From 50f2edc154591d5b233f2969116483c7e02d65d7 Mon Sep 17 00:00:00 2001
From: Programmable <309941960+0xprogrammable@users.noreply.github.com>
Date: Sun, 9 Aug 2026 00:04:58 +0200
Subject: [PATCH] docs: redesign Hookbuilder public presentation
---
README.md | 352 +++++++++++----------------
assets/README.md | 10 +-
assets/repository-cover.jpg | Bin 403024 -> 111474 bytes
assets/repository-social-preview.jpg | Bin 0 -> 84289 bytes
4 files changed, 154 insertions(+), 208 deletions(-)
create mode 100644 assets/repository-social-preview.jpg
diff --git a/README.md b/README.md
index 4abfa81..02491b5 100644
--- a/README.md
+++ b/README.md
@@ -1,228 +1,177 @@
- 
+
Programmable v4 Builder
- Describe a Programmable product—including a Uniswap v4 hook, token, app, game, service, or standalone settlement. Give the same portable skill to your coding agent. Build, check, and prepare one exact GitHub application.
+ A portable Agent Skill for turning a product idea or an existing repository into a complete, reviewable Programmable project with explicit evidence.
- 
- 
- 
- 
+ Maintained by Programmable · Developed in public · MIT licensed
-Programmable v4 Builder is an evidence-first [Agent Skill](https://agentskills.io/specification) for open-ended
-Programmable projects, including but not limited to Uniswap v4. It combines protocol mechanics, Programmable's platform
-contract, reusable planning blocks, deterministic checks, adversarial evaluations, and a GitHub application workflow in
-one versioned package.
+
+ 
+ 
+ 
+ 
+
+
+
+ Install ·
+ Use the Builder ·
+ Contribute ·
+ Security
+
+
+> [!IMPORTANT]
+> **Release status**
+>
+> `v0.4.0` is the latest published release. `main` contains the public `v0.5.1` development source. Its
+> integration on `main`, package version, or green CI does not make it a stable release. It does not activate
+> Application V3 intake, Registry acceptance, deployment, or launch.
-The Open-World V2, Fee V2, Application V3, and Launch Bundle V2 work described below is an unreleased local candidate.
-Its local tools do not make Registry V3 intake active, accept a project, authorize launch, deploy code, or prove provider
-availability. The public `v0.4.0` release remains current until the candidate release gates pass.
+Programmable v4 Builder is an evidence-first [Agent Skill](https://agentskills.io/specification). It accepts a plain
+idea or an existing public repository. Hooks, tokens, apps, games, services, standalone settlement systems, and mixed
+projects can all be modeled.
-It is built for ordinary token launches and unfamiliar ideas alike: custom hooks, games, maps, wallet quests, auctions,
-dynamic fees, transfer taxes, automatic liquidity, external services, active-liquidity markets, wrapped assets,
-position automation, and architectures that do not have a catalog name yet. Templates accelerate repeated work; they
-never define what builders are allowed to invent.
+The Builder does not limit unfamiliar work to a fixed catalog. Starters and capability packs speed up repeated work;
+unknown ideas remain eligible for architecture review. Local checks stay separate from maintainer review, Registry
+acceptance, launch authority, deployment, provider support, and public availability.
-## Install
+## Install the published release
-The shortest interactive command is:
+Preview the exact published Skill before installing it:
```bash
-gh skill install 0xprogrammable/programmable-v4-builder
+gh skill preview 0xprogrammable/hookbuilder \
+ programmable-v4-hook-builder@v0.4.0
```
-For a reproducible Codex installation pinned to this release:
+Install the same immutable release for Codex:
```bash
-gh skill install 0xprogrammable/programmable-v4-builder \
- programmable-v4-hook-builder@v0.4.0 \
+gh skill install 0xprogrammable/hookbuilder \
+ skills/programmable-v4-hook-builder \
--agent codex \
- --scope user
+ --scope user \
+ --pin v0.4.0
```
-Replace `codex` with `claude-code`, `github-copilot`, `cursor`, `gemini-cli`, or another host supported by `gh skill`.
-The GitHub CLI skill commands are currently a preview feature. Installation grants no wallet, deployment, publication,
-approval, or external-account authority.
+GitHub's `gh skill` commands are in preview. Clean package placement has been checked for Codex, Claude Code, and
+GitHub Copilot; host behavior remains a separate question. See
+[Portability and lifecycle](docs/PORTABILITY_AND_LIFECYCLE.md) before choosing another destination.
-Then tell the agent:
+From the installed Skill directory, verify its package:
-```text
-Use the Programmable v4 Builder skill. Turn this idea into a complete checked project and prepare its GitHub application:
+```bash
+node scripts/verify-skill.mjs --installed
```
-For an existing project:
+Then give your agent one clear request:
```text
-Use the Programmable v4 Builder skill. Inspect this repository, preserve the intended product, repair objective findings, and prepare one exact GitHub application revision:
+Use the Programmable v4 Builder skill. Start from this idea or repository: . Preserve the intended product, choose the smallest complete architecture, run the available checks, and prepare one exact GitHub application. Ask before any external write.
```
-The agent keeps the mechanics behind that request explicit: it validates and freezes the project, derives the unique
-Application V3 revision/lineage with GET-only GitHub reads, builds the complete source-assessed package locally with no
-network, then shows a read-only submission/update plan. A GitHub draft is written only after separate confirmation of
-that exact current plan digest. No local success is called acceptance, deployment or launch.
-
-## Why this is more than a documentation page
-
-Documentation explains. This package also gives the agent deterministic machinery:
-
-- a local knowledge router that loads only the chapters required for the current job;
-- live, hash-bound search of the canonical public Programmable project Registry;
-- an explicit offline Registry fallback proven against one already-public commit through a portable raw-Git receipt,
- always labeled as a stale point-in-time baseline rather than live state;
-- composable starters and capability packs with hash-bound provenance;
-- a plain-language intent contract and open-world product graph with owner-defined capabilities;
-- first-class trade-capability outputs: explicit `tradable`/`no-market`/`unresolved` classification, one closed
- `NOT_APPROVED` route manifest per tradable market, and typed quote/execution evidence;
-- deterministic scaffolding, semantic validation, security assessment and intent-fidelity checks;
-- Fee V2 profiles for AMM, zero-AMM, async/batched and custom-reviewed settlement designs;
-- exact multi-repository source closure with content-addressed manifests for large projects;
-- staged, post-commit security and source-verification evidence that avoids a source hash cycle;
-- a derived, content-bound, GitHub-only Application V3 with GET-only revision discovery, zero-network package building,
- inline and manifest closure modes;
-- duplicate-key-safe bounded JSON handling before privacy, integrity or semantic review;
-- read-only application status plus two-step confirmation for an authorized draft submission;
-- upstream source drift detection, migrations, release planning, and installed-package verification; and
-- adversarial evals for protocol, SDK, liquidity, signing, claims, creativity, repository and external-action failures.
-
-## One idea, progressive context
-
-The agent does not preload the entire knowledge base.
+## What the Builder does
+
+| Stage | Output |
+| --- | --- |
+| Understand | Preserves the stated product intent and isolates the owner choices that materially change it. |
+| Design | Produces the smallest complete architecture, capability graph, value flow, and repository plan. |
+| Build | Uses composable starters and capability packs without treating them as an allowlist. |
+| Check | Runs deterministic validation, intent-fidelity checks, security boundaries, and executable evidence where available. |
+| Bind | Closes over the exact public source repositories, commits, trees, files, and required evidence. |
+| Prepare | Builds a local, source-bound application package and a read-only GitHub plan before any authorized write. |
+
+The Builder loads only the protocol, runtime, liquidity, service, and security material relevant to the confirmed
+project.
```mermaid
-flowchart LR
- A["Plain-language idea or repository"] --> I["Verbatim intent contract"]
- I --> H["Live canonical project archive"]
- H --> B["Local context router"]
- B --> C["Starter plus Lego capabilities"]
- C --> D["Design and value-flow checks"]
- D --> T["Trade classification and route contract"]
- T --> E["Implementation and executable evidence"]
- E --> R["GET-only revision and lineage derivation"]
- R --> F["Zero-network exact application package"]
- F --> G["Programmable maintainer review"]
+flowchart TD
+ A["Idea or public repository"] --> B["Intent and material owner choices"]
+ B --> C["Architecture and repository plan"]
+ C --> D["Implementation and local checks"]
+ D --> E["Exact source-bound application"]
+ E --> F["Programmable maintainer review"]
+ F -. "separate authority and evidence" .-> G["Registry acceptance"]
+ G -. "separate authority and evidence" .-> H["Deployment and public availability"]
```
-The router reports exact bytes and a reproducible `ceil(UTF-8 bytes / 4)` estimate for every profile. The 18,000-token
-initial target is a routing budget, not a safety cap or model-tokenizer claim. More complex products add only the
-confirmed protocol, runtime, service, liquidity, or security context. Unknown capabilities remain eligible and route to
-architecture review; routing follows confirmed capabilities and surfaces, never one prompt word or a fixed taxonomy.
-Required context is reported and retained when it exceeds the target.
-
-Try the router locally:
+Inspect the local router on the development source:
```bash
node skills/programmable-v4-hook-builder/scripts/cli.mjs context --mode explore
-node skills/programmable-v4-hook-builder/scripts/cli.mjs discover search "ordinary token with automatic liquidity"
```
-## Product rules that do not disappear inside templates
-
-First derive fee applicability from the exact project graph. A proposal is `unresolved`. A prototype with at least one
-`programmable-canonical` scope is `applicable` and binds a real Fee V2 instance. An exact zero-scope prototype is
-`not-applicable`, keeps instance fields null, and must not invent a market, PoolKey, hook or fee receipt.
-
-Every Programmable-canonical execution scope must preserve Programmable's inclusive 10 bps share of executed gross
-quote-side swap or fill volume exactly once. If a builder selects a 3% total charge, the total remains 3%:
-0.1% Programmable and 2.9% project. The immutable claim authority is
-`0x4957f49620AFf3Adbbe8195a4f633E49cc93376c`. The separate platform administration wallet,
-`0x2Bb333d48DFAF1596D9036671d2E43168994249E`, can exercise only independently authorized platform controls; it cannot
-claim, replace, redirect, sweep, or net the fee liability.
-
-V2 encodes every EVM `chainId` as a canonical positive `uint256` decimal string (`"1"`, never the JSON number `1`). Fee
-collection uses the settlement-specific `standard-amm`, `sync-custom-zero-amm`, `async-fill-batch`, or
-`custom-reviewed` profile. The bundled Solidity kernel evidences only `standard-amm`; the other profiles need their own
-implementation, custody proof, tests, and review.
-
-When applicability is `applicable`, prototype conformance uses a typed receipt plus canonical vector-set bytes, not a
-free-form evidence string. A complete record carries `evidenceRefs`, exact `evidenceDigests[]`, and one typed
-receipt/vector pair in `scopeArtifacts[]` for every fee scope. It binds the implementation, source commit/tree,
-scope/profile, execution-surface coverage, exposed modes, vectors and evidence. User-funded modes remain below 100%; a
-high-rate custom profile requires exact segregated external funding and solvency evidence. Structural conformance is not
-an audit, deployment or approval.
-
-Application V3, Registry Acceptance V3, and Launch V2 likewise encode `applicationRevision` as a canonical positive
-decimal string, without the historical V1 integer, `1,000,000`, or JavaScript safe-integer ceiling. Frozen V1 records
-retain their original integer revision shape. A new V3 application starts at `"1"`; each update binds the exact prior
-revision and increments it by one with arbitrary-precision semantics.
-
-Application V3 transports the derived applicability. Registry Acceptance V3 may accept an exact zero-scope project as
-`not-applicable` with null fee-instance fields. Launch Bundle V2 remains a separate canonical fee-bearing contract: N/A
-cannot be converted into a fabricated instance or launch authorization, and every local launch result stays unsigned
-and `NOT_AUTHORIZED`.
-
-The Builder's Registry Acceptance V3 bridge is also preflight, not platform authority. It resolves the stable central
-Registry repository ID, closed-and-merged Application PR, central pull ref, current-head OWNER approval and exact
-raw-Git package/change-set digests under aggregate request, byte and deadline bounds. The same protected pass reads
-central `refs/heads/main` twice and replays the exact bound acceptance blob from that stable current commit. Builder identity follows the
-immutable GitHub user ID, so a renamed login does not break lineage. Maintainer and Registry authority also follow their
-immutable numeric IDs; current login, full name and URI remain internally consistent observed metadata, not authority.
-Mocked or injected transports return inspection-only results, caller JSON never authenticates, and production must
-independently re-run the verification before any later authorization.
-
-Application V3 may downgrade an exact intentionally public financial-identifier candidate from privacy block to mandatory
-human review only through an exact pointer, substring digest, application binding, owner-stated purpose and matching
-authorization record. That attestation proves neither ownership nor approval and can never waive a secret, credential,
-private key, seed phrase, password or API/auth token.
-
-The Builder keeps these states separate:
-
-1. idea and design eligibility;
-2. implementation conformance;
-3. Programmable maintainer review;
-4. deployment and source/runtime verification;
-5. provider-specific indexing, quoting, simulation, and execution; and
-6. public availability.
-
-A local green check proves only the exact local check. It is not an audit, approval, deployment, provider statement,
-Uniswap endorsement, or guarantee that code is safe or rug-free.
-
-The public application path accepts exact public GitHub repositories only. A local directory, ZIP, pasted source,
-private repository, or another Git host can support exploration but cannot satisfy Application V3; it stays idea
-eligible as `INTEGRATION_PENDING`, with no public package or external write. Application transport, maintainer
-acceptance, launch preparation, admin authorization, deployment/runtime verification, indexing, quoting, and public
-availability remain separate evidence states.
-
-The current source-closure-manifest V1 transport supports SHA-1 Git object databases, separately records SHA-256 content
-digests, and requires UTF-8 committed paths. A Git SHA-256 object database or non-UTF-8 path is an
-`INTEGRATION_PENDING` transport case with `ELIGIBLE_FOR_REVIEW` idea status. A UTF-8 path above the generator's current
-16 KiB path-byte budget is `HOLD_SPLIT_REVIEW` under the same eligibility. None is an unsafe or ineligible product;
-other object formats or path encodings require a new versioned closure contract rather than widening V1 in place.
+## Build it with us
+
+The Builder is maintained by Programmable and developed in public. Contributions are welcome when they improve its
+accuracy, composability, efficiency, testability, portability, or documentation without weakening evidence bounds.
+
+| If you found or need | Start here |
+| --- | --- |
+| A reproducible defect | [Open a bug report](https://github.com/0xprogrammable/hookbuilder/issues/new?template=bug.yml) |
+| A missing capability, source, or integration | [Propose a capability](https://github.com/0xprogrammable/hookbuilder/issues/new?template=capability.yml) |
+| A code, documentation, template, or evaluation change | [Read the contribution guide](CONTRIBUTING.md) |
+| A possible vulnerability | [Report it privately](SECURITY.md) |
+
+Each rule has one owning layer. Change the canonical contract and its tests instead of copying policy into several
+files. Pull requests should name the problem, owning layer, checks run, and remaining evidence boundary.
+
+## Evidence boundaries
+
+| A result can establish | It does not establish |
+| --- | --- |
+| A specific local check passed for exact bytes | An independent audit or a guarantee that code is safe |
+| A package was placed and verified for one host directory | That the host selected or executed the Skill correctly |
+| An exact GitHub application package was prepared | Maintainer acceptance, Registry activation, or launch authority |
+| Deployment or source/runtime evidence exists | Provider indexing, quoting, simulation, execution, or public availability |
+| A project resembles an existing Registry entry | Duplication, ineligibility, or unsafe behavior |
+
+The Builder builds and checks. It does not approve routes, execute trades, audit, deploy, list, endorse, or launch.
+Every external write, signature, deployment, publication, credentialed provider action, and material cost requires
+separate authority.
+
+
+Technical contracts kept explicit on main
+
+- Fee applicability follows the exact project graph. Zero-scope work must not invent a market, PoolKey, hook, or fee
+ receipt.
+- Each Programmable-canonical execution scope preserves the inclusive 10 bps share of executed gross quote-side swap
+ or fill volume exactly once. Claim authority and platform administration remain separate roles.
+- Fee V2 names four settlement profiles. The bundled Solidity kernel evidences only `standard-amm`; every other profile
+ needs its own implementation, custody proof, tests, and review.
+- Application V3 requires exact public GitHub source. Local, private, ZIP-packaged, pasted, or non-GitHub source can
+ support exploration but remains `INTEGRATION_PENDING`.
+- Local Launch Bundles remain unsigned and `NOT_AUTHORIZED`. The Registry Acceptance bridge is a bounded preflight,
+ never Registry or production authority.
+
+Read the canonical details in [Open-World V2 architecture](docs/OPEN_WORLD_V2_ARCHITECTURE.md),
+[Security and review](docs/SECURITY_AND_REVIEW.md), and
+[Programmable platform boundaries](docs/PLATFORM_INTEGRATION.md).
+
+
## Repository map
```text
-skills/programmable-v4-hook-builder/ portable source installed by agents
- SKILL.md small operating contract and router entry
- references/ progressively loaded knowledge and schemas
- assets/starter-catalog/ starters and composable capability packs
- assets/reference-kernels/ reviewed implementation candidates and tests
- scripts/ deterministic local tools
- scripts/test/ portable package tests
-evals/ adversarial agent evaluation suite
-docs/ architecture, usage, source and review records
-config/plugin.json single host-metadata source
-.codex-plugin/ generated Codex manifest
-.agents/plugins/marketplace.json generated Codex marketplace
-.claude-plugin/ generated Claude manifest and marketplace
-.mcp.json and mcp/ Codex-only local MCP companion and tests
-plugins/programmable-v4-builder/ generated, byte-verified Codex plugin payload
-scripts/verify-repository.mjs complete local repository gate
+skills/programmable-v4-hook-builder/ canonical portable Skill
+ references/ routed knowledge, schemas, and policy
+ assets/ starters, capability packs, reference kernels
+ scripts/ deterministic tools and package tests
+evals/ adversarial agent evaluations
+docs/ usage, architecture, security, release records
+config/plugin.json canonical host metadata
+plugins/programmable-v4-builder/ generated, byte-verified Codex payload
+mcp/ and .mcp.json canonical Codex-only MCP companion
```
-The portable skill and root MCP files are canonical. Host manifests and the Codex marketplace payload are generated;
-the payload mirrors canonical bytes and cannot redefine policy. The Claude marketplace is isolated to the canonical
-Skill subtree and cannot package the root Codex-only MCP companion. See [Architecture](docs/ARCHITECTURE.md) for the ownership and upgrade boundaries and
-[Portability and lifecycle](docs/PORTABILITY_AND_LIFECYCLE.md) for evidence-qualified host support.
-
-[`programmable-registry`](https://github.com/0xprogrammable/programmable-registry) is the separate public application
-ledger and project archive. The Builder reads it; project source remains in each builder's own public GitHub repository,
-and the platform repository remains responsible for contracts and the Explorer.
+The portable Skill and root MCP files are canonical. Generated manifests and the Codex payload remain byte-aligned
+with them. See [Architecture](docs/ARCHITECTURE.md) before changing package or host metadata.
## Verify from source
@@ -233,7 +182,7 @@ npm test
gh skill publish --dry-run
```
-The historical and current reference fee kernels additionally use Foundry:
+The two bundled reference fee kernels additionally use Foundry:
```bash
cd skills/programmable-v4-hook-builder/assets/reference-kernels/programmable-volume-fee-v1
@@ -242,33 +191,24 @@ cd ../programmable-volume-fee-v2
forge test
```
-Model-backed eval execution is deliberately separate because it requires an explicit provider credential and can cost
-money. Structural eval validation is included in `npm test`; provider selection is supplied through the explicit
-subject/judge provider contract rather than embedded in semantic cases or rubrics. No model result or credential is
-committed. Provider-neutral definitions do not prove equivalent behavior across Codex, Claude Code, GitHub Copilot or
-other hosts; cross-provider claims require separately authorized, versioned run receipts.
+Model-backed evaluations are separate because they require provider credentials and may cost money. `npm test` still
+validates their structure; an unavailable model result or provider receipt is never treated as passed.
## Documentation
-- [Use the Builder with an agent](docs/AGENT_SKILL.md)
-- [GitHub application journey](docs/PUBLIC_GITHUB_PR_BETA.md)
-- [Open-World V2 architecture](docs/OPEN_WORLD_V2_ARCHITECTURE.md)
-- [Open-World V2 release gates](docs/OPEN_WORLD_V2_RELEASE_GATES.md)
-- [Architecture and surgical change map](docs/ARCHITECTURE.md)
-- [Knowledge routing and token efficiency](docs/KNOWLEDGE_SYSTEM.md)
-- [Templates and extension model](docs/TEMPLATES_AND_EXTENSIONS.md)
-- [Security and review model](docs/SECURITY_AND_REVIEW.md)
-- [Security and audit readiness](docs/SECURITY_AUDIT_READINESS.md)
-- [Code maturity snapshot](docs/CODE_MATURITY.md)
-- [Uniswap source coverage](docs/UNISWAP_MASTER_SKILL_ADOPTION.md)
-- [Programmable platform boundary](docs/PLATFORM_INTEGRATION.md)
-- [Portability and lifecycle](docs/PORTABILITY_AND_LIFECYCLE.md)
-- [Release process](docs/RELEASING.md)
-
-## Contributing and security
-
-Read [CONTRIBUTING.md](CONTRIBUTING.md) before changing knowledge, policy, templates, schemas, validators, or release
-metadata. Security reports belong in [private GitHub vulnerability reporting](SECURITY.md), not a public issue.
-
-Programmable v4 Builder is independent open-source software. Uniswap is a source protocol and ecosystem referenced by
-the Builder; this repository does not claim affiliation with or endorsement by Uniswap Labs or Uniswap Foundation.
+| Goal | Documents |
+| --- | --- |
+| Use the Builder | [Agent Skill guide](docs/AGENT_SKILL.md) · [GitHub application journey](docs/PUBLIC_GITHUB_PR_BETA.md) |
+| Understand the system | [Architecture](docs/ARCHITECTURE.md) · [Open-World V2](docs/OPEN_WORLD_V2_ARCHITECTURE.md) · [Knowledge routing](docs/KNOWLEDGE_SYSTEM.md) |
+| Extend it safely | [Templates and extensions](docs/TEMPLATES_AND_EXTENSIONS.md) · [Security and review](docs/SECURITY_AND_REVIEW.md) · [Contribution guide](CONTRIBUTING.md) |
+| Inspect portability and releases | [Portability and lifecycle](docs/PORTABILITY_AND_LIFECYCLE.md) · [Release gates](docs/OPEN_WORLD_V2_RELEASE_GATES.md) · [Release process](docs/RELEASING.md) |
+
+[`programmable-registry`](https://github.com/0xprogrammable/programmable-registry) is the separate public application
+ledger and project archive. The Builder reads it; each project stays in its builder's own public GitHub repositories,
+and the platform repository remains responsible for contracts and the Explorer.
+
+## License and independence
+
+Programmable v4 Builder is independent open-source software under the [MIT License](LICENSE). Uniswap is a source
+protocol and ecosystem referenced by the Builder. This repository does not claim affiliation with or endorsement by
+Uniswap Labs or the Uniswap Foundation.
diff --git a/assets/README.md b/assets/README.md
index 6834c0f..e47f7df 100644
--- a/assets/README.md
+++ b/assets/README.md
@@ -1,6 +1,12 @@
# Repository assets
-`repository-cover.jpg` is a Programmable-owned brand asset reused for this repository. It depicts connected watercolor
-islands and is used only as the README cover. The human-readable alt text is maintained in the README.
+`repository-cover.jpg` is the 1600 × 720 README cover. It shows paper-cut ants assembling a hook in Programmable's
+floral night garden. The human-readable alt text is maintained in the README.
+
+`repository-social-preview.jpg` is a 1280 × 640 image prepared for the repository's social preview. Committing this
+file does not change GitHub's repository setting; uploading it is a separate, authorized publication action.
+
+Both files are Programmable-owned brand assets derived from the accepted Night Garden composition and the exact
+Warm Ivory loop mark. Keep each repository asset below the repository's 1 MB file limit.
Do not infer Uniswap affiliation, protocol behavior, security, approval, deployment or availability from branding.
diff --git a/assets/repository-cover.jpg b/assets/repository-cover.jpg
index 76b152a810c9c87a49aa82aff0898e6331d258fb..11f6aaa5036dd3aa27564523a017d3be7dee0741 100644
GIT binary patch
literal 111474
zcmbTec|26@|37}l5-D4$?g&E*ie$?=3<;Ht=uWa#k`QBGXP8P^%Fvyd{k9g)zHuu9t82fF}VJJ
z4)PHa=H>XY4b6!VhB$-~oWh7dpCNGwfq)R);PAhK;Njrn-p0$h9aM{f3Qhzl|Eq-%
z9Gsw<=TAShlM?}P2yqI5x{W_@18xXH^P%}5DYTRc{K>^FWdg}^aU&s!4{eS>pt&H(
zf+Jpp05`UP%n=Z%%7#z>7CtltiQGrZB4r^V5ur>W$QZ2yE2iLup
zg&+fh!CeuE7n0)xLqZUs4UA}xHaCc855tX=1@l0%NNz}og8&*p0v0lll3+4H;kXM9
zA{?MUA2{JP2ZP0DLSP=yLrNc#g*YHSFk&{~1%e0=LLYnp`nZu$CI^HtF-NnnEkp=l
zP75%Yz)hJ4N#Gv1WS2}J0t7-I5XQkl;I)AG3<$h1m1rFZ2lOO`I
zsm_M`f&=2GWp6uN!@j|BmxIj`;_5R{35J6O1or_2&6XzMn%5khg30az5dbp5EWB_%
zEJOl)f&eoC@k2cF=}%5JA|Yr8S0=c_%>@o%B3NsLOhPv32Ny0NdWeq~bOI+}M$1E&
zLPWiAz2VQGOdv7@EHnfV1A-68j|-*=P;{RQi~%Y*(ig*?Xc@eLr{hL~8-EK-%6*6f
z&;;ua+ygR)Kj7gIeCVC2)|<*rxJK*#6hYw`;@#!u3DBDx!Sx>Q0ECAG^z$P0`5+`W
zJj%8=x8y)?SLIaOPula3?;aE=A&Gi@Wh(%H3nm{_!-4__4sZbA0Nud_5DNPShj@m}
ziZ(H;&5CqKt+~L|nPtPM6yFHT4Hf8>0?h#IVD%$qVfad+q5E7A7#0-%8>$3IU2a~<
zW1FLlBAIl%;(X!Cp4f;Sovlyr8d>3;fO60tpbHIf&z3GY0M)=60(b`>XJ8Bn0Tse9
zms;3a&fs@vO4W;^Rp#7TIS>9ok##x)<`5LXM|pK6lNXkw2n3{L4j};nFyrjPkrwTv
zratR~jaEB1@B2IiH}%?1OZn67Ag;OA6mBjk`)@OF1}g(t3b+Qd1G?n+&=w*4PK0JaV!38;Z}3M*0wwm~2zDOS^i
zty^_E4Z)!=94kW$wgTM~(~2WA(7~1h#I2&rLV48I1*KU^!AT}}o5i+`|vRZUy;tD0tH1Lz@
z@50xxPa%K~z%>A30Dxe@W&>$(5CD>dvbm)S7&Sd~?Qb1E+v8V4wOE}GYa;5yM3-^?
zy_Fb6_gu~#svfiumsCzR0V)9Vix4>O0$sqt0*nR&5x6$CKQ;g-0B@OuY*@&W1(&fC
z4?Dl4IKzGdkPIX%Ldb+*+JyjDTx_2NA;A#PTuh0M-_<^PlWWv!v0!NM_su5LEzfCf
z{~*Ph_VMc|Yg!QgwPtzzl8ljL`sP14;7?%mnXvr|8=rtHfN|K)MDD{Q!bS=TQVxOU
znVuJ8zGN)6B}YOK>?#4W`3)gt67cSzJC^({X9(I${s4W?Vn7)z~)ZRtGT%Fk%2#ut5GQiS{yZI|25c
z9I&ad3&0O>1Q7TLh|H)_`V2z|4d?|v4|rxNz!J<9uvb{Lz}@IWkZ7EF9p;2iolNa-
z%7BJ@W#Nx*wb3~V_855bQAzxmno0vrKqfQaA&n|II)_P(IP0?@#QIs{4jvYe)Zq9*)r5&LwH*I%mW
zz>Ro4pJLwYYyRPGb4WG$ZsMlV*woy`ilItolEc|hi`CzgP&OD72o`aq0F)#8sXL;ObW%?{mW66Cr}+%OYCZH!e!!{$oJ1fv38!Qi04R+|Yj6WET~
z1ysY~nAaR263O-yK!urqLpodq94x2;xex?#YHKgrJS8}@9Gg3*2K*jt35qSm(H5*u
z5+I%*n~tM1t@~XnqewId2gRrAUtsYe01$xh{F~0e4jj%|z(~MYU`E+;CV-%94zdhT
zfQJPKwvB)jK;(aVfe;uY42c#=H`#O@ooTLeyRcjyrpyZ3_)vj{wv|$K26a1VR2o-P
zl=`QScv}`*=wJdg7X)|%$O3AF&4R;(4e}gS0uTeReOTnMAAty&AW;Kq2LLnyWE