# Index of tessera

Every file in this repo with a one-line description taken from the file itself. Generated by `scripts/build_index.py`; run `make index` after adding files, never edit by hand.

## Top level (12)

| File | What it is |
|---|---|
| [.gitignore](.gitignore) | Secrets and private material never leave this machine |
| [.python-version](.python-version) |  |
| [AGENTS.md](AGENTS.md) | Tessera: instructions for every coding agent |
| [CHANGELOG.md](CHANGELOG.md) | Changelog |
| [CLAUDE.md](CLAUDE.md) | Worker execution direction |
| [LICENSE](LICENSE) | MIT License |
| [Makefile](Makefile) | Tessera: atomic work queue and atomic storage. |
| [README.md](README.md) | Tessera – the work and storage engine |
| [architecture.json](architecture.json) | tessera |
| [keywords.txt](keywords.txt) | Terms for KEYWORDS.md, one per line: "term" or "term: alias, alias". |
| [package.json](package.json) | tessera-worker-tests |
| [pyproject.toml](pyproject.toml) |  |

## .claude/ (4)

| File | What it is |
|---|---|
| [.claude/guard.json](.claude/guard.json) | JSON object: allow, deny_read, deny_write, deny_commands |
| [.claude/hooks/format.py](.claude/hooks/format.py) | Claude Code PostToolUse hook: format the file that was just edited. |
| [.claude/hooks/guard.py](.claude/hooks/guard.py) | Claude Code PreToolUse guard: enforces the hard rules in every permission mode. |
| [.claude/settings.json](.claude/settings.json) | JSON object: $schema, permissions, hooks |

## .github/ (2)

| File | What it is |
|---|---|
| [.github/workflows/ci.yml](.github/workflows/ci.yml) |  |
| [.github/workflows/deploy.yml](.github/workflows/deploy.yml) |  |

## briefs/ (20)

| File | What it is |
|---|---|
| [briefs/L11.json](briefs/L11.json) | L11: Making-of export: event log and a board snapshot each evening |
| [briefs/L13.json](briefs/L13.json) | L13: Run the remaining L tasks through Tessera as briefs |
| [briefs/L16.json](briefs/L16.json) | L16: Model-job lane: job endpoints and cancel-on-lapse against the Meta Model API |
| [briefs/L18.json](briefs/L18.json) | L18: Receipts on every admitted result, with the thin gate attached |
| [briefs/L19.json](briefs/L19.json) | L19: Read-only board at /board on the counts record |
| [briefs/L20.json](briefs/L20.json) | L20: Deploy to one Fly machine with a volume; keys in the secret store |
| [briefs/L21.json](briefs/L21.json) | L21: Soak and restart runs on the deployed service |
| [briefs/L23.json](briefs/L23.json) | L23: Move Shamash's run loop onto the service |
| [briefs/L25.json](briefs/L25.json) | L25: Resume effect-class gate: require the task's effect class and admin for write-repo (T22) |
| [briefs/L26.json](briefs/L26.json) | L26: Versioned receipt hash test vectors for cross-language verification |
| [briefs/L33-execution-recall.md](briefs/L33-execution-recall.md) | L33: Execution evidence and reusable solution candidates |
| [briefs/done/L12.json](briefs/done/L12.json) | L12: Split-keep-parent and cancel endpoints |
| [briefs/done/L14.json](briefs/done/L14.json) | L14: Counts, health and events endpoints |
| [briefs/done/L15.json](briefs/done/L15.json) | L15: Queue definitions endpoint and Shamash's five queues |
| [briefs/done/L17.json](briefs/done/L17.json) | L17: MCP door over Streamable HTTP with claim, heartbeat, submit and split |
| [briefs/done/L24.json](briefs/done/L24.json) | L24: Atomic storage endpoints: records by compare-and-set, batches, blobs |
| [briefs/done/L6.json](briefs/done/L6.json) | L6: HTTP service skeleton: FastAPI app, problem+json errors, health and counts |
| [briefs/done/L7.json](briefs/done/L7.json) | L7: Worker keys: hashed storage, scopes, 14-day expiry, a command-line issuer |
| [briefs/done/L8.json](briefs/done/L8.json) | L8: Claim, heartbeat and submit endpoints with fencing tokens and idempotency keys |
| [briefs/done/L9.json](briefs/done/L9.json) | L9: Reaper loop, backoff and dead letter in the running service |

## conformance-draft/ (8)

| File | What it is |
|---|---|
| [conformance-draft/SPEC.md](conformance-draft/SPEC.md) | Tessera specification, v0.2 (draft) |
| [conformance-draft/conftest.py](conformance-draft/conftest.py) | Fixtures for the Tessera conformance suite (SPEC.md §10). |
| [conformance-draft/receipt_vectors.json](conformance-draft/receipt_vectors.json) | JSON object: schema_version, spec_section, version, vectors |
| [conformance-draft/test_model_jobs.py](conformance-draft/test_model_jobs.py) | T12: model jobs (Meta Model API background responses) are tied to the lease that started them. |
| [conformance-draft/test_queue_semantics.py](conformance-draft/test_queue_semantics.py) | Queue semantics every backend must meet (SPEC §5). |
| [conformance-draft/test_service_surface.py](conformance-draft/test_service_surface.py) | Service-only requirements (SPEC §7): keys, POST-only writes, health over HTTP, the MCP door and secret hygien… |
| [conformance-draft/test_storage.py](conformance-draft/test_storage.py) | ST01 to ST04: atomic storage. |
| [conformance-draft/test_task_graph.py](conformance-draft/test_task_graph.py) | Task graphs every backend must meet (SPEC §5.2 and §5.6): dependencies (fan-in), children that run in waves,… |

## deploy/ (28)

| File | What it is |
|---|---|
| [deploy/README.md](deploy/README.md) | Deploy |
| [deploy/bootstrap-hosting.sh](deploy/bootstrap-hosting.sh) | One-time hosting setup for one service on Fly.io behind a Cloudflare Worker: the Fly app and its |
| [deploy/cloudflare/README.md](deploy/cloudflare/README.md) | Cloudflare: the Tessera cloud runtime and edge |
| [deploy/cloudflare/access_openapi.json](deploy/cloudflare/access_openapi.json) | JSON object: components, info, openapi, paths |
| [deploy/cloudflare/agent_start.js](deploy/cloudflare/agent_start.js) | Versioned onboarding instructions. |
| [deploy/cloudflare/claim_admission.js](deploy/cloudflare/claim_admission.js) | Ephemeral scheduling hints, never lease or authorization authority. |
| [deploy/cloudflare/cloud_access.js](deploy/cloudflare/cloud_access.js) | Persistent principal metadata, anonymous limits and redacted access receipts. |
| [deploy/cloudflare/cloud_access_routes.js](deploy/cloudflare/cloud_access_routes.js) |  |
| [deploy/cloudflare/cloud_buckets.js](deploy/cloudflare/cloud_buckets.js) | Persistent cloud blob & bucket storage engine + canonical JSON receipt hashing for Tessera Cloud. |
| [deploy/cloudflare/cloud_executions.js](deploy/cloudflare/cloud_executions.js) | Execution nouns are immutable result evidence, never generic CAS rows or an alternate queue. |
| [deploy/cloudflare/cloud_frontier_publication.js](deploy/cloudflare/cloud_frontier_publication.js) | Publication is checked against immutable, actor-bound queue results inside SQL transactions. |
| [deploy/cloudflare/cloud_intelligence.js](deploy/cloudflare/cloud_intelligence.js) | 0-call progressive L0 scent search, RFC 9110 HTTP 206 byte-range slicing, |
| [deploy/cloudflare/cloud_mcp.js](deploy/cloudflare/cloud_mcp.js) |  |
| [deploy/cloudflare/cloud_pages.js](deploy/cloudflare/cloud_pages.js) | Agent door & HTML pages for Tessera Cloud (/connect, /llms.txt, /llms-full.txt, |
| [deploy/cloudflare/cloud_publication.js](deploy/cloudflare/cloud_publication.js) | Completed queue evidence is checked in the same transaction as promotion. |
| [deploy/cloudflare/cloud_store.js](deploy/cloudflare/cloud_store.js) |  |
| [deploy/cloudflare/d1-buckets.sql](deploy/cloudflare/d1-buckets.sql) | Persistent Cloudflare D1 content-addressed blob & named-bucket store (ta-amps-buckets), Schema v3. |
| [deploy/cloudflare/d1-fabric.sql](deploy/cloudflare/d1-fabric.sql) | Persistent Cloudflare D1 schema for Tessera cloud fabric (ta-amps-fabric), Schema v3. |
| [deploy/cloudflare/execution_matching.js](deploy/cloudflare/execution_matching.js) | Bounded execution-report validation and retrieval candidates, never reuse authorization. |
| [deploy/cloudflare/execution_observations.js](deploy/cloudflare/execution_observations.js) | Execution observations and problem fingerprints: core infrastructure, no learning policy. |
| [deploy/cloudflare/ontology_catalog.js](deploy/cloudflare/ontology_catalog.js) | Generated by scripts/buildontologycatalog.py; do not edit catalog data by hand. |
| [deploy/cloudflare/ontology_engine.js](deploy/cloudflare/ontology_engine.js) |  |
| [deploy/cloudflare/storage_quota.js](deploy/cloudflare/storage_quota.js) | Provider capacity failures are availability errors, not client claim throttling. |
| [deploy/cloudflare/worker.js](deploy/cloudflare/worker.js) |  |
| [deploy/cloudflare/wrangler.jsonc](deploy/cloudflare/wrangler.jsonc) | Tessera Cloud Worker: 24/7 persistent atomic work queue, queryable CAS noun stores, |
| [deploy/fly/Dockerfile](deploy/fly/Dockerfile) |  |
| [deploy/fly/README.md](deploy/fly/README.md) | Fly.io: the Tessera service |
| [deploy/fly/fly.toml](deploy/fly/fly.toml) | Tessera service on one Fly.io machine with a volume (brief L20), reached through the Cloudflare edge |

## docs/ (30)

| File | What it is |
|---|---|
| [docs/WORKER-FIRST-EXECUTION.md](docs/WORKER-FIRST-EXECUTION.md) | Worker-first execution and private work |
| [docs/WORKER-VERB-EXAMPLES.md](docs/WORKER-VERB-EXAMPLES.md) | Work suitable for claimed verbs |
| [docs/access-modes.md](docs/access-modes.md) | External agent access |
| [docs/agent-start.md](docs/agent-start.md) | Start an agent from a website |
| [docs/architecture.md](docs/architecture.md) | Tessera: architecture |
| [docs/claim-admission.md](docs/claim-admission.md) | Claim admission |
| [docs/comparison.md](docs/comparison.md) | Tessera: comparison |
| [docs/continuous-tuning.md](docs/continuous-tuning.md) | Continuous tuning, learning and evaluation |
| [docs/decisions.md](docs/decisions.md) | Tessera: decisions |
| [docs/evidence/native-history-observation.json](docs/evidence/native-history-observation.json) | JSON object: schema_version, observed_date, source_revision, worker_version, harness, provider |
| [docs/execution-learning.md](docs/execution-learning.md) | Execution learning: observations, fingerprints and swappable learners |
| [docs/execution-memory.md](docs/execution-memory.md) | Execution reports and permitted recall |
| [docs/expansion.md](docs/expansion.md) | Seeding & expanding atomic nouns and verbs with Muse / Meta agents |
| [docs/fine-tuning-workers.md](docs/fine-tuning-workers.md) | Optional fine-tuning workers |
| [docs/frontier-publication.md](docs/frontier-publication.md) | Frontier publication receipts |
| [docs/generalizing.md](docs/generalizing.md) | Tessera: generalizing it |
| [docs/glossary.md](docs/glossary.md) | Tessera: glossary |
| [docs/handoff.md](docs/handoff.md) | Tessera: handoff |
| [docs/making-of.md](docs/making-of.md) | The making-of |
| [docs/opencode-and-ollama.md](docs/opencode-and-ollama.md) | Bilateral swarm testing with OpenCode & Ollama Cloud (muse-spark-open) |
| [docs/operations.md](docs/operations.md) | Operations |
| [docs/opportunity-lab.md](docs/opportunity-lab.md) | Opportunity Lab |
| [docs/raw-blob-governance.md](docs/raw-blob-governance.md) | Raw blob governance and the first uploader |
| [docs/research-seeds.md](docs/research-seeds.md) | Research seeds and proposed recurring work |
| [docs/review.md](docs/review.md) | Tessera: review guide |
| [docs/setup.md](docs/setup.md) | Tessera: setup |
| [docs/storage-availability.md](docs/storage-availability.md) | Storage availability and activity timestamps |
| [docs/summary.md](docs/summary.md) | Tessera: summary |
| [docs/work-design-and-evidence.md](docs/work-design-and-evidence.md) | Nouns, verbs and evidence |
| [docs/work-routing.md](docs/work-routing.md) | Route agent work through nouns and verbs |

## examples/ (73)

| File | What it is |
|---|---|
| [examples/continuous_tuning/__init__.py](examples/continuous_tuning/__init__.py) | Continuous tuning, evaluation and hard-example mining as Tessera nouns and verbs. |
| [examples/continuous_tuning/augment.py](examples/continuous_tuning/augment.py) | Schema-perturbation augmentation: teach reading the offered schema, not memorizing names. |
| [examples/continuous_tuning/backend_hf.py](examples/continuous_tuning/backend_hf.py) | Optional torch/Transformers/PEFT worker backend; never imported by Tessera's service. |
| [examples/continuous_tuning/catalog.py](examples/continuous_tuning/catalog.py) | Verb templates and the execution report every verb submits. |
| [examples/continuous_tuning/cli.py](examples/continuous_tuning/cli.py) | Command line surface: tick on demand, show status and the learning curve, build the pack. |
| [examples/continuous_tuning/dataset.py](examples/continuous_tuning/dataset.py) | Curation: fingerprints, deduplication, contamination checks and dataset manifests. |
| [examples/continuous_tuning/decide.py](examples/continuous_tuning/decide.py) | The promotion verb: apply the pre-registered rule and move the champion pointer. |
| [examples/continuous_tuning/evidence/llama-3.2-1b-instruct-2026-10-29.json](examples/continuous_tuning/evidence/llama-3.2-1b-instruct-2026-10-29.json) | JSON object: benchmark_tools_sha256, captured, code, curve, format, hardware |
| [examples/continuous_tuning/evidence/llama-3.2-3b-instruct-2026-10-29.json](examples/continuous_tuning/evidence/llama-3.2-3b-instruct-2026-10-29.json) | JSON object: benchmark_tools_sha256, captured, code, curve, format, frozen_sets_sha256 |
| [examples/continuous_tuning/fabric.py](examples/continuous_tuning/fabric.py) | The learner's use of a local Tessera store: principals, private nouns, blobs and claims. |
| [examples/continuous_tuning/fixture_backend.py](examples/continuous_tuning/fixture_backend.py) | A deterministic, dependency-free stand-in for the model backend (tests and CPU demos). |
| [examples/continuous_tuning/learner_spec.json](examples/continuous_tuning/learner_spec.json) | JSON object: format, learner, strategy, version, facet, reads |
| [examples/continuous_tuning/loop.py](examples/continuous_tuning/loop.py) | One on-demand tick: write spec nouns, enqueue a tick verb, then claim and run ready verbs. |
| [examples/continuous_tuning/mcp_tools.json](examples/continuous_tuning/mcp_tools.json) | JSON object: format, served_by, source_files_sha256, tools_sha256, tools |
| [examples/continuous_tuning/mcp_tools.served.json](examples/continuous_tuning/mcp_tools.served.json) | JSON object: format, served_by, source_files_sha256, tools_sha256, tools |
| [examples/continuous_tuning/metrics.py](examples/continuous_tuning/metrics.py) | Strict scoring of one generated reply against its computed label, plus a failure taxonomy. |
| [examples/continuous_tuning/mining.py](examples/continuous_tuning/mining.py) | Hard-example mining: turn dev failures into new, labeled training rows. |
| [examples/continuous_tuning/model_verbs.py](examples/continuous_tuning/model_verbs.py) | Model verbs: tune hyperparameters, train adapters and evaluate through a swappable backend. |
| [examples/continuous_tuning/pack.py](examples/continuous_tuning/pack.py) | Seed pack: the learner's nouns (spec, trigger, rule, recipes, base models, tool snapshot, verb templates) as… |
| [examples/continuous_tuning/pack/manifest.json](examples/continuous_tuning/pack/manifest.json) | JSON object: base_model_license, categories, does_not, evaluation, files, format |
| [examples/continuous_tuning/pack/records/base_model.35f7440ce8870ef2.json](examples/continuous_tuning/pack/records/base_model.35f7440ce8870ef2.json) | JSON object: allowed_surfaces, expected_version, key, namespace, store, tags |
| [examples/continuous_tuning/pack/records/base_model.a7110b742efbc097.json](examples/continuous_tuning/pack/records/base_model.a7110b742efbc097.json) | JSON object: allowed_surfaces, expected_version, key, namespace, store, tags |
| [examples/continuous_tuning/pack/records/learner_spec.c4fc7ab6d0eb99c5.json](examples/continuous_tuning/pack/records/learner_spec.c4fc7ab6d0eb99c5.json) | JSON object: allowed_surfaces, expected_version, key, namespace, store, tags |
| [examples/continuous_tuning/pack/records/pack_index.0adb151fcd9eae2d.json](examples/continuous_tuning/pack/records/pack_index.0adb151fcd9eae2d.json) | JSON object: allowed_surfaces, expected_version, key, namespace, store, tags |
| [examples/continuous_tuning/pack/records/promotion_rule.520abffa83ef6080.json](examples/continuous_tuning/pack/records/promotion_rule.520abffa83ef6080.json) | JSON object: allowed_surfaces, expected_version, key, namespace, store, tags |
| [examples/continuous_tuning/pack/records/search_space.099ed35b5dcc4f07.json](examples/continuous_tuning/pack/records/search_space.099ed35b5dcc4f07.json) | JSON object: allowed_surfaces, expected_version, key, namespace, store, tags |
| [examples/continuous_tuning/pack/records/tool_snapshot.04b90e2ce2e7b3b0.json](examples/continuous_tuning/pack/records/tool_snapshot.04b90e2ce2e7b3b0.json) | JSON object: allowed_surfaces, expected_version, key, namespace, store, tags |
| [examples/continuous_tuning/pack/records/training_recipe.c538f95f751732b8.json](examples/continuous_tuning/pack/records/training_recipe.c538f95f751732b8.json) | JSON object: allowed_surfaces, expected_version, key, namespace, store, tags |
| [examples/continuous_tuning/pack/records/trigger.06b5614255fea258.json](examples/continuous_tuning/pack/records/trigger.06b5614255fea258.json) | JSON object: allowed_surfaces, expected_version, key, namespace, store, tags |
| [examples/continuous_tuning/pack/records/verb_template.090e0a551bfb3b91.json](examples/continuous_tuning/pack/records/verb_template.090e0a551bfb3b91.json) | JSON object: allowed_surfaces, expected_version, key, namespace, store, tags |
| [examples/continuous_tuning/pack/records/verb_template.0dfc5a4aaf86364a.json](examples/continuous_tuning/pack/records/verb_template.0dfc5a4aaf86364a.json) | JSON object: allowed_surfaces, expected_version, key, namespace, store, tags |
| [examples/continuous_tuning/pack/records/verb_template.0eca35dba157fbe0.json](examples/continuous_tuning/pack/records/verb_template.0eca35dba157fbe0.json) | JSON object: allowed_surfaces, expected_version, key, namespace, store, tags |
| [examples/continuous_tuning/pack/records/verb_template.10b97ac15340c5ee.json](examples/continuous_tuning/pack/records/verb_template.10b97ac15340c5ee.json) | JSON object: allowed_surfaces, expected_version, key, namespace, store, tags |
| [examples/continuous_tuning/pack/records/verb_template.34d9a5878c6d511a.json](examples/continuous_tuning/pack/records/verb_template.34d9a5878c6d511a.json) | JSON object: allowed_surfaces, expected_version, key, namespace, store, tags |
| [examples/continuous_tuning/pack/records/verb_template.3dbf10fcd634a9fd.json](examples/continuous_tuning/pack/records/verb_template.3dbf10fcd634a9fd.json) | JSON object: allowed_surfaces, expected_version, key, namespace, store, tags |
| [examples/continuous_tuning/pack/records/verb_template.a625527e2177c54e.json](examples/continuous_tuning/pack/records/verb_template.a625527e2177c54e.json) | JSON object: allowed_surfaces, expected_version, key, namespace, store, tags |
| [examples/continuous_tuning/pack/records/verb_template.aa85127ec175f228.json](examples/continuous_tuning/pack/records/verb_template.aa85127ec175f228.json) | JSON object: allowed_surfaces, expected_version, key, namespace, store, tags |
| [examples/continuous_tuning/pack/records/verb_template.f7654361e9bb3583.json](examples/continuous_tuning/pack/records/verb_template.f7654361e9bb3583.json) | JSON object: allowed_surfaces, expected_version, key, namespace, store, tags |
| [examples/continuous_tuning/pack/tasks/tick.llama-3.2-1b-instruct.json](examples/continuous_tuning/pack/tasks/tick.llama-3.2-1b-instruct.json) | JSON object: allowed_surfaces, brief, effect_class, idempotency_key, kind, namespace |
| [examples/continuous_tuning/pack/tasks/tick.llama-3.2-3b-instruct.json](examples/continuous_tuning/pack/tasks/tick.llama-3.2-3b-instruct.json) | JSON object: allowed_surfaces, brief, effect_class, idempotency_key, kind, namespace |
| [examples/continuous_tuning/promotion.py](examples/continuous_tuning/promotion.py) | The pre-registered promotion rule, applied to candidate and champion evaluations. |
| [examples/continuous_tuning/promotion_rule.json](examples/continuous_tuning/promotion_rule.json) | JSON object: format, rule_id, version, registered, selection, promotion |
| [examples/continuous_tuning/recipes/base-llama-3.2-1b-instruct.json](examples/continuous_tuning/recipes/base-llama-3.2-1b-instruct.json) | JSON object: format, base_id, model_id, source, files_sha256, license |
| [examples/continuous_tuning/recipes/base-llama-3.2-3b-instruct.json](examples/continuous_tuning/recipes/base-llama-3.2-3b-instruct.json) | JSON object: format, base_id, model_id, source, weights, files_sha256 |
| [examples/continuous_tuning/recipes/lora-sft-r16.json](examples/continuous_tuning/recipes/lora-sft-r16.json) | JSON object: format, recipe_id, version, method, rank, alpha |
| [examples/continuous_tuning/recipes/search-lr3.json](examples/continuous_tuning/recipes/search-lr3.json) | JSON object: format, search_id, version, objective, max_trials, trials |
| [examples/continuous_tuning/report.py](examples/continuous_tuning/report.py) | Read-only learning curve from the stored nouns and tick logs: no model, no writes. |
| [examples/continuous_tuning/sanity.py](examples/continuous_tuning/sanity.py) | A small general-capability sanity set with computed answers (no tools, no system prompt). |
| [examples/continuous_tuning/serve.py](examples/continuous_tuning/serve.py) | Several specialists over one base: one loaded base, adapters attached by name. |
| [examples/continuous_tuning/task.py](examples/continuous_tuning/task.py) | Deterministic typed tool-call examples: request, offered tools, label and features. |
| [examples/continuous_tuning/templates.py](examples/continuous_tuning/templates.py) | Repository-authored request templates for typed Tessera tool calls (MIT, no model text). |
| [examples/continuous_tuning/tools.py](examples/continuous_tuning/tools.py) | Tessera's real MCP tool definitions, their families, and a strict schema checker. |
| [examples/continuous_tuning/trigger.json](examples/continuous_tuning/trigger.json) | JSON object: format, trigger_id, version, conditions, counted_origins, min_new_accepted_rows |
| [examples/continuous_tuning/verbs.py](examples/continuous_tuning/verbs.py) | Deterministic verbs: generate examples, curate datasets and mine hard examples. |
| [examples/finetuning/__init__.py](examples/finetuning/__init__.py) |  |
| [examples/finetuning/author.py](examples/finetuning/author.py) | Preview fine-tuning artifact nouns and worker verbs using existing Tessera contracts. |
| [examples/finetuning/data.py](examples/finetuning/data.py) | Owned synthetic research observations and family-separated dataset manifests. |
| [examples/finetuning/encoding.py](examples/finetuning/encoding.py) | Completion-only tokenization with explicit fit checks and no truncation. |
| [examples/finetuning/evidence/smoke-2026-10-29.json](examples/finetuning/evidence/smoke-2026-10-29.json) | JSON object: format, checked_at, verification, scope, runs, not_run |
| [examples/finetuning/glimmer.py](examples/finetuning/glimmer.py) | Prepare, but do not launch, a pinned official Glimmer QLoRA recipe on an H100 worker. |
| [examples/finetuning/kaggle.sh](examples/finetuning/kaggle.sh) | Run inside an already authorized Kaggle GPU session; this script starts no notebook or paid job. |
| [examples/finetuning/local_worker.py](examples/finetuning/local_worker.py) | Run a real tiny training verb through a private, task-owned local Tessera store. |
| [examples/finetuning/peft_backend.py](examples/finetuning/peft_backend.py) | Optional torch/PEFT backend; never imported by Tessera's service or data authoring. |
| [examples/finetuning/recipes/llama-1b-qlora.json](examples/finetuning/recipes/llama-1b-qlora.json) | JSON object: version, mode, model_id, revision, license, quantization |
| [examples/finetuning/recipes/llama-3b-qlora.json](examples/finetuning/recipes/llama-3b-qlora.json) | JSON object: version, mode, model_id, revision, license, quantization |
| [examples/finetuning/recipes/tiny-smoke.json](examples/finetuning/recipes/tiny-smoke.json) | JSON object: version, mode, model_id, revision, license, quantization |
| [examples/finetuning/requirements.txt](examples/finetuning/requirements.txt) | Optional worker environment only. |
| [examples/finetuning/run.py](examples/finetuning/run.py) | Prepare synthetic data or run one explicitly bounded optional worker training job. |
| [examples/research/collector.json](examples/research/collector.json) | JSON object: version, namespace, queue, output_store, stage, profile |
| [examples/research/discover.json](examples/research/discover.json) | JSON object: version, namespace, queue, output_store, stage, profile |
| [examples/research/scan.json](examples/research/scan.json) | JSON object: version, namespace, queue, output_store, stage, profile |
| [examples/research/schedule.json](examples/research/schedule.json) | JSON object: version, namespace, queue, output_store, stage, profile |
| [examples/showcase/opportunity-lab.json](examples/showcase/opportunity-lab.json) | Opportunity Lab |

## harness/ (3)

| File | What it is |
|---|---|
| [harness/README.md](harness/README.md) | Work harness |
| [harness/limits.toml](harness/limits.toml) | Starting limits for the work harness (workstream W). |
| [harness/rubrics.toml](harness/rubrics.toml) | Task rating rubric and priority weights. |

## scripts/ (7)

| File | What it is |
|---|---|
| [scripts/build_access_openapi.py](scripts/build_access_openapi.py) | Generate the shared access OpenAPI subset from actual Python route and model schemas. |
| [scripts/build_index.py](scripts/build_index.py) | Build this repo's indexes from its own files. |
| [scripts/build_ontology_catalog.py](scripts/build_ontology_catalog.py) | Generate the Worker ontology catalog from the canonical Python category/profile models. |
| [scripts/check_architecture.py](scripts/check_architecture.py) | Check that architecture.json still describes this repo truthfully. |
| [scripts/check_public.py](scripts/check_public.py) | Fail if a file headed for a public repo holds a secret or a private name. |
| [scripts/dev_stack.sh](scripts/dev_stack.sh) | Run Tessera on this machine behind the real Cloudflare edge code (wrangler dev), with the origin |
| [scripts/suite_http_local.py](scripts/suite_http_local.py) | Run the whole conformance suite over HTTP against a real tessera serve process on this machine. |

## src/ (64)

| File | What it is |
|---|---|
| [src/tessera/__init__.py](src/tessera/__init__.py) | Tessera: atomic work queue and atomic storage for agent swarms. |
| [src/tessera/_access_boundary.py](src/tessera/_access_boundary.py) | One authenticated request boundary for HTTP and MCP; discovery remains public. |
| [src/tessera/_access_contract.py](src/tessera/_access_contract.py) | Public access contract shared by Python HTTP, MCP and the connection page. |
| [src/tessera/_access_models.py](src/tessera/_access_models.py) | Public access response records shared by generated HTTP/OpenAPI descriptions. |
| [src/tessera/_access_routes.py](src/tessera/_access_routes.py) | Anonymous session, self identity, audit, key lifecycle and long-poll HTTP routes. |
| [src/tessera/_access_schema.py](src/tessera/_access_schema.py) | Additive local principal/audit schema and periodic anonymous state cleanup. |
| [src/tessera/_base.py](src/tessera/_base.py) | Plumbing shared by the store's operation mixins: one SQLite connection per thread, IMMEDIATE transactions, na… |
| [src/tessera/_buckets.py](src/tessera/_buckets.py) | Queryable noun stores and persistent named file buckets (SPEC §6 extension). |
| [src/tessera/_execution_capture.py](src/tessera/_execution_capture.py) | Atomic execution noun capture; indexes never stand in for source authorization. |
| [src/tessera/_executions.py](src/tessera/_executions.py) | Permission-filtered execution recall over immutable nouns and rebuildable indexes. |
| [src/tessera/_frontier_guard.py](src/tessera/_frontier_guard.py) | Bind artifact publication to completed repair, verification and independent review. |
| [src/tessera/_graph.py](src/tessera/_graph.py) | Task-graph plumbing shared by the verbs: releasing dependents when a task resolves, waking parked tasks, and… |
| [src/tessera/_intelligence.py](src/tessera/_intelligence.py) | 0-call progressive L0 scent search, monotonic per-(namespace, feed) delta feeds, and dual-resolution reflex r… |
| [src/tessera/_jobs.py](src/tessera/_jobs.py) | Model jobs (SPEC §6): each Meta Model API background response is tied to the task and the fencing token that… |
| [src/tessera/_keys.py](src/tessera/_keys.py) | Worker keys (SPEC §7): random bearer secrets stored only as SHA-256 hashes, scoped to queue patterns, task ki… |
| [src/tessera/_legacy_privacy.py](src/tessera/_legacy_privacy.py) | One-time restriction of automatically derived legacy task tombstones. |
| [src/tessera/_mcp_access.py](src/tessera/_mcp_access.py) | MCP access tools use the same principal context as HTTP. |
| [src/tessera/_mcp_storage.py](src/tessera/_mcp_storage.py) | Storage, bucket, 0-call L0 search, live feed, and reflex-recall MCP tools for tessera.mcpdoor. |
| [src/tessera/_observe.py](src/tessera/_observe.py) | Read-only views: tasks, results, the counts record, health, the event log, and Board v2 snapshots (SPEC §5.7… |
| [src/tessera/_principals.py](src/tessera/_principals.py) | Durable principal metadata, bounded anonymous quotas and content-free access audit. |
| [src/tessera/_publication_guard.py](src/tessera/_publication_guard.py) | Verify publication against immutable completed queue evidence inside the write transaction. |
| [src/tessera/_requesters.py](src/tessera/_requesters.py) | Requester keys: self-serve sign-up for organizations that ask for work, with no operator step. |
| [src/tessera/_service_buckets.py](src/tessera/_service_buckets.py) | HTTP routes for queryable noun stores, persistent named file buckets, and remote key management. |
| [src/tessera/_service_executions.py](src/tessera/_service_executions.py) | Execution nouns retain task permissions across HTTP and MCP. |
| [src/tessera/_service_intelligence.py](src/tessera/_service_intelligence.py) | HTTP routes for progressive L0 scent search, HTTP 206 byte-range slicing, live delta feeds, and 0-call dual-r… |
| [src/tessera/_storage.py](src/tessera/_storage.py) | Atomic storage (SPEC §6): versioned records written by single-statement SQL compare-and-set, with every versi… |
| [src/tessera/_tasks.py](src/tessera/_tasks.py) | The five verbs (claim, heartbeat, split, submit, cancel), park and resume, plus enqueue with dependencies, qu… |
| [src/tessera/_version.py](src/tessera/_version.py) |  |
| [src/tessera/access.py](src/tessera/access.py) | What every door shares: the service settings, the claim rate limiter, reading the bearer key, and the Gate th… |
| [src/tessera/backend.py](src/tessera/backend.py) | The contract every Tessera backend meets: the in-process Store, the HTTP Client, and any adapter an app puts… |
| [src/tessera/bridge.py](src/tessera/bridge.py) | Reusable Surface Bridge wrapper for stateless product frontends and services. |
| [src/tessera/claim_admission.py](src/tessera/claim_admission.py) | Async, process-local claim pacing; leases and access rules remain in the Store. |
| [src/tessera/cli.py](src/tessera/cli.py) | tessera: drive the in-process store on a SQLite file from a shell. |
| [src/tessera/client.py](src/tessera/client.py) | HTTP client for a Tessera service (SPEC §7). |
| [src/tessera/clock.py](src/tessera/clock.py) | Time source for leases and backoff. |
| [src/tessera/errors.py](src/tessera/errors.py) | Errors every Tessera backend raises. |
| [src/tessera/execution_matching.py](src/tessera/execution_matching.py) | Bounded execution-report validation and candidate matching, never reuse authorization. |
| [src/tessera/fingerprints.py](src/tessera/fingerprints.py) | Deterministic problem fingerprints: a small pure library, not a learning policy. |
| [src/tessera/grants.py](src/tessera/grants.py) | What a worker key may do: its queues, kinds, effect classes and scopes, and the rules that narrow a request t… |
| [src/tessera/learners/__init__.py](src/tessera/learners/__init__.py) | Swappable learners over execution observations. |
| [src/tessera/learners/catalog.py](src/tessera/learners/catalog.py) | Learner specs for every learning facet: three implemented, the rest proposed designs. |
| [src/tessera/learners/contract.py](src/tessera/learners/contract.py) | The common learner contract: spec nouns, config nouns and run verbs. |
| [src/tessera/learners/memory.py](src/tessera/learners/memory.py) | Memory learner, strategy minhash-lsh v1: asynchronous fingerprint index plus recall. |
| [src/tessera/learners/pack.py](src/tessera/learners/pack.py) | Deterministic seed-pack export of learner spec nouns, a default config and verb templates. |
| [src/tessera/learners/path_statistics.py](src/tessera/learners/path_statistics.py) | Path statistics learner, strategy wilson-grouping v1: what worked, per path. |
| [src/tessera/learners/routing.py](src/tessera/learners/routing.py) | Routing-weights learner, strategy beta-bandit v1: advisory selection weights per family. |
| [src/tessera/learners/runtime.py](src/tessera/learners/runtime.py) | Shared runtime for learner run verbs: pinned spec checks, bounded paging, output writes. |
| [src/tessera/learners/stats.py](src/tessera/learners/stats.py) | Deterministic outcome statistics over execution observations, shared by learners. |
| [src/tessera/learners/worker.py](src/tessera/learners/worker.py) | Run one claimed learner verb with its registered deterministic runner, then submit. |
| [src/tessera/mcp_door.py](src/tessera/mcp_door.py) | The MCP door (SPEC §7): atomic verbs, progressive L0 noun search, CAS records, buckets, delta feeds, and refl… |
| [src/tessera/models.py](src/tessera/models.py) | Records that cross Tessera's boundary: what callers send and what every backend returns. |
| [src/tessera/observations.py](src/tessera/observations.py) | The optional report.learning block and the stable execution observation view. |
| [src/tessera/ontology/__init__.py](src/tessera/ontology/__init__.py) | Ontology, Ranking, Fragility, Conflict-Reconciliation, and Profile-Adaptive Engine. |
| [src/tessera/ontology/categories.py](src/tessera/ontology/categories.py) | Categorized Noun and Verb taxonomy, typed dimensions, relations, and worker guidance. |
| [src/tessera/ontology/fragility.py](src/tessera/ontology/fragility.py) | Fragility scoring, exponential staleness decay, recheck planning, and drift detection. |
| [src/tessera/ontology/profiles.py](src/tessera/ontology/profiles.py) | Caller profile policies and adaptive rendering for Nouns and Verb task guidance. |
| [src/tessera/ontology/rankings.py](src/tessera/ontology/rankings.py) | Empirical-Bayes shrunk rankings, Wilson bounds, composite scores, and Pareto frontiers. |
| [src/tessera/ontology/reconcile.py](src/tessera/ontology/reconcile.py) | Deterministic 0-call 3-way CAS merge and Sybil-fenced calibration-weighted voting. |
| [src/tessera/pages.py](src/tessera/pages.py) | The agent door: /connect, /llms.txt, /llms-full.txt, /favicon.svg, the agent card, robots rules, Console Boar… |
| [src/tessera/receipts.py](src/tessera/receipts.py) | Canonical JSON and receipt hashes (SPEC §5.4). |
| [src/tessera/schema.sql](src/tessera/schema.sql) | Tessera schema v3 (SPEC §3). |
| [src/tessera/service.py](src/tessera/service.py) | Tessera's HTTP service (SPEC §7). |
| [src/tessera/store.py](src/tessera/store.py) | The in-process Tessera store: the reference implementation of SPEC.md. |
| [src/tessera/testing.py](src/tessera/testing.py) | Factories the conformance suite loads by name (TESSERABACKENDFACTORY="module:function"). |

## tests/ (69)

| File | What it is |
|---|---|
| [tests/agent_start.test.mjs](tests/agent_start.test.mjs) |  |
| [tests/claim_admission.test.mjs](tests/claim_admission.test.mjs) |  |
| [tests/edge_access_harness.mjs](tests/edge_access_harness.mjs) | No remote HTTP, credentials, or persistent production data: real Worker handlers, synthetic SQLite. |
| [tests/edge_access_loader.mjs](tests/edge_access_loader.mjs) | Minimal Cloudflare DurableObject class for deterministic Node SQLite tests. |
| [tests/edge_access_publication.mjs](tests/edge_access_publication.mjs) | Storage-level approvals bind an authenticated draft, reviewer, receipt, and exact catalog value. |
| [tests/edge_access_review.mjs](tests/edge_access_review.mjs) | Regressions for authorization before durable writes and public observation boundaries. |
| [tests/edge_category_queries.mjs](tests/edge_category_queries.mjs) | Actual coordinator, SQL and Worker ranking handlers; temporary synthetic data only. |
| [tests/edge_core_boundaries.mjs](tests/edge_core_boundaries.mjs) | Actual Worker routes with the harness's synthetic in-memory SQLite bindings. |
| [tests/edge_dead_letter_privacy.mjs](tests/edge_dead_letter_privacy.mjs) | Real tombstone SQL and shared Worker ACL/search code; only synthetic local state. |
| [tests/edge_execution_history.mjs](tests/edge_execution_history.mjs) |  |
| [tests/edge_execution_mcp.mjs](tests/edge_execution_mcp.mjs) | MCP adapter fixture only: authorization/storage are covered by the real runtime suite. |
| [tests/edge_frontier_publication.mjs](tests/edge_frontier_publication.mjs) |  |
| [tests/fixtures/access_parity.json](tests/fixtures/access_parity.json) | JSON object: about, steps, queues, cases |
| [tests/fixtures/execution_learning_vectors.json](tests/fixtures/execution_learning_vectors.json) | JSON object: fingerprints, learning_invalid, learning_valid, observations, profile, reports |
| [tests/fixtures/execution_matching_vectors.json](tests/fixtures/execution_matching_vectors.json) | JSON object: profile, valid, invalid |
| [tests/test_access_modes.py](tests/test_access_modes.py) | Full anonymous/keyed agent loops, durable limits and adversarial isolation. |
| [tests/test_access_parity.py](tests/test_access_parity.py) | The access cases in tests/fixtures/accessparity.json, run against the Python service. |
| [tests/test_agent_tool_listings.py](tests/test_agent_tool_listings.py) | Every agent-facing tool listing on the Python service matches what MCP tools/list serves. |
| [tests/test_blob_first_uploader.py](tests/test_blob_first_uploader.py) | Raw blob governance is one row per namespace and digest, written by the first uploader. |
| [tests/test_board.py](tests/test_board.py) | Console Board v2 (/board, L19), UI chrome, and agent discovery routes. |
| [tests/test_buckets.py](tests/test_buckets.py) | Tests for queryable noun stores, persistent named file buckets, external blobdir, and MCP tools. |
| [tests/test_category_queries.py](tests/test_category_queries.py) | Category projections agree before bounded scans, without granting visibility. |
| [tests/test_check_public.py](tests/test_check_public.py) | checkpublic finds secrets and private names, masks them, and honours the allow marker. |
| [tests/test_claim_admission.py](tests/test_claim_admission.py) | Real async cancellation, fairness, resource bounds and HTTP default admission. |
| [tests/test_client_addresses.py](tests/test_client_addresses.py) | Per-IP limits key on an address the client cannot choose (finding 6), counted in the hour its daily-rotated h… |
| [tests/test_continuous_tuning.py](tests/test_continuous_tuning.py) | Continuous tuning example: real tool definitions, held-out splits, contamination checks, strict metrics, mini… |
| [tests/test_core_boundaries.py](tests/test_core_boundaries.py) | Privacy over version history, strict outcomes, deadline boundaries and category plans. |
| [tests/test_dead_letter_privacy.py](tests/test_dead_letter_privacy.py) | Failure diagnostics remain private even when the source verb was shared. |
| [tests/test_execution_adapters.py](tests/test_execution_adapters.py) | HTTP client framing and real local MCP execution-report flow, without inference. |
| [tests/test_execution_distillation.py](tests/test_execution_distillation.py) | Bounded task authoring and private local claim fixtures, without model execution. |
| [tests/test_execution_history.py](tests/test_execution_history.py) | Execution recall uses accepted results and rechecks source permissions and expiry. |
| [tests/test_execution_learning.py](tests/test_execution_learning.py) | Learning reports, the observation read interface and swappable learners, end to end. |
| [tests/test_execution_learning_vectors.py](tests/test_execution_learning_vectors.py) | Problem fingerprints, learning-block checks and observations match across Python and Worker. |
| [tests/test_execution_matching.py](tests/test_execution_matching.py) | Execution fingerprints are portable retrieval evidence, not completion proofs. |
| [tests/test_execution_output_budget.py](tests/test_execution_output_budget.py) | Large historical results are loaded only for selected hits and bounded in recall. |
| [tests/test_external_access_edge.py](tests/test_external_access_edge.py) | External-agent access through real Worker HTTP/MCP handlers and local SQLite bindings. |
| [tests/test_finetuning_example.py](tests/test_finetuning_example.py) | Behavioral checks for optional training data, fit rejection, and existing noun/verb contracts. |
| [tests/test_frontier_publication.py](tests/test_frontier_publication.py) | Publication requires exact, namespace-bound evidence from independent principals. |
| [tests/test_governance_and_isolation.py](tests/test_governance_and_isolation.py) | Universal 6-field governance envelope, namespace isolation (/ns/{namespace} & X-Namespace), Layer 1 surface o… |
| [tests/test_guard_hook.py](tests/test_guard_hook.py) | The guard hook blocks protected paths and risky commands, and fails closed. |
| [tests/test_handshake_and_bridge.py](tests/test_handshake_and_bridge.py) | Tests for Surface Handshake v1.0 (POST /v1/handshake + MCP handshake) and SurfaceBridge. |
| [tests/test_intelligence.py](tests/test_intelligence.py) | Tests for progressive L0 scent search, HTTP 206 byte ranges, delta feeds, dead-end tombstones, and 0-call dua… |
| [tests/test_legacy_tombstone_migration.py](tests/test_legacy_tombstone_migration.py) | Existing diagnostic heads are restricted on upgrade, not only newly failed tasks. |
| [tests/test_ontology_and_reconcile.py](tests/test_ontology_and_reconcile.py) | Tests for Tessera Wave-2 Categorized Ontology, Empirical-Bayes Rankings, Fragility Decay, 3-Way CAS Reconcili… |
| [tests/test_open_verbs.py](tests/test_open_verbs.py) | Open verbs: a tracked anonymous session claims, heartbeats and submits only verbs whose producer opened them,… |
| [tests/test_phase4_atomicity_and_vectors.py](tests/test_phase4_atomicity_and_vectors.py) | Phase 4 atomicity, migration, T22 gates, L26 receipt vectors, and Edge parity tests. |
| [tests/test_requester_keys.py](tests/test_requester_keys.py) | Requester keys: a self-serve sign-up mints a hashed key that writes nouns only in its own org.<id> namespace,… |
| [tests/test_research_seed.py](tests/test_research_seed.py) | Bounded authoring, pinned inputs and a real local task loop; no provider calls. |
| [tests/test_service.py](tests/test_service.py) | The HTTP service: keys, scopes, POST-only writes, holders, rate limits, the edge lock, the agent door, the we… |
| [tests/test_showcase_seed.py](tests/test_showcase_seed.py) | Creative seeds remain bounded and idempotent through actual Store operations. |
| [tests/test_spec_events.py](tests/test_spec_events.py) | SPEC §5.5 lists every event type, and the code emits only listed types: a new event can't ship without a line… |
| [tests/test_tools.py](tests/test_tools.py) | The Muse worker keeps its safety flags whatever the brief says, and briefs load in dependency order. |
| [tests/test_work_routing.py](tests/test_work_routing.py) | Explicit grants, team-private routing and owner-only negative controls. |
| [tests/workerd_access_descriptions.mjs](tests/workerd_access_descriptions.mjs) | Real Worker: every published access description (connect.json, the agent card, llms-full.txt, |
| [tests/workerd_access_parity.mjs](tests/workerd_access_parity.mjs) | Real Worker/SQLite: the access cases in tests/fixtures/accessparity.json, which |
| [tests/workerd_access_writes.mjs](tests/workerd_access_writes.mjs) | Real Worker/SQLite: requests that never authenticate write nothing and create no namespace |
| [tests/workerd_agent_tool_listings.mjs](tests/workerd_agent_tool_listings.mjs) | Real Worker: every agent-facing tool listing matches what MCP tools/list serves. |
| [tests/workerd_anonymous_cost.mjs](tests/workerd_anonymous_cost.mjs) | Real Worker/SQLite: anonymous sessions stay available but cost little until used, and one |
| [tests/workerd_blob_first_uploader.mjs](tests/workerd_blob_first_uploader.mjs) | Real Worker/SQLite: a key can always read back bytes it uploaded, whoever stored them first, |
| [tests/workerd_blob_identity.mjs](tests/workerd_blob_identity.mjs) | Real Worker/SQLite: content identity comes from the uploaded bytes, never from a client header. |
| [tests/workerd_client_addresses.mjs](tests/workerd_client_addresses.mjs) | Real Worker/SQLite: per-IP limits count in the hour their daily-rotated IP hash was computed, so a |
| [tests/workerd_execution_history.mjs](tests/workerd_execution_history.mjs) | Actual Worker handlers and ephemeral SQLite Durable Objects; no provider or cloud access. |
| [tests/workerd_execution_learning.mjs](tests/workerd_execution_learning.mjs) | Actual Worker and SQLite Durable Objects: learning reports, observations and row costs. |
| [tests/workerd_open_verbs.mjs](tests/workerd_open_verbs.mjs) | Real Worker/SQLite: a tracked anonymous session claims, heartbeats and submits only verbs whose |
| [tests/workerd_reaper.mjs](tests/workerd_reaper.mjs) | Real Worker/SQLite: the lease reaper keeps running in the main namespace even when an alarm |
| [tests/workerd_requester_keys.mjs](tests/workerd_requester_keys.mjs) | Real Worker/SQLite: a self-serve requester key writes nouns only in its own org.<id> namespace, |
| [tests/workerd_requester_storage.mjs](tests/workerd_requester_storage.mjs) | Real Worker/SQLite with persisted objects: a requester sign-up writes only the primary object; the |
| [tests/workerd_storage_exhaustion.mjs](tests/workerd_storage_exhaustion.mjs) | Real Worker/SQLite under a faithful storage-exhaustion simulation: after the provider's daily |
| [tests/workerd_storage_quota.mjs](tests/workerd_storage_quota.mjs) | Real Worker/SQLite behavior; fixture-only RPC methods never ship in the product. |

## tools/ (11)

| File | What it is |
|---|---|
| [tools/execution_distillation.py](tools/execution_distillation.py) | Prepare one bounded external-worker distillation verb. |
| [tools/learner.py](tools/learner.py) | Learner seed-pack and run helper; offline commands are the default. |
| [tools/load_briefs.py](tools/load_briefs.py) | Put briefs on the build queue once their dependencies are done. |
| [tools/meta_open_bridge.py](tools/meta_open_bridge.py) | Local Meta Model API /v1/responses bridge backed by Meta open-weight models (muse-spark-open) on any OpenAI/O… |
| [tools/muse_worker.py](tools/muse_worker.py) | Run one Muse Code worker on one Tessera brief. |
| [tools/opencode_bilateral.py](tools/opencode_bilateral.py) | Bilateral multi-agent test harness for Tessera over MCP (/mcp) + HTTP (/v1/), pairing OpenCode (opencode v2.x… |
| [tools/research_seed.py](tools/research_seed.py) | Prepare one bounded research seed verb; preview is the default. |
| [tools/seed.py](tools/seed.py) | Seed a Tessera database for Muse agents: the standard queues, one key per agent, and first tasks. |
| [tools/showcase_seed.py](tools/showcase_seed.py) | Author bounded creative seed nouns and verbs. |
| [tools/swarm_expander.py](tools/swarm_expander.py) | Run a Muse / Meta Model API swarm worker that seeds, expands, processes, and promotes atomic nouns and verbs… |
| [tools/team_private_work.py](tools/team_private_work.py) | Local executable work-routing example, using HTTP handlers and temporary state. |