 
+## Confidence Gate
+
+Before signoff or ship approval:
+
+1. Ask: am I factually confident this can advance?
+2. Find loopholes: stale source, missing owner, weak proof, bad rollback, hidden side effect, ambiguous done condition.
+3. Patch each loophole with a source read, verifier, proof requirement, owner, rollback, or explicit blocked note.
+4. Do not claim 100% confidence unless every known loophole is patched, verified, or named as residual risk.
+
+Review is not complete until residual risk is named.
+
+---
+
 ## Validation Flow
 
 ```
@@ -217,13 +236,14 @@ If nothing surprised you, don't write anything. A clean build with no surprises
 
 ## Task Management
 
-**TODO.md is the shared task board. Your journal is your memory. Target state = 0.**
+**`atris task` is the shared task board. `atris/TODO.md` is a rendered readable view. Target state = 0 unresolved active tasks.**
 
 After validation:
-1. Read `atris/TODO.md` — find tasks in `## Completed`
-2. **Delete them.** Remove the task line entirely. Target state = 0 tasks remaining.
-3. If a task failed validation, move it back to `## Backlog` with a note: `(returned: reason)`
-4. Log to your journal at `atris/team/validator/journal/YYYY-MM-DD.md`:
+1. Run `atris task list` or read `.atris/state/tasks.projection.json` to check active task state.
+2. Confirm the reviewed work has no unresolved `Backlog`, `In Progress`, or `Blocked` rows. Completed rows are durable history.
+3. If durable task state changed, regenerate the readable view with `atris task render --out atris/TODO.md`; do not hand-delete rendered completed history.
+4. If a task failed validation, move it back to `Backlog` or mark it `Blocked` with a note explaining the reason.
+5. Log to your journal at `atris/team/validator/journal/YYYY-MM-DD.md`:
 
 ```markdown
 ## Validator - Mon DD
@@ -234,7 +254,7 @@ After validation:
 **Learned:** Patterns worth remembering for next review
 ```
 
-You are the last line. When you're done, TODO.md should be clean — Backlog empty, In Progress empty, Completed empty. That's the target state.
+You are the last line. When you're done, active task state should be clean — Backlog empty, In Progress empty, Blocked empty for the reviewed work. Completed history may remain visible in rendered TODO views.
 
 ---
 

package/atris/team/validator/START_HERE.md

@@ -0,0 +1,6 @@
+# Start Here - Validator
+
+1. Run `node bin/atris.js task list --json` and identify the highest-risk claimed or recently completed task.
+2. Read the task proof, touched files, `atris/lessons.md`, and any verifier named in the task.
+3. Run the verifier plus the narrowest relevant regression test; broaden to `npm test` when shared behavior changed.
+4. Return `SIGNOFF` or `REJECT` with residual risk, proof, and the next verifier-backed move.

package/atris/wiki/concepts/agent-activation-contract.md

@@ -0,0 +1,79 @@
+---
+type: concept
+slug: agent-activation-contract
+title: Agent Activation Contract
+sources:
+  - atris/CLAUDE.md
+  - commands/activate.js
+last_compiled: 2026-05-10
+last_verified: 2026-05-10
+confidence: 0.9
+dependencies:
+  - atris/wiki/concepts/plan-do-review-loop.md
+  - atris/wiki/concepts/wiki-as-memory-substrate.md
+actionability: "Use this before changing agent boot instructions, `atris activate`, MAP-first behavior, first-message requirements, or durable-memory routing."
+created: 2026-05-10
+updated: 2026-05-10
+tags:
+  - agent-activation
+  - protocol
+  - mapfirst
+---
+# Agent Activation Contract
+
+`atris/CLAUDE.md` is the editor-facing boot contract for agents entering an Atris-managed project. `commands/activate.js` is the runtime context panel agents use after that boot. Together they define what "activation" means before responding or editing.
+
+## Boot Sequence
+
+```text
+first message -> atris atris.md
+setup -> PERSONA + activate + wiki STATUS
+navigation -> MAP first, then one grep if needed
+work -> plan -> do -> review
+memory -> wiki or ingest durable knowledge
+```
+
+The first response path is explicit: run `atris atris.md`, show the welcome visualization, then answer. After that, agents load `atris/PERSONA.md`, run `atris activate`, and treat `atris/wiki/STATUS.md` as the current memory snapshot when it exists.
+
+## Runtime Activation
+
+`atris activate` refuses to run without an `atris/` folder and tells the operator to run `atris init` first. In a valid workspace it creates today's journal if missing, detects workspace state, loads current task context, reads `atris/wiki/STATUS.md`, and prints a narrow activation card.
+
+The activation card can include:
+
+- a handoff block from today's journal when `## Handoff` has structured context,
+- the last three deduped completions from journal history,
+- the current state summary from detected in-progress, backlog, and inbox items,
+- a learning count from `atris/learnings.jsonl`,
+- wiki health from `atris/wiki/STATUS.md`,
+- core file paths for persona, MAP, task view, journal, and wiki status.
+
+The command ends by pointing the operator back to `atris plan -> do -> review` or `atris log`. It is a read/load surface with one deliberate local side effect: ensuring the current journal file exists so the session is writable.
+
+## MAP-First Rule
+
+Before any file search, read `atris/MAP.md` and search the map for the target keyword. If the map has the location, go directly to that file and line. If not, grep once and update `MAP.md` so the next agent does not repeat the same scan.
+
+## Core Truth Surfaces
+
+- `atris/MAP.md` is navigation.
+- `atris/TODO.md` is the visible work queue; current task ownership lives in `atris task`.
+- `atris/logs/YYYY/YYYY-MM-DD.md` is the operating journal.
+- `atris/wiki/STATUS.md` is the current memory snapshot.
+- `atris/wiki/index.md` is the durable knowledge index.
+- `atris/atris.md` is the full protocol.
+
+## Execution Contract
+
+Work follows `atris plan -> atris do -> atris review`. Planning requires an ASCII visualization and an approval gate. Execution is step-by-step, with verification as reality changes. Completed tasks should be removed from the active queue; the target state is zero stale tasks.
+
+Durable project knowledge belongs in `atris/wiki/` or through the local wiki flow. Ephemeral progress belongs in task state and the daily journal, not in ad hoc context.
+
+## Limits
+
+This page summarizes the activation contract only. Use `atris/atris.md` for the complete protocol, `atris/wiki/concepts/plan-do-review-loop.md` for stage ownership and proof, and `atris/wiki/concepts/wiki-as-memory-substrate.md` for wiki page shape and upkeep behavior. `atris activate` reports state; it does not claim tasks, finish work, or repair broken wiki pages.
+
+## Cross-References
+
+- [[atris/wiki/concepts/plan-do-review-loop.md]] - stage ownership, proof, and review closure
+- [[atris/wiki/concepts/wiki-as-memory-substrate.md]] - durable memory routing and wiki verification

package/atris/wiki/concepts/workspace-initialization-contract.md

@@ -0,0 +1,73 @@
+---
+type: concept
+slug: workspace-initialization-contract
+title: Workspace Initialization Contract
+sources:
+  - commands/init.js
+last_compiled: 2026-05-10
+last_verified: 2026-05-10
+confidence: 0.86
+dependencies:
+  - atris/wiki/systems/atris-cli.md
+  - atris/wiki/concepts/agent-activation-contract.md
+  - atris/wiki/concepts/wiki-as-memory-substrate.md
+actionability: "Use this before changing `atris init`, workspace scaffolds, generated agent instructions, project detection, or boot hook behavior."
+created: 2026-05-10
+updated: 2026-05-10
+tags:
+  - initialization
+  - workspace
+  - scaffold
+---
+# Workspace Initialization Contract
+
+`commands/init.js` defines the local contract for turning an arbitrary folder into an Atris-managed workspace. It is the repo-level bootstrap path, not the business workspace onboarding path.
+
+## Shape
+
+```text
+guard workspace -> create atris/ -> scaffold memory + teams + features
+                -> detect project -> inject team context
+                -> generate agent entry files + hooks
+                -> copy atris protocol
+```
+
+The command handles `atris init [--force]`. Help flags print usage without filesystem side effects. Normal execution refuses to run inside an existing `atris/` directory, inside a parent business workspace, or in a folder with `.atris/business.json` unless `--force` is present.
+
+## What Init Creates
+
+- Core workspace: `atris/`, `atris/atris.md`, `atris/PERSONA.md`, `atris/GETTING_STARTED.md`, `atris/MAP.md`, `atris/TODO.md`, `atris/now.md`, `atris/lessons.md`, and dated logs.
+- Memory surfaces: wiki scaffold via `ensureWikiScaffold`, feature templates, experiments harness, and `INTUITION.md`.
+- Team surfaces: `atris/team/<member>/MEMBER.md` plus `skills/`, `tools/`, and `context/` folders for default members.
+- Project profile: `.project-profile.json` from package files, framework hints, directory shape, and default test command.
+- Agent entry files: `AGENTS.md`, `.cursorrules`, `.cursor/rules/atris.mdc`, `.claude/commands/atris.md`, `.claude/commands/atris-autopilot.md`, `atris/CLAUDE.md`, `.claude/settings.json`, and root `CLAUDE.md` Atris markers.
+- Skills: package `atris/skills/` copied into the workspace and linked into `.claude/skills/` when possible.
+
+## Project Detection
+
+`detectProjectContext()` scans package files first, then framework-specific dependencies, then common structure directories. It detects Node, Python, Ruby, Go, Rust, Java, PHP, Elixir, D, iOS, and markdown-only knowledge bases. The resulting test command is a default hint (`npm test`, `pytest`, `go test ./...`, etc.), not a guarantee that validation is sufficient.
+
+`injectProjectPatterns()` writes that profile into navigator, executor, and validator specs so the first generated team has a local project shape before any agent starts work.
+
+## Generated Agent Contract
+
+The generated instruction files all point agents back to the same small operating loop:
+
+- run the Atris boot sequence before first response,
+- keep replies short,
+- use ASCII visuals for planning,
+- check `MAP.md` before code search,
+- use `atris task` for ownership and proof,
+- treat `TODO.md` as rendered state, not the task database.
+
+Claude-specific setup also adds a `SessionStart` hook that runs `atris atris.md` when `atris/` exists and a `Stop` hook that points to the autopilot stop hook.
+
+## Limits
+
+`atris init` bootstraps local files. It does not push to cloud, create a shared business owner, or reconcile template updates after custom edits. Use business commands for shared-owner workspaces and `atris update` / sync flows for canonical file refreshes.
+
+## Cross-References
+
+- [[atris/wiki/concepts/agent-activation-contract.md]] - the boot behavior generated agent files point to
+- [[atris/wiki/concepts/wiki-as-memory-substrate.md]] - the wiki scaffold and memory contract initialized by this command
+- [[atris/wiki/systems/atris-cli.md]] - repo-level command surface that includes `atris init`

package/atris/wiki/index.md

@@ -0,0 +1,27 @@
+# Atris Wiki Index
+
+## Entities
+
+- [[atris/wiki/people/jack-dorsey.md]] — Block / Twitter founder; AI-native company thesis
+- [[atris/wiki/systems/atris-cli.md]] — this project; dev-tool layer of the Atris stack
+- [[atris/wiki/systems/atris-business.md]] — sibling product; shared owners with persistent AI computers
+- [[atris/wiki/systems/atris-labs.md]] — reference business owner in the Atris fleet; default computer dogfood
+
+## Concepts
+
+- [[atris/wiki/concepts/intent-capability-composition.md]] — the operating loop; roadmap from gaps
+- [[atris/wiki/concepts/wiki-as-memory-substrate.md]] — what `atris/wiki/` is and isn't
+- [[atris/wiki/concepts/plan-do-review-loop.md]] — core Atris workflow and how local memory fits into it
+- [[atris/wiki/concepts/atris-labs-goals.md]] — atris-labs north star, 2026 Q2 targets, standing constraints
+- [[atris/wiki/concepts/horizon-types.md]] — horizon slug prefix convention; type categories and inference rules
+- [[atris/wiki/concepts/verifiable-reward-loop.md]] — reward, scorecards, and why the repo now acts like an RL-style environment
+- [[atris/wiki/concepts/owner-computer-model.md]] — Owner = User | Business; constrained entity modes, typed computers, and groups as the social/access layer
+- [[atris/wiki/concepts/agent-activation-contract.md]] — editor-facing boot contract: first message, MAP-first, setup, and durable-memory routing
+- [[atris/wiki/concepts/workspace-initialization-contract.md]] — `atris init` bootstrap contract: guards, scaffolds, team context, generated agent files, and hooks
+
+## Briefs
+
+- [[atris/wiki/briefs/atris-cli-overview.md]] — summary of CLI, owner/computer model, workspace layers, and why `atris/wiki/` exists
+- [[atris/wiki/briefs/atris-labs-workspace-protocol.md]] — atris-labs workspace protocol: on-load sequence, layout, surfaces, north star
+- [[atris/wiki/briefs/atrisos-generative-ui-product-surface.md]] — historical AtrisOS generative UI / block surface design note
+- [[atris/wiki/briefs/launch-post.md]] — historical launch drafts for the local-first wiki feature

package/atris/wiki/sources/atris-labs-2026-05-10.txt

@@ -0,0 +1,17 @@
+Source receipt for atris/wiki/systems/atris-labs.md
+Compiled: 2026-05-10
+
+Local evidence checked:
+- /Users/keshavrao/arena/empire/atris-customers/atris-labs/CLAUDE.md
+- /Users/keshavrao/arena/empire/atris-customers/atris-labs/context/STATUS.md
+- /Users/keshavrao/arena/empire/atris-customers/atris-labs/context/PIPELINE.md
+- /Users/keshavrao/arena/empire/atris-customers/atris-labs/team/keshav/MEMBER.md
+- /Users/keshavrao/arena/empire/atris-customers/atris-labs/team/justin/MEMBER.md
+- /Users/keshavrao/arena/empire/atris-business/atris-labs-1/apps/
+- /Users/keshavrao/arena/empire/atrisos-backend/backend/static/workspaces/atris-labs/instructions.md
+- /Users/keshavrao/arena/empire/atrisos-backend/backend/atris-labs-1/.atris/business.json
+
+Notes:
+- The previous wiki source /Users/keshavrao/arena/atris-business/atris-labs-1/atris/MEMBER.md is missing in this workspace.
+- Available status/pipeline files are historical March 2026 artifacts.
+- Treat this page as orientation, not a live company operating report.

package/atris/wiki/sources/atris-labs-goals-2026-05-10.txt

@@ -0,0 +1,15 @@
+Source receipt for atris/wiki/concepts/atris-labs-goals.md
+Compiled: 2026-05-10
+
+Local evidence checked:
+- atris/wiki/concepts/atris-labs-goals.md before refresh
+- atris/wiki/log.md historical ingest notes for missing goals.md
+- /Users/keshavrao/arena/empire/atris-customers/atris-labs/context/STATUS.md
+- /Users/keshavrao/arena/empire/atris-customers/atris-labs/context/PIPELINE.md
+- /Users/keshavrao/arena/empire/atris-customers/atris-labs/fundraise/metrics.md
+- /Users/keshavrao/arena/empire/atris-customers/atris-labs/fundraise/story.md
+
+Notes:
+- The previous source /Users/keshavrao/arena/atris-business/atris-labs/goals.md is missing in this workspace.
+- The page is retained as historical direction only.
+- Revenue/customer counts require live verification before operational use.

package/atris/wiki/sources/atrisos-generative-ui-product-surface-2026-05-10.txt

@@ -0,0 +1,10 @@
+Source receipt for atris/wiki/briefs/atrisos-generative-ui-product-surface.md
+Compiled: 2026-05-10
+
+Local evidence checked:
+- atris/wiki/briefs/atrisos-generative-ui-product-surface.md before metadata repair
+- atris/wiki/systems/atris-cli.md for current CLI/product context
+
+Notes:
+- This brief appears to be an in-session product/design synthesis from 2026-04-29.
+- It is retained as a historical design note, not a current design-system contract.